While MPG was testing Photoshop performance for the 2013 Mac Pro real-world performance, several filters were crashing the tests and had to be removed from the test suite; others made it through without crashing, but destabilized Photoshop, leading to a later crash. Ultimately some had to be run “solo” in order to get a timing figure.
MPG reported the crash issues to Adobe, but it quickly became apparent that the crashes were peculiar to the 2013 Mac Pro only, easily seen by disabling GPU support (uncheck) and/or running the same test suite on other Macs (MacBook Pro and/or 2010 Mac Pro) and/or hearing the same thing from other trusted parties.
Some of these crashes remain extant. Adobe is currently working with Apple and AMD to get this issue resolved as soon as possible
For the improvements, Adobe has chosen to undo GPU support for fast sharpening, something MPG picked up on while retesting Photoshop CC speed on the 2013 Mac Pro a few days ago, observing 3X to 4X losses in sharpening speed. MPG agrees that this is the right thing for Adobe to do at this this time. But it is also MPG’s position that Adobe should make a formal statement on where things stand now, since the improvements were announced with 14.2. Unannounce them, along with the reason. (Perhaps MPG has missed such an announcement?).
It is also the MPG position that Apple, having broken sleep and display support with some displays is also (to MPG’s knowledge) not coming clean on the state of the graphics drivers with the 2013 Mac Pro, which MPG testing shows make Photoshop CC too unstable to use on the 2013 Mac Pro with certain other Photoshop filters. The 2013 Mac Pro is a high-end machine that professionals rely upon and invest huge sums in, and hence Apple must step up to the plate first by fixing the broken testing processes at Apple (hardware and software with some extremely dangerous failures), 2nd by having the integrity to acknowledge bugs, and 3rd by providing a timeline for fixes.
The 2013 Mac Pro offers much ballyhooed dual* AMD graphics. Apple just barely shipped it at the end of 2013 to meet the promised deadline, but was it shipped prematurely to meet that deadline with inadequate testing of the dual-GPU graphics system? Readers can ponder that for themselves, perhaps in the context of Apple Core Rot.
* To date, “dual” is a theoretical benefit, as very few programs can use more than one of the two GPUs.
OS X 10.9.2 has been changed to break longstanding behavior with some 3rd-party displays. MPG doesn’t know how widespread the issue is among 3rd-party displays, but it definitely affects NEC displays (e.g., it might or might not affect Eizo or Dell or other brands).
Update March 10: The lightbulb clicks on—the 90° rotation appears to be an AMD thing, meaning AMD graphics and/or graphics drivers. Using the 2010 Mac Pro back in February, I noted a failure of the AMD Sapphire HD 7950 video card to function properly at 90° rotation even as the stock Radeon HD 5870 was working fine and worked fine for years. So the 90° rotation bug might be restricted to AMD graphics, but on any model Mac Pro.
MPG is working with Apple high level Support and also contacts at NEC to get this issue nailed down. Apple support has also communicated the issue to Apple engineers.
Any fix would likely go into 10.9.3 (now in beta), but MPG declines to access 10.9.3 beta, as the Apple NDA restrictions would preclude discussing it here.
MPG has used the NEC wide-gamut displays for ~6 years now, most recently the 30" 2560 X 1600 NEC PA302W and NEC PA301W and also a 2560 X 1440 NEC PA271W (mainly as a spare on a 2nd system). All of them are highly recommended for their calibrated wide-gamut performance.
After years of functioning properly in a dual display setup, along comes OS X 10.9.2 and the 2013 Mac Pro. The following issues are extant*.
* These are issues that reasonable testing by Apple ought to catch (when code is changed, there is that testing thing to be done). OS X releases are breaking more and more things throughout the system at greater and greater frequency, not just display support.
Sleep mode broken by OS X 10.9.2
Various sleep behaviors no longer work correctly. MPG configures NEC displays like this for sleep mode.
Example, occurs with a single display or dual displays—
When the display is supposed to sleep after the Energy Saver duration specified, it ought to go blank (off). Instead with 10.9.2, the image on the display freezes, and remains at full brightness forever, the image burning itself in. Waking the machine up, normal operation resumes.
Since other users are reporting sleep issues with NEC displays on other Macs, this seems to be an OS X bug not involving the graphics card or Mac model.
Dual-display support with 90° rotation
The NEC displays are easily pivoted to be a portrait-orientation. MPG uses dual 30-inch displays, the main one in landscape orientation and the 2nd one in portrait orientation. This is about both desk space (and neck swivel capability) but also about workflow efficiency.
Substituting a 27" 2560 X 1440 NEC PA271W, portrait orientation works—sort of—reboots often force it to some lower scaled resolution while also swapping around the menu bar for good measure, forcing a redo to fix everything (which works ultimately). The flaky behavior strongly suggests a software bug (graphics drivers). And that is on two different Mac Pros (D300 and D700 models).
But the 30" display will not work at all at 90° rotation on the D700 GPU 2013 Mac Pro (after working for weeks on a D300 2013 Mac Pro, then beginning to fail after a swap between machines, suggesting a flaky graphics driver bug).
As it stands, one 30" display has now been rendered useless, costing me the better part of a frustrating day (including a call to Apple). Desk space and neck-swiveling make dual 30" displays side by side in landscape mode for both unusable for MPG. Having a $2200 investment idled is unfunny, but it is also an ongoing workflow impairment.
This 90° rotation bug appears to be limited to the 2013 Mac Pro seems; the same scenario worked flawlessly for the past ~4 years with NEC 30" displays on the 2010 Mac Pro.
AHA! The lightbulb clicks on—it might be an AMD thing, meaning AMD graphics and/or graphics drivers. Using the 2010 Mac Pro back in February, I noted a failure of the AMD Sapphire HD 7950 video card to function properly at 90° rotation (the stock Radeon HD 5870 was working fine and worked fine for years). So this 90° rotation bug might be restricted to AMD graphics.
Additional (non-NEC) graphics issues
The graphics drivers for OS X and the 2013 Mac Pro GPUs are prone to crash; crashes in Photoshop are still occurring with some filters. That display support (see above) has been multiply-broken makes sense in context.
These crashes started with the 2013 Mac Pro, and do not occur for MPG with other Macs running the identical tests.
The crashes have been partially addressed Adobe undid GPU support for fast sharpening, something MPG picked up on while retesting Photoshop CC speed on the 2013 Mac Pro yesterday.
Adobe has not weighed in on the remaining crashes, but to MPG it appears that Adobe code is not involved, e.g., that it is a bug in the Apple drivers for the AMD GPUs in the 2013 Mac Pro. It’s up to Adobe and Apple to publicly clarify the situation and the timeline for fixing it. MPG’s position is that silence on this and other such issues is unacceptable.
With dual Thunderbay units housing eight drives I can house all my original image files and similar (Archive0, Archive1) as well as keep clone backups of both of those as well as my Boot and Master volumes*, and a TimeMachine volume too. Moreover, each archive volume can be a dual drive RAID-0 stripe for high performance in both usage and backup and verification time.
Boot = system and applications, miscellaneous data
Master = current projects (past 6 months or so), high performance SSD
Archive0 = oldest projects
Archive1 = next oldest project
Many users don’t have enough data to require additional archive volumes housing older data, so this simplifies to Boot and Master, with respective backups.
* Online always attached backups are not a substitute for other detached backups stored safely away from the computer. MPG keeps both nearline and separate backups to address both the “I’m too busy to go get those backup drives” and the worst-case “everything attached is stolen/burned up” scenarios.
When the data grows too large
When Master nears capacity, the concept needs to be extended to Archive1, Archive2, etc (your choice of naming convention). Each of these can be backed-up by cloning to single (non RAID) external drives.
The structure can be extended indefinitely without requiring any new mental model. MPG deems this far preferable to larger and larger and more expensive 4/6/8 bay units; these generate the same issues for backup, and end up being more complex, not less.
Configuring the dual Thunderbays
By using dual Thunderbay units, I can “cross configure” so that each Thunderbay houses some originals and some backups, so that if one unit were to fail, the loss would affect some originals and some backups, but the other unit would have a backup of what was lost. Note that a lightning strike or other hazard could take out both units (and the computer), so this approach is NOT a subsitute for additional backups out of harm’s way.
An approach equivalent to this is mandatory as a professional competence issue for anyone whose livelihood depends on data safety. Consulting can walk you through a suitable approach tailored to current and growing needs.
Simplified (omitting Boot and Master):
Unit #1: Archive0, Archive1.Clone
Unit #2: Archive0.Clone, Archive1
A failure of Unit #1 leaves intact copies on Unit #2, and vice versa. Perfect.
The MPG system has a lot of data to house and it’s growing steadily, so more volumes are required, but the concept is unchanged:
Internal SSD: Master, Boot
Unit #1: Archive0, Archive1.Clone, Master.Clone1, Boot.Clone1
Unit #2: Archive0.Clone, Archive1. Master.Clone2, Boot.Clone1
Don’t forget that data validation is an essential part of any backup system. Assuming that backups are intact is a fundamental error.
Tests are one thing, and there the 1TB SSD in the 2013 Mac Pro tests out better than any single (non-RAID) SSD I’ve yet tested.
But real-world results are the only thing that actually matter. Here’s one example.
Second, verifying data integrity with IntegrityChecker is incredibly fast*, encouraging doing so as a matter of course (no assumptions about “should be fine”). See also How to Safely Transfer Data or Verify Backups (IntegrityChecker).
diglloyd2:DIGLLOYD lloyd$ ic verify Master
IntegrityChecker(tm) v1.2.1 64-bit, diglloydTools 2.2.2, 2013-10-26 13:20
Copyright 2006-2013 DIGLLOYD INC. All Rights Reserved ... Using threads = 6, read buffer size = 4096K, num buffers = 36 Looking for files in "/Volumes/Master".. Selecting files for hashing...
Preparing to hash 601014 files...
Files prepared, hashing 601014 files...
0%: 48 files @ 1090.9MB/sec, processed 1.07GB
1%: 70 files @ 1105.1MB/sec, processed 2.17GB
1%: 89 files @ 1110.0MB/sec, processed 3.27GB
1%: 110 files @ 1114.4MB/sec, processed 4.42GB
2%: 128 files @ 1117.9MB/sec, processed 5.54GB
2%: 148 files @ 1120.5MB/sec, processed 6.68GB
2%: 168 files @ 1122.2MB/sec, processed 7.81GB
3%: 189 files @ 1123.5MB/sec, processed 8.97GB
3%: 210 files @ 1123.6MB/sec, processed 10.09GB
3%: 231 files @ 1124.0MB/sec, processed 11.2GB
4%: 250 files @ 1125.5MB/sec, processed 12.3GB
4%: 269 files @ 1120.0MB/sec, processed 13.4GB
4%: 288 files @ 1118.3MB/sec, processed 14.5GB
5%: 308 files @ 1118.8MB/sec, processed 15.6GB
5%: 328 files @ 1119.4MB/sec, processed 16.7GB
5%: 349 files @ 1118.8MB/sec, processed 17.8GB
6%: 370 files @ 1119.2MB/sec, processed 18.9GB
6%: 389 files @ 1119.6MB/sec, processed 20.0GB
6%: 410 files @ 1119.5MB/sec, processed 21.2GB
7%: 431 files @ 1119.9MB/sec, processed 22.3GB
7%: 452 files @ 1120.3MB/sec, processed 23.5GB
7%: 472 files @ 1120.4MB/sec, processed 24.6GB
8%: 493 files @ 1120.0MB/sec, processed 25.7GB
8%: 513 files @ 1118.4MB/sec, processed 26.8GB ...
* IntegrityChecker is one of the most efficient multi-threaded programs available, using all CPU cores with near perfect scalability.
Refer to the chart below; results were with OS X 10.9.1 and Photoshop CC v14.2.0.
With OS X 10.9.2 and Photoshop CC v14.2.1, the Sharpen filter now runs 3X to 4X SLOWER on the 2013 Mac Pro. This reverses the mid-January speedup.
Sharpening was touted as greatly improved in 14.2.0, and MPG duly confirmed and reported on that (see Two GPUs to Rule Them All, and in the Software Bind Them—Photoshop CC Now Sharpens 4X Faster). That improvement has been “un-made”, in ring parlance.
MPG does not hold Adobe culpable here, rather it is Apple that is responsible in two ways: poorly-tested graphics drivers, and a failure to provide a key software partner with adequate machines to test with.
Let’s take the 8-core 3.3 GHz case:
10.9.1 / CC 14.2.0: 2.6 seconds using D500 GPU
10.9.2 / CC 14.2.1: 7.15 seconds using D700 GPUs
That’s right: faster GPUs are running ~3X slower. (the 12-core with D700 GPUs clocked in at 2.4 seconds when tested on 10.9.1 as the chart shows, so it’s not some weird D700 vs D500 vs D300 thing).
Even more dismal results are seen with the 4-core (3.5 seconds to 12.4 seconds and that’s with GPUs used!).
MPG re-ran and rebooted and re-ran with the same shocking results each time.
- The MacBook Pro tests at the ~same speed as before! (5.9 seconds now vs 6.0.
- Other filters test the same or even a little better.
- Turning OpenGL off, the time increases form 7.15 seconds to 11.6 seconds, so clearly the GPU support is not just disabled entirely.
My guess is that the crash bugs in Photoshop (still not fixed, some filters still crash if OpenCL is enabled) have been dealt with by disabling or detuning something.
An inquiry is in to Adobe. It’s unclear if Photoshop CC is directly responsible, because the filter crashes I observe(d) is/are a graphics driver bug according to information received.
The late 2013 MacBook Pro performs the same (5.9 seconds now, vs 6.0 ~= the same), this sure seems like more half-baked Apple Core Rot. And along with severe dual display problems, the MPG advice for professionals is to tread cautiously on the “upgrade” front.
Click for larger chart.
Just tested— the 1TB flash storage (SSD) option in the 2013 Mac Pro is a big bump up in write speed over the 256GB and 512GB SSDs.
Observe a write speed of ~929 MB/sec vs ~755 MB/sec for the lower capacity options. I can’t rule out OS X 10.9.1 vs 10.9.2, but OWC tells me that this difference is their same observation.
My upgraded Mac Pro is here.
Update March 10: The lightbulb clicks on—the 90° rotation bug might be an AMD thing, meaning AMD graphics and/or graphics drivers. Using the 2010 Mac Pro back in February, I noted a failure of the AMD Sapphire HD 7950 video card to function properly at 90° rotation (the stock Radeon HD 5870 was working fine and worked fine for years). So this 90° rotation bug might be restricted to AMD graphics.
MPG is working with Apple high level Support and also contacts at NEC and AMD to get these issues nailed down (sleep is probably an Apple thing, 90° rotation an AMD thing). So all parties are aware of these issues.
Any fix would likely go into 10.9.3 (now in beta), but MPG declines to access 10.9.3 beta, as the Apple NDA restrictions would preclude discussing it here.
But I’ve run into a serious problem involving dual 30-inch displays ( NEC PA302W and NEC PA301W). On the 4-core/D300 Mac Pro I just swapped out, the displays were set up as follows, and working perfectly for a month or so:
PA302W: 2560 X 1600 main display
PA301W: 2560 X 1600 2nd display, in portrait orientation (90° rotation)
This same pairing worked together for the past ~6 months on the 2010 Mac Pro, and a similar pairing for 3.5 years prior to that. And the past ~3 weeks on the 2013 Mac Pro with D300 GPUs.
So along comes the 8-core 2013 Mac Pro with D700 GPUs. I swapped the display cables into the same ports on the 8-core, and now the 2nd display will not run at full resolution in portrait mode! It will do full 2560 X 1600 resolution in landscape mode, but only ugly scaled resolutions in portrait mode (even the option-click trick for more resolutions will not show 1600 X 2560 as an option).
Go figure. The D300 (cheaper) Mac Pro works properly and the “better” D700 Mac Pro does not. The only thing I can figure is that the “better” D700 GPUs have a bug, and the D300 GPUs do not. But given the new batch of display sleep bugs in 10.9.2 (after years of no issues) and GPU crashes in Photoshop (the AMD driver for Apple), the issue seems to fall into “Apple bug” land. Seems to at least—bugs are that.
What I tried
The things I’ve done/tried have wasted the better part of a day, with no solution in sight. I’ve gone through all A/B variables, etc. All efforts point to a bug:
- I’ve unplugged all additional Thunderbolt devices.
- Tried two different Mini DisplayPort cables to rule out cable issue.
- Rebooted several times.
- Reset PRAM (twice).
- Created a different admin account.
- Downloaded and installed the OS X 10.9.2 update over existing install.
- Update not working, so installed and booted off a brand-new OS X Mavericks install on a separate drive to rule out some oddity in my system.
- Cloned the system from the 4-core/D300 Mac Pro to an external SSD, and booted both Mac Pros booted off that same drive/system. The D300 Mac works perfectly and the D700 Mac Pro fails as described.
- A/B swapped the two Mac Pros several times (same displays with same cables in matching ports). The Mac Pro with D300 GPU works perfectly (as did 2010 Mac Pro for years), the Mac Pro with the D700 GPUs malfunctions as described.
- Tried a NEC PA271W (2560 X 1440) 27-inch display, which works fine with the D700 Mac Pro.
- Spent an hour or so with Apple 1st and 2nd level technical support: “first report we’ve had”.
It appears that there is a software or hardware bug with a D700-based Mac Pro. Or that my PA301W and the Mac Pro with D700 GPUs just dislike each other, for unknown reasons. Or maybe it’s a PA301W bug only or an Apple OS X bug only or a 2013 Mac Pro D700 bug only! I don’t know.
This bug is a problem for my work efficiency, so it’s driving me crazy.
More below! Symptoms run deeper and wider.
More problems, and worse than thought
Subsequent to the above, I observed other things that make it clear that there is something seriously unstable with multiple display support in OS X 10.9.2:
- Rebooting sometimes swaps the menu bar to the other screen and simultaneously changes the screen resolution to a scaled low-res variant. I observed this happen on BOTH Mac Pros with the 2560 X 1440 PA271W, which otherwise was working as expected. Using the Displays control panel, I was able to restore full resolution at 90° for the PA271W and to put the menu bar back on the main display. A reboot was OK, but another reboot screws it all up again. Or doesn’t—it’s flaky.
- Re-attaching the two displays to the Mac Pro D300 (vs D700 model), the same display resolution problem was observed with the PA271W and also with rebooting. But again, flaky, and not always.
- All of this after weeks of trouble-free use on the D300 Mac Pro, and years of trouble free use on the 2010 Mac Pro. Something flaky is in the innards of OS X here. Reboots apparently trigger it.
Note also that Apple has broken display sleep in 10.9.2: I’ve been dealing with this as well as hearing reports from readers with NEC displays (though it did not start with the 2013 Mac Pro, suggesting something broken at the OS level, not hardware level).
Also, graphics drivers for the new Mac Pro GPUs still crash Photoshop in some cases (these are apparently written by AMD for Apple). No wonder 10.9.3 reportedly includes graphics driver fixes (unspecified).
With two 2013 Mac Pros both showing the symptoms discussed using displays proven to work for long periods of time (years), something deeper and more troublesome is at work: the increasing trend to shoving sloppy untested changes out the door. When a company can break a fundamental and critical security underpinning, can one reasonably expect that multiple display support escape unscathed? BTW, full screen viewing mode to does not function in any useful way on two displays, which should come as no surprise in this context.
Now published is a summary and discussion of the finalized MPG 2013 Mac Pro configuration for a intensive digital photography usage.
See the MPG buying links to B&H photo for the 2013 Mac Pro. Thanks for using!
MPG has the privilege of being the first advance customer of the new OWC CPU upgrade program* - which OWC will soon be opening up for all.
Now upgraded, it’s on its way back to me, where it will have 64GB memory installed to yield the fastest 2013 Mac Pro possible ( as per the tests). Still, it’s only a little faster than the stock 6-core CPU, which is the MPG recommendation for the vast majority of users.
MPG uses the Mac Pro mainly for intensive photography, every day. Few users need this much gear, but some do, and most users will need some of it.
The key things are optimizing workflow, a well-conceived strategy for storage, and a robust backup approach. Expert advice can be helpful in all those areas, but also for saving money by not over-configuring for little or negative benefit and/or tips on how and where to buy stuff.
- 3.3 GHz 8-core CPU.
- 1TB flash storage (SSD).
- D700 GPUs.
- 64GB memory.
- Dual OWC Thunderbay external 4-drive enclosures.
- CalDigit T3 Thunderbolt enclosure.
- OWC Helios 2 with dual 960GB OWC Mercury Accelsior PCIe SSD (as soon as the Helios 2 is released).
- CalDigit Thunderbolt Station.
- TRIPP LITE USB3 expansion hub.
- FirmTek Q6G Thunderbolt to eSATA (for backward compatibility with existing eSATA enclosures).
Most drivers are prudent, but it’s clear that distracted driving is a serious problem too. Of those situations, some are aggressive drivers, a few are mentally disturbed malicious individuals*, but taking the troublesome minority of drivers as a group, most are simply careless or impatient (an “accident”). Hence my concern about iOS in a car leading to distracted driving.
Along comes Apple with the idea that iOS in the car is a good idea (Apple CarPlay). Since when are messaging and videos and similar distractions a good combination with driving, particularly with teenagers? Teenagers tweeting in cars and watching videos and status updates from friends with built-in iOS makes me nervous. As did a recent incident with two teenage girls gyrating wildly (including the driver) to “do you wanna be my lover” while tailgating the car ahead and passing me on my bicycle. OMG.
Though one might already bow in defeat to the reality of iPhone (ab)use in vehicles already, it is a fair question to ask how many people will end up dead or mangled as a result of inappropriate usage once the idea is accepted by virtue of it being a built-in part of driving. That number is not going to be zero.
To be fair to Apple, it is a general issue and someone is going to do it anyway. I don’t have the answers here, but driving and iOS-like technologies aren’t likely to ever be a good idea. Maybe a multi-way proximity sensor that shuts off functionality would make some mitigation sense, and while I abhor tracking, a “black box” recorder noting usage of messaging and similar features could be appropriate for determining culpability, so “accident” is not an excuse. It is a question that the legislature will have to decide in 50 states. But here in California, text messaging or phone usage is a ticketable offense (wireless link to phone is OK, for both hands on the wheel).
* In one encounter I was hit by a disturbed individual who the CHP neglected to even ticket in spite of a witness shocked by what he saw. That disturbed driver caused an indirect death (heart attack) in a road-rage incident ~2 years later.
My kids can’t understand why they shouldn’t use things like FaceBook et al, but to be fair to them, the vast majority of adults think the same way. Publishing details of one’s personal life online poses risks for years to come (surveillance, criminals, job searches, “privacy”). Perhaps overseas where thugs rule, people understand this better than in the USA (though there is very uncomfortable presumption in that sentence).
But what about another area that presumably most users consider private?
Britain's surveillance agency GCHQ, with aid from the US National Security Agency, intercepted and stored the webcam images of millions of internet users not suspected of wrongdoing, secret documents reveal.
The use of very strong peer-to-peer encryption that does not involve any certificate or other authority is the right way to go for such things. That a company like Yahoo would even provide a service that could be intercepted is a good reason not to use it. I don’t want anyone to know I’m going to the grocery store, let alone more sensitive areas like the emerging area of doctor/patient video chats.
Having once worked at PGP, I am disappointed to see how little progress has been made in getting away from the fragile infrastructure of official “trusted” parties (encryption or otherwise) for the transmission of private information.
The internal 2013 Mac Pro PCIe SSD (flash storage) is a very high performance device, at present the clearly preferable solution for high performance needs such as housing Lightroom catalogs, or actively editing large Photoshop files, scratch files, etc.
In some cases it might be useful to partition it into two partitions, such as Boot (system and applications) and Master (storage for files that benefit from high performance).
See the MPG buying links to B&H photo for the 2013 Mac Pro. Thanks for using!
The MPG Mac Pro has arrived. It was supposed to go directly to OWC for the 3.3 GHz 8-core CPU upgrade, but in spite of four phone calls to Apple with assurances of the shipping address being updated, the shipping address was not changed. No big deal (sarcasm), only a $350 screwup for me—I have to have it right away, since the loaner I am using has to go back and I cannot be without a computer. So it’s overnight shipping two way. Nice job Apple—ship it late and ship it wrong.
Update March 7: I have to hand it to Apple for doing something entirely unexpected and praiseworthy: out of the blue I received a call from Apple apologizing for the mishandling, along with an offer to refund some of the shipping (about half, e.g. ~$175). While I am not made entirely whole, I expected nothing at all, certainly not coming out of nowhere like that. It’s a class act and deserves mention here.
So tomorrow out it goes to OWC for that CPU upgrade*. I have the privilege of being the first customer of this new OWC program - which OWC will soon be opening up for all.
It should be back by Friday or Saturday, where it will be put into “production” use, primarily for intensive photography. As shown below, I ordered the 4-core CPU with 12GB; it will become an 8-core with 64GB memory and the D700 GPUs. It doesn’t get any faster. Still, it’s only a little faster than the stock 6-core CPU, which is the MPG recommendation for the vast majority of users.
* OWC is performing the CPU upgrade on the MPG Mac Pro in advance of public availability of the upgrade program (expected soon). Rolling out such a service requires logistics and support to do it right, so anyone looking to do a CPU upgrade should appreciate that OWC is taking the time to implement the service robustly.
How those thunderbolt ports are used matters—see 2013 Mac Pro: Thunderbolt Performance Tips.
Deciding which Mac Pro variant is best for a particular workflow is not obvious—it’s easy to overspend on a system that might actually run slower than a less expensive one. Spending $2K more for a slower system is not an appealing thought. And in some cases an alternative might be preferred.
Overspending is not the only risk; choosing the wrong system can mean having to buy the right system later and that’s a really expensive mistake. A trip to the Apple Store slants the odds against you by the glitz and pressure to decide, just as walking into a car dealership uninformed can be a costly experience (read this).
Just yesterday I walked a photographer through the choice of systems, this client having almost turned away from the 2013 Mac Pro thinking it had to cost $10K. As it turned out, the ideal system was half that cost for her needs, and will run her workflow faster than the more expensive choices.
What actually matters for performance?
Moreover, a professional needs a computing system that takes into account current and future needs, reliability, backup and expansion.
The wrong choices can mean spending more money for less performance, inadequate storage capacity, confusing backup, or workflow configuration that does not utilize the best approach.
For all these considerations, I offer personalized consulting.
This article by Bruce Schneier (a true security expert) is excellent.
Last year, Ars Technica gave three experts a 16,000-entry encrypted password file, and asked them to break as many as possible. The winner got 90% of them, the loser 62% -- in a few hours. It's the same sort of thing we saw in 2012, 2007, and earlier. If there's any new news, it's that this kind of thing is getting easier faster than people think.
See the MPG buying links to B&H photo for the 2013 Mac Pro. Thank you for using!
Something is rotten in the city of Cupertino, or rather, Austin TX, where the Mac Pro is manufactured. It’s just strange to see Apple execute poorly.
I’ve never seen such amorphous delivery dates, and now my order has clearly slipped: I received a phone call from Apple in Texas at 15:55 informing me that my Mac pro would ship in “5 to 7 days”. MacRumors has a similar report.
Oddly, a CC hold was put on my card a full week ago, a hold that has since elapsed, which is even more strange: how can a company think it’s time to ship, pre-charge the full amount, then miss it by 5-7 days?
Apple is clearly having logistics and planning difficulties, but the nature of the issue(s) is unclear. Perhaps there is a component shortage that is delaying production, or perhaps there are problems in the production itself. But that is only a symptom—to think it does not go deeper makes for a comforting but naive thought, no matter what public excuse or rationalization is offered.
MPG sees it as another of many symptoms of Apple Core Rot, now spreading to the execution level. Apple reached its zenith with Steve Jobs (a true genius), believe it or not. All great companies gain steam, but like a 2-mile-long train train loaded with iron ore (the spillage great ammo for my childhood slingshot), inertia carries it a very long way, and a cooked noodle at the helm cannot steer a visionary course. It will take a long time for that to be perceived (being first always looks wrong), but that changes nothing. Nor does it mean I don’t prefer Apple products over competing brands—it isn’t relevant to the foregoing—for now. But I would not be 'long' on Apple, even if there are short and intermediate term gains likely.
UPDATE Mar 1: to their credit Apple updated shipping to overnight and I did not ask, they just did it. That is a good way to operate and deserves kudos.
Now my challenge is to get the shipment rerouted to OWC for the 3.3 GHz 8-core CPU upgrade before it actually goes onto a truck/plane. (Among other considerations, Apple can’t charge sales tax for destination vs buyer, e.g., Texas vs California, so I had to wait until my CC was actually charged, there is now a pending charge for the full amount).
Update, Mar 4: A phone call to Apple Saturday Mar 1st after my CC was charged led to assurances of the shipping address being updated as requested, and that was while the Mac Pro was still at Apple. A subsequent phone call on Monday led to “it should be rerouted” which led to noting— the Mac Pro was delivered today to the address I did not want. So now I have to ship it out to where it needs to go. This follows two other phone calls to Apple the week prior.
2013 Mac Pro: USB3 Performance Impaired Relative to MacBook Pro (Single PCIe Lane Throttles Bandwidth)
The USB3 implementation on the late 2013 Mac Pro is a non-starter for high performance applications because its four USB3 ports are restricted to 500MB/sec theoretical maximum speed (in each direction). Actual limits are even lower.
A RAID-0 stripe of SSDs on USB3 is no faster than a single SSD on the Mac Pro.
By comparison, the late 2013 MacBook Pro offers far higher aggregate bandwidth, and superior performance even with a single SSD.
OS X 10.9.2 is out with a stated fix for the TLS/SSL bug.
Software Update failed to work for me, insisting there was no update at all (even after rebooting and multiple checks).
I had to go and manually download the OS X Mavericks 10.9.2 update.
UPDATE: OS X 10.9.2 is out with a fix for the TLS/SSL bug.
Apple has plenty of engineers for eye candy and destructive changes to usability.
But what about real usability, respect for compatibility and just plain fixing bugs? OS X Mavericks introduces more new bugs than I have ever seen in an OS release (including a new file system hang that I personally encountered*.
But that’s all trivial stuff in context—
In a nutshell, this particular bug allows a MITM attack (man in the middle); an attacker can interpose between the two ends, fooling each into believing the conversation is direct, undetected by both, capture everything in unencrypted form.
This exploit is made possible by the failure to validate the certificate (the nature of the Apple bug), that validation step being fundamental to the security. It is so critical and fundamental that no implementation could reasonably be released without a test suite to validate that it functions properly, well that’s my view when tens of millions of devices are in use. Obviously, Apple did not have that test suite in place.
The bug is so simple that it’s surprising a compiler warning was not issued: a 'goto' statement in the middle of a series of 'if' statements. With security code, one ought to enable all the warnings and have rigorous code-review processes in place.
What’s puzzling is given that the bug is trivially simple to fix (take out the errant 'goto'), Apple still has not issued an OS X fix. Fix the bug, and get that test suite written yesterday Apple.
“researchers faulted Apple for inadequate testing” 
This one is unforgiveable**. It could have compromised interactions with tens of millions of devices, had hackers exploited it (have they?), and that fact remains true for some time to come because plenty of people won’t update their devices and OS X doesn’t even have a fix as this was written.
You just don’t break a core security protocol like this. Who is in charge over there? Test suites should validate such stuff; it’s not exactly a new protocol. Heads ought to roll on this one and right up to high levels perhaps.
A security fix is in iOS 7.0.6 which is now available (affects iOS 6 also, v6.1.6 might fix it). Apple is mum on the full timeline and extent of the issue. See http://support.apple.com/kb/HT6147.
* Engineers at Apple have yet to test the Mavericks file system hang I reported, which is reproducible by highly competent other parties using different gear. I’ve filed a Bug Reporter bug and given specific instructions for a 100% reproducible test taking 5 minutes, but so far the response amounts to no action. I see it as indicative of serious internal issues at Apple in quality control.
** Your author worked for at time at Pretty Good Privacy (PGP), is familiar with TLS/SSL, and has implemented SSL on server software.
OS X is affected and a fix is yet to arrive
UPDATE: OS X has the same flaw and a fix HAS NOT been issued as of Feb 22. Which means that hackers everywhere are game to exploit this while they can. Details at Reuters.
Apple released a fix Friday afternoon for the mobile devices running iOS, and most will update automatically. Once that fix came out, experts dissected it and saw the same fundamental issue in the operating system for Apple's mainstream computers.
That started a race, as intelligence agencies and criminals will try to write programs that take advantage of the flaw on Macs before Apple pushes out the fix for them.
The flaw is so odd in retrospect that researchers faulted Apple for inadequate testing and some speculated that it had been introduced deliberately, either by a rogue engineer or a spy. Former intelligence operatives said that the best "back doors" often look like mistakes.
The bug has been present for months, according to researchers who tested earlier versions of Apple's software. No one had publicly reported it before, which means that any knowledge of it was tightly held and that there is a chance it hadn't been used
The issue is a "fundamental bug in Apple's SSL implementation," said Dmitri Alperovitch, chief technology officer at security firm CrowdStrike Inc
NOTE: “researchers faulted Apple for inadequate testing”.
Big surprise? Hardly. I noticed this starting years ago, and it’s why I wrote Apple Core Rot a little over a year ago; it just became so obvious that software quality was being compromised in so many ways. It’s one reason why Apple’s proprietary hidden code presents a very serious concern in a worldwide sense up to and including the spectre of a national security threat: one vendor with a proven track record of extreme secrecy depended upon by tens of millions of users, if not more. Open source can be examined by experts; closed source code cannot.
BTW, it is NOT true that iOS devices update automatically. Mine didn’t; I had to connect them up, launch iTunes and make it happen. I have a friend with iPhones who doesn’t even get it near a computer or WiFi for months = no update. He can’t be alone.
This kind of flaw makes the Target fiasco look like a minor spat in a kiddie sandbox: the effects of a TSL/SSL compromise in Apple iOS and OS X are worldwide and could be devastating if exploited properly by opportunistic hackers. What is Apple’s liability if something happens, e.g., access to Citbank, Chase, Schwab, and similar financial sites?
Moreover, the ramifications extend well beyond now and well beyond a fix: systems compromised by this bug could harbor malware lying in wait undetected. It is a very ugly potentiality. It’s not just about your access, it’s about that financial institution employee whose system was compromised... an employee with access to critical systems. Ditto for internet service providers. I wonder if anyone in the military uses an iPhone or iPad or Mac?
It’s not clear at all if use of Google Chrome or Mozilla Firefox avoids the security issue, but Apple kicks Mac users in the teeth by not IMMEDIATELY making that point clear (so users can avoid Safari). Apple should be on paid television telling users exactly how to safeguard their internet use, how to play it safe. It’s unconscionable. The core rot extends to ethics apparently.
Wow. It’s a huge prize and if hackers everywhere aren’t salivating and working 24 X 7 on this gift from heaven... consider that they don’t have to compromise your Mac or iOS device, they just have to compromise the right accounts at the right companies, insert their hooks and all heck could break loose in the Garden of Apple.
Is the 'cloud' and swarms of devices attached to it an inherently flawed design ?
Your author worked for a few years at PGP, where security does not rely on a single certificate authority (“web of trust”). Unfortunately, the world adopted the hierarchical X509 system, which fails utterly when any top level signer is compromised, and forces users into no backstop (no ability to require more than one signer). The hierarchical vs distributed trust model is not the issue per se here, but it is related in that the bug involves failing to check the certificates properly. Moreover, a cloud-based future makes distributed trust (more than one signer) the only really intelligent choice, just in case an authority is compromised (and that has happened).
In an age where millions of always-on devices are at risk, you don’t screw up fundamentally critical things like this. It’s one reason I abhor gatekeeper type services like the Apple App Store: one screwup and the entire system is at risk worldwide for tens or hundreds of millions of devices. I wrote about this months ago, and while some readers poo-pooed my remarks as alarmist, I repeat that warning even more emphatically now.
It's a near certainty that the NSA has had that exploit at its disposal for a while, and the NSA sniffs the backbone of the internet, so it is perfectly situated for a MITM exploit (and surely has a well developed kit of tools to do just that). Other spy agencies or simply sophisticated hackers with compromises to backbone systems, or at least wireless networks are hardly to be ruled out, which puts them into a nifty position also. When TLS/SSL is undermined, the entire internet is undermined.
Joshua C writes:
Just noting that iOS 6.1.6 also addresses this fix. So it's not just limited to iOS 7 and therefore goes a ways back.
MPG: the TLS/SSL protocol has been around a while. It is unconscionable to see such a basic check on a key underpinning (signature checking on certificates) not incorporated into a validation test suite. Put plainly, it is gross incompetence. It is so unbelievable that one might contemplate the “deliberate action” or “intentional” scenarios. I wouldn’t want to be the engineer who committed that source code change into the SCC system.
Add two eSATA and two USB3 ports to your older Mac Pro with the NewerTech MAXPower eSATA 6G + USB3 PCIe Card. This card will be tested soon here at MPG.
UPDATE: there are two variants of this card, one driverless xHCI card that does not support sleep, and one card that requires a driver and supports sleep. Details.
Card readers are ideal for downloading gigabytes of the digital camera raw files that I shoot (JPEG is mediocre at best).
Connecting the camera is often awkward and they are usually awfully slow via a USB2 port. The last thing I want to do after shooting all day is to wait for 20 gigabytes to download over USB2.
See the MPG buying links to B&H photo for the 2013 Mac Pro. Thank you for using those links.
February wanes, but maybe it will ship in time like Apple’s barely-met “new Mac Pro in 2013 promise”. At least it’s not a leap year.
UPDATE, February 21: my CC has been charged, so unit should presumably ship with in a few days.
UPDATE 2: CC has not been charged after all, it was a “hold” for the charge that ought to be coming. But as of Feb 28, still says “available to ship February”, which is what it has said since Jan 6.
Update 3, Mar 4: Four phone calls over four different days eventually led to assurances of the shipping address being changed as I requested, and that was while the Mac Pro was still at Apple. Another one as it went to UPS... all of which means the box is on the truck for delivery to the wrong address.
I’ve made the leap to the 2013 Mac Pro for my primary “production” workstation (intensive photography is the primary use). Currently that means the 4-core 3.7 GHz with 64GB memory and D300 GPUs, on loan from B&H Photo, which is an authorized Apple dealer (30 day loan period, so I’m testing it well!).
The MPG Mac Pro order still reads “February” for delivery. It will have its CPU upgraded to a 3.3 GHz 8-core by OWC (announcement of that upgrade program is not yet official, but it is coming). Apple Store lead times have extended to April, so I’m glad I got this order in back in January. But what “February” means even the customer service representative at Apple could not say.
Apple planning and delivery systems for the Mac Pro are in apparent disarray; the online system failed a dozen times over two days with a mysterious “unexpected error” when I tried to change my shipping address. Calling Apple, they could not change it either, their own systems balking. I was told some nonsense story about which warehouse it was shipping from was pre-planned so it could not be changed—which is nonsense because these are all being built in one place (Austin, Texas) and this is a build-to-order order. The logistics for the Mac Pro are thus the creakiest I’ve seen from Apple: no specific ship dates and everything locked down on the order. A highly credible source told of receiving a new Mac Pro with 1TB SSD spec'd and noted on the packing list, but arriving with a 256GB SSD installed.
No Mac has more than four USB3 ports, and I have about 10-12 devices to connect, so what to do?
After a search, I’ve settled on a USB3 expansion hub that meets all my requirements, no easy task in the huge haystack of crap quality hubs, USB 2 hubs, underpowered hubs, hubs with funky shapes and lights, etc.
Also in use are three USB3 ports on the CalDigit Thunderbolt Station. Together, the TRIPP LITE unit and the CalDigit solves the expansion challenge.
Benchmarks can deceive for real work. While 4 cores is enough for many things, 6 cores provides some depth.
Never in the history of the Mac have I seen so many bugs crop up in an OS release—OS Mavericks. They are there by the dozens, from obvious GUI bugs to nasty file system hangs (the file system hang I filed with Bug Reporter, but the Apple won’t bother to take 5 minutes to test it using the simple and precise steps I provided).
Also, and perhaps most offensive, arbitrary changes and removals of features and behaviors that users come to rely upon, a slap in the face approach to software updates first established back with Final Cut Pro.
Bottom line is that I have lost trust in the quality of system software releases, which show increasingly sloppy work and careless thinking. I guess “everyone” has one Mac with one drive whose purpose is to watch movies and play games and fry eardrums. BTW, indications are that printing and display sleep and Mail will be even more broken soon.
And still, after many years of waiting and after this very expensive new trashcan Mac Pro with 4K video: no 10-bit video support.
My advice: defer unnecessary system updates by 3-6 months unless a bug that is acknowledged is listed as fixed in the next release (and you need that fix). Professionals are unwise to “upgrade” production systems to Mavericks, barring some mandatory need (e.g. new computer) or pressing need for a new feature (highly unlikely). By the way, professional photographers with color managed and calibrated printing needs should exercise extreme care in the printing support area (it’s unclear to MPG if internal colorspace conversion breakage a few OS X releases ago has been fixed).
Apple Core Rot is accelerating.
One huge time waster example
This bug affects both the Finder and the Open/Save dialogs: columns in list view are always hidden or partially hidden *way* over to the right, off the edge of the window.
The only reason that the columes are visible here in this screen shot is that the window was resized to maximum make them (partly) visible.
The Finder forgets any attempt to correct the column widths ; close the window, reopen it, start over. Add or remove a column; start over. Ditto for Open/Save dialogs.
Accordingly, each and every day, I have to size windows to 2500 pixels or so a few dozen times per day in order to be able to see the columns I need to see. It’s a constant waste of time, and it happens on all my Macs and a fresh install does not cure it. Perhaps it is related to screen size (2560 width on desktop, 1440/2880 Retina display on laptop).
The particular window seen below is a MacBook pro with Retina display at normal "Best” screen scaling; the window is the full 2880 pixels wide. It’s unclear how even rudimentary testing could fail to see this bug.
Tried and failed to fix:
rm ~/Library/Preferences/com.apple.finder.plist&&killall Finder
It could be related to the sidebar and/or combinations of what is shown, though that would not explain the Open/Save Dialogs.
For the right tasks, the Sapphire HD 7950 is right in the D300/D500 GPU range on the new Mac Pro. With all GPUs (video cards) in general, many tasks see no benefit, but some do, and a fast GPU (video card) can be a huge improvement. If those are your tasks, it’s a great investment.
All or nearly all tests have been updated now to include the base-model 2013 Mac Pro 3.7 GHz 4-core with D300 GPUs. It‘s a darn fine machine; no one should feel that the base model is 2nd rate; it gives up a lot less than one might think on most work. The gains with 6 and 8 cores are modest in most cases, mainly due to software that needs improvement.
I’ve made the leap to the 2013 Mac Pro for my primary “production” workstation (intensive photography is the primary use). Currently that means the 4-core 3.7 GHz with 64GB memory and D300 GPUs, on loan from B&H Photo, which is an authorized Apple dealer (30 day loan period, so I’m testing it well!).
See the MPG buying links to B&H photo for the 2013 Mac Pro. Thank you for using those links.
The MPG Mac Pro still reads “February” for delivery, now 32 days since ordered. It will have its CPU upgraded to a 3.3 GHz 8-core by OWC (announcement of that upgrade program is not yet official, but it is coming).
The switch over process was no trivial exercise to think through for the conversion, make the backups, incorporate some forward planning etc. And with all the hard drives and SSDs and a need to stay operable with very little downtime and the Thunderbolt considerations
As yet, four 480GB SSDs are idled (no home for them yet), and one Accelsior PCIe SSD sits idle, though I hope to resolve both of those handicaps soon.
I had a terrible time with the Promise Pegasus J4, but OWC has graciously agreed to take it back for a refund. It is highly unlikely that MPG will ever consider Promise Technology products again, in good measure because of an infuriating support experience. Moreover, varied sources have filtered my my way on quality issues, and this all “clicks” for me with the J4 experience and in more ways than I’ve documented.
Performance seems at least as good as the prior 3.33 GHz 12-core Mac Pro, which is remarkable for a 4-core system.
UPDATE: the Promise Pegasus J4 is giving me fits. Taking 4 proven OWC 6G SSDs, the J4 will disappear 4 or 3 or 2 or 1 of them each time I reboot the system, or power cycle the J4. Totally flaky. Yet the same SSDs work perfectly in my old Mac Pro and have done so for 18 months. And it’s a configuration I previously tested. So at the least I suspect a bad Pegasus J4.
But OMG what a tangle of power cables and USB cables and Thunderbolt cables, and I still could use several more USB3 ports even with the CalDigit Thunderbolt Station attached. No one can say this is elegant; it’s a mess. It’s like owning a car consisting of an engine, but to drive it one must attach a trailer for the driver and another trailer for passengers and the tire modules... elegant it is not, not when actually configured to be useful. But it is highly expandable, and when Thunderbolt 2 devices arrive and more variety too, it will start feeling its oats.
Nor are the power savings there when one looks at the total system power draw: the Mac Pro itself saves 120 watts or so, but that is all eaten up by all the many required add-ons. So many dilettante reviewers just don’t “get” the reality of a working system and regurgitate the official PR on power usage; it’s like testing for gas mileage driving down a 10° slope.
But I do like the end result: once cabled and stowed out of the way under the desk (a big plus of Thunderbolt), things are working great. Except for Apple OS X Mavericks bugs, which infest the system by the dozens.
What does an enterprised-grade hard drive buy you exactly? Perhaps some consistency, vibration resistance for multi-drive RAID setups, etc, but it ought to also be faster, right? Well, not necessarily.
I’ve just started verifying performance on the 2013 Mac Pro with Thunderbolt devices, in preparation for a switch-over to it as my main workstation. As it turns out, it’s not a trivial thing to configure for those using high performance devices.
It turns out that attaching the Helios/Accelsior to the same Thunderbolt bus as the 30-inch display meant that not enough bandwidth remained for full performance! Attaching it to a different Thunderbolt bus yielded full performance.
The MPG review of the OWC Thunderbay 4-bay Thunderbolt enclosure is now up.
Included are an overview as well as various pages of performance analysis. The OWC Thunderbay is a top notch performing unit.
Joshua C writes:
I've been reading your site for years and following a modified version of your backup strategy (no offsite, just a drive in a firebox away from the computer).
My current RAID enclosure was getting flaky so I decided to replace it. The problem was temperature and fan related so I was migrating the drives to the new enclosure
I knew I would have to reformat but I've got two backups so that's OK. That is until I accidentally installed my clone drive as one of the drives in the RAID without noticing. Bang! My original and my clone were gone. My heart stopped. This RAID contains ALL my image files.
But since I have another backup I am rebuilding as I write this. One human error could have wiped me out. Following your guidelines has kept me going.
MPG: See the backup pages for critical backup strategy discussions.
Two backups are far better than one: with one backup, a failure means you’re down to a single master copy, possibly out of date. With two backups, loss of one backup means you’re down to one remaining master copy (possibly out of date), as with Joshua C.
Anything less than three complete backups should be a source of uneasiness for anyone who values their data. For starters, spreading the risk (physical risks, bonehead mistakes, out of date backups, drive failures, splitting across sites, etc).
Time Machine backups should be considered strictly a nice-to-have additional redundancy for short term protection. It does not count for the “three”. For one thing, Time Machine has had historical bugs that left data unprotected (and I don’t trust Apple anymore, what with the massive increase in bug count in Mavericks), for another it is unbootable, and it’s also typically subject to the same physical risks. Also, if it fills up, you’re out of luck until something is done, whereas clone backups used properly can never fill up provided they are as large or larger than the original (e.g. some amount of data on a 4TB original can never overflow a 4TB clone backup).
Here is a good general purpose backup drive suitable for all Macs. 4TB or at least 3TB capacity recommended.
You do NOT need RAID for backup; it is far better to have two independent drives than a RAID mirror. RAID is a fault tolerance or performance solution, NOT a backup solution.
The 6-core 3.5 GHz CPU is an outstanding performer and should be the model of choice for most everyone, as the tests show. Match the model choice to the work you actually do (never to generalized benchmarks); see the tests along with 2013 Mac Pro: How Many Cores for Your Workflow?.
Gains with other Mac Pro models for Photoshop and Lightroom are modest, or actual performance declines, though there are other applications where more than 6 cores offers some advantage. Be sure to note that the Apple 8-core 3.0 GHz offering will be somewhat slower than the CPU-upgraded 8-core 3.3 GHz shown in the tests.
My use of Photoshop is intensive (big files, hundreds of hours of use over the year), so I’m looking to squeeze out every drop of performance. So my approach is to order the 4-core Mac Pro with the intent of upgrading the CPU to the 8 core 3.3 GHz.
Given recent support in Photoshop CC for the GPU with huge speed gains in certain areas, I’ve also ordered the 4-core Mac Pro with the D700 GPUs. I deem the odds low for meaningful gains over the D500 GPUs, but given my intensive usage, it’s a cost I’ll incur.
As for order status of the above, Apple just sent a notice saying in essence “we don’t know when it will ship, but it will be in February as stated nearly 4 weeks ago”. It seems odd for Apple’s well-honed supply chain excellence to have such a vague grasp of delivery time frames. The message just feels frustrating by being so vague.
Dear Lloyd Chambers,
Thank you for your recent order. We know you’re excited to get started with your new Mac Pro, so we’d like to give you a quick update on the status of your order.
Your Mac Pro is scheduled to ship in February. When it does ship, we’ll email you with a more specific delivery date.
For the most up-to-date delivery information, visit Order Status on the Apple Online Store. If you have an iOS device, you can check your order status using the Apple Store app for iPad or iPhone.
We appreciate your patience and look forward to getting the new Mac Pro to you as soon as possible.
The Apple Online Store
For better or worse (Apple keeps making it worse), I use Apple Mail for my email. But Apple’s junk mail filter is buggy and still sometimes blips on when disabled; it’s buggy even when disabled!
Enter Michael Tsai’s SpamSieve, which I’ve used for some years now. With a little training, it’s highly effective at passing through the good stuff and blocking the unwanted stuff, which I can quickly skim just in case (no filter is ever perfect).
SpamSieve is indispensible for my high volume email needs!
A point driven home regularly by Apple: every time Apple Mail changes, Apple intentionally disables Apple Mail plugins. But with SpamSieve, an update is promptly made available, generally within a day. I value that kind of support strongly.
Thanks to authorized Mac dealer B&H Photo, a 4-core 3.5 GHz Mac Pro is arriving at MPG for testing next week.
Note that B&H Photo has substantial savings on AppleCare versus the Apple price.
This is the base model, the least expensive and it ought to make an excellent reference for all the other models. See the test results for the 2013 Mac Pro. In particular, how does a fast 4-core CPU stack up, and how well do the D300 GPUs perform on GPU-intensive tasks?
Received my new Mac Pro from B&H last week. Everything works great and I’m beginning the process of transferring over from my Windows PC.
When I ordered my Mac Pro from B&H I also ordered AppleCare. The problem is that Apple started the 3-year AppleCare clock on the day I pre-ordered which was Dec 19, 2013. The unit did not ship until Jan 17, 2014. I actually received it on Jan 22, 2014.
My AppleCare "Welcome Kit” was emailed directly to me from Apple and indicated a Dec 19, 2016 Coverage Period End Date. Basically, I lost an entire MONTH of coverage.
I contacted B&H about the issue and they suggested I contact Apple to get the start date corrected. I spoke with an Apple technical support person on the phone and he agreed that the date should be changed but, he said only Customer Service could do that. He put me on hold, spoke with customer service, came back and apologized. He was told that AppleCare ALWAYS starts on the order date. That’s Apple policy and cannot be changed. Have a nice day.
I called B&H Customer Service back and told them what Apple told me. B&H said they would handle it... and they did! My AppleCare agreement now shows a Coverage Period End Date of Jan 18, 2017.
You can check AppleCare status here:
Please advise your readers to contact B&H if their AppleCare start date on their new Mac Pro is incorrect. Kudos to B&H for OUTSTANDING customer support!
I experienced a similar issue, but I ordered directly from Apple. I received my MacBook Pro with Retina display on Dec 31, having ordered it on Dec 24; it was shipped on Dec 27. The AppleCare term as per Apple expires on Dec 27, 2016. It’s only a loss of four days (ambiguous, “expires” means that day at 00:00 or does it include that day?). Anyway, the warranty term should start upon delivery, not upon shipment.
Isn’t it obvious that a computer shipped from Shanghai, China cannot be delivered to the USA the same day? How can a warranty even start if the user has not yet obtained the product? This is not an honest way to do business. If it’s Apple policy, it sounds like fodder for a class action lawsuit. It stinks in any case.
There is no reason that products shipped from Apple cannot apply the delivery date as the start date of the warranty.
The B&H case is much more troublesome (pre-order date). That Apple would not adjust when the date of delivery can be demonstrated is dishonest at best.
For my Mac Pro, I’m going to have OWC do the CPU upgrade, as I’m good at losing small screws and similar. However, it’s as yet unclear when OWC will offer a CPU upgrade service.
Also, to prevent bending the memory latches (I’m guilty), OWC now ships a plastic spudger tool with each kit:
OWC 1866MHZ DDR3 Kits are available now for upgrading all of Apple’s new ‘2013’ Mac Pro 3.5GHz 6-Core, 3.0GHz 8-Core, and 2.7GHz 12-Core models to up to 64GB.
The Spudger tool now included with all OWC Mac Pro Upgrade kits is the same type of tool used by Apple technicians to aid in the disassembly of iPods, iPads and iPhones without damaging or scratching any other component.
This is a great video processing machine.
HAS BEEN SOLD
My photographic work as well as taking Mac Performance Guide forward mean that it’s time to sell my 2010 Mac Pro, which has been my solid and stalwart workhorse since 2010:
- August 2010 Apple Mac Pro 'Westmere', upgraded to 12 core 3.33 GHz CPUs.
- 80GB memory as 6 X 8GB + 2 X 16GB.
- ATI Radeon 5870 video card + spare NVidia GT120 video card (not installed, but included).
- 3 X 1TB hard drives in internal bays (freshly tested, no remapped blocks).
- Configured with one Apple 120GB SSD (nearly new) as boot drive with fresh system install in Bay 1 in Apple drive sled. Extra 120GB SSD included.
- FirmTek 4-port eSATA card (installed).
- FirmTek 5-bay eSATA port-multiplied (single cable) enclosure with 5 empty drive carriers.
- OWC Mercury Elite Pro quad-interface (FW800, eSATA) external drive enclosure.
- Keyboard and mouse and power cord, shipping in original box.
- Firewire 800 CompactFlash card reader.
- Four extra internal bay drive sleds (handy for swapping drives in bays, e.g., for backup).
- Negotiable: NEC 3090WQXi 30-inch display (dual-link DVI-D, predecessor to the PA301W).
System available around Feb 14 (depends on arrival of my new system).
All offers over $3500 considered. Buyer pays FedEx 3-day insured shipping or local pickup. Cashier's check or cash (if local sale). Contact.
An OWC 64GB memory kit was installed for the tests.
Test results have all been updated to include the 2.7 GHz 12 core 2013 Mac Pro with D700 GPUs.
The Unigine Valley application uses 3D rendering. Here the D700 GPUs show a clear speed improvement over the D500 GPUs. Regrettably the Valley application uses only one of the two GPUs (like most applications as yet).
The test results have been updated to include the 2.7 GHz 2013 Mac Pro in about half the tests; the remainder will follow shortly.
So far, the 12-core and the D700 GPUs are both duds or only marginally better than the D500 GPUs with most photography applications.
Put another way, you’d better have a darn good reason for going past 6 cores and/or getting the D700 GPUs. The sitution is fluid and might well evolve over time, but for now it’s clear that the 6-core is the ideal machine for most users, and that those looking to eke out modestly more performance should go with an 8-core 3.3 GHz CPU upgrade (from a 4 core).
Bleeding edge—the D700 GPUs show a curiously inferior performance on some tests to the D500 GPUs. Barefeats.com reports a similar confounding result in the FCPX Render Gaussian Blur test and others (D700 GPUs test slower than D500 GPUs). Perhaps it is a drive issue or other issue that will get fixed, because the Unigine Valley demo, shows frame rates about 50% higher with the D700 GPUs.
The test results have been updated to include the 2.7 GHz 2013 Mac Pro in about half the tests; the remainder will follow shortly.
The only place the D700 GPUs impresses so far (versus the D500 GPUs) is the Unigine Valley demo, where frame rates are about 50% higher. Regrettably the program uses only one of the two GPUs (no program tested here at MPG uses both as yet).
When software of any kind is added to a system, a new vector for infection or malware or exploit is part of the deal. Sometimes it’s the new software itself, and sometimes there are things that combine to create a new weakness.
It’s never a good idea to add extensions you do not need. Ars Technica reports on a new trend in Adware vendors buy Chrome Extensions to send ad- and malware-filled updates.
Arriving tomorrow, the 12-core Mac Pro with D700 GPUs will offer a look at performance with a variety of photography applications, pitted against the 6-core and the 2010 12-core 3.33 GHz Mac Pro. The test graphs will be updated to include results starting tomorrow.
DxO Optics Pro shows outstanding scalability across CPU cores.
This graph will have the 2013 12 core Mac Pro added soon. It is a fascinating study in both CPU core usage and efficiency of the 2010 generation of Mac Pros vs the 2013 generation.
DxO Optics Pro deserves kudos for utilizing CPU cores in near-perfect scalability. Best of breed in a technical sense; no more efficient raw conversion out there in terms of CPU utilization. Now if only it could use the dual GPUs and cut that clock time down from ~5 seconds per file to ~1 second per file.
The 2013 Mac Pro is an expensive system to assemble. It’s not just deciding which Mac Pro variant is best for a particular workflow, but also the type and capacity and speed and build quality and fault tolerance of external storage and backup. All of which can matter a great deal for some workflows.
The wrong choices can mean spending more money for less performance, inadequate storage capacity, confusing backup, or workflow configuration that does not utilize the best approach.
A complete system is not likely to come in under $6K even for the base model 4-core: a 4-core with with 512GB flash and 32GB and D500 GPUs is already $4199 and that’s before a display, AppleCare, external storage, backup drives, etc.
Professionals in particular might look at all these costs and wonder if they are making the right decision. For that I offer personalized consulting.
B&H Photo also has substantial savings on AppleCare versus the Apple price.
The 6-core is the “sweet spot” model for most uses and users as the tests show. The choice of SSD (flash size) speaks to particular needs (I prefer 1TB, but 256GB is ample for many uses).
Version 5.5.3 of Sigma Photo Pro is a 32-bit program, and runs like molasses on all machines, taking 7.9 seconds per file on the fastest Mac Pro.
Click for larger image.
Apple Aperture might be expected to utilize the dual GPUs of the 2013 Mac Pro. It’s fast but it’s also unclear if it does so.
There is quite a long backlog at Apple, yet the 6-core is readily available at B&H Photo as I write this. Don’t forget to upgrade the 32GB or 64GB memory at OWC.
Reader Bill reports:
Just got an email from B&H that my new Mac Pro is being drop shipped directly from Apple to me today. 6-core CPU, D500 GPUs, 512GB SSD, 32GB RAM. I placed my order with B&H on Dec 19.
B&H also included a great deal on AppleCare and free Parallels Desktop 9.
Until Adobe implements GPU support for more of the no GPU benefit to a good number of the filters with Photoshop CC 14.2.menu, it’s clear that there is
It’s not a numbers thing so much as what is used in any particular workflow.
Once a Mac Pro system is purchased, then what?
Photographers and professionals need to think about a l0t more than CPU and GPU choices; how to configure for optimal workflow, working space vs main storage, backup and fault tolerance, how to evolve storage as needs grow, software choices and workflow efficiency, color calibration, and so on.
Consult with MPG on the best approach for your needs.
Testing continues on the 6-core and 8-core 2013 Mac Pro.
Preliminary conclusions on CPU and GPU choices for photographers:
- The ideal system for photography is the 6-core. The only choice is whether to go with the stock Apple 3.5 GHz, or to aim for a small bonus speed increase with a 3.7 GHz 6-core via CPU upgrade (or 3.3 GHz 8-core).
- The Apple 8-core at 3.0 GHz is unlikely to be any faster than the 3.5 GHz 6-core for the vasty majority of photographic applications, and will be slightly slower for some tasks: it’s not an upgrade over the 6-core.
- The emergence of huge speedups via the GPUs holds great promise (as yet unfulfilled in Photoshop). Still unclear is how much difference the D300/D500/D700 GPU choice makes even for functions that use the GPUs (no system to test yet), but it looks wise to make the D500 choice over the stock D300. Whether the D700 choice can bump up real-world gains is unclear as yet, but those who want the fastest should opt for the D700 GPUs and hope for more GPU support.
PhaseOne CaptureOne Pro offers “hardware acceleration”, meaning the GPU(s).
The late 2013 Mac Pro handily outperforms the 2010 Mac Pro, even a 12-core 3.33 GHz monster with the ATI Radeon 5870 video card.
But what about the “sweet spot” of 6 CPU cores at 3.5 GHz (Apple stock CPU) vs an 8-core CPU at 3.3 GHz (OWC upgraded CPU)?
Not sure what to get, how much memory, drives, backup, storage, etc? Consult with me about your new Mac Pro system.
The 6-core 3.5 GHz CPU is an outstanding performer (as predicted). It should match the 8-core 3.0 GHz on most things, based on testing so far.
Hence I have canceled my order shown below in favor of the 4-core 3.7 GHz Mac Pro with D700 GPUs—to be modified to a faster 6-core or 8-core. See discussion below.
This order for a 4-core has been in place since January 6th, so it is in the queue. My reasoning for going this way:
- I want to test a 4-core and this gives me that opportunity.
- I want to test the D700 GPUs, and Adobe’s recent first and very excellent (crashes aside!) stab at using GPUs suggests that the D700 option might pay off even more. It’s probably not worth it for most users, but I use Photoshop a great deal, and I doubt I'll get a hold of a D700 GPU model any other way.
- The 4-core is the most cost effective model for upgrading the CPU to either a 3.7 GHz 6-core or a 3.3 GHz 8-core.
My needs call for the fastest possible system, hence the upgrade path. The choice between those 6-core 3.7 GHz and 8-core 3.3 GHz is still pending additional testing. I suspect that the 6-core 3.7 GHz will be just as fast as the 8-core 3.3 GHz on almost everything, based on testing so far, and perhaps even a little faster on some tasks.
It has become clear that the D700 GPUs might be worthwhile now that support is emerging for huge speed gains in certain areas—more is likely to come. But how much more the D700 GPUs will offer over the D500 GPUs is uncertain as yet (specs vs real world usage).
With the release of Adobe Photoshop CC v14.2 as of Jan 15, 2014 we now have an existence proof of the huge potential for speedup using the GPUs. If more functions follow this path, this could be the biggest gain in performance in a decade or more.
But it’s not just the 2013 Mac Pro that’s a winner; the 2010 Mac Pro sees similar gains, at least with the NVidia 5870 video card, which has dual GPUs on board.
But are both GPUs being used, or just one? Not yet clear.
Nice job, Adobe!
Update March 7 2014: this improvement has apparently been reversed—OS X 10.9.2 and Photoshop CC 14.2.1 is about 3X slower (with GPU) than shown here, yet CPU-only test is identical.
Themenu in Photoshop contains many different functions, but in total it is used heavily by most Photoshop users.
PhotoZoom Pro 5 support GPUs via OpenCL explicitly and even has an optimization function for configuring for best performance.
My question to BenVista is whether both GPUs are utilized.
One thing the 2013 Mac Pro is rather nice for: the warm air at top is a pleasant hand warmer (my office gets cold in the morning). Just keep it close by your keyboard, and you’re Good, for one hand at least.
Also, at least with the lighting in my office, the polished finish has all the allure of black plastic covered with saran wrap. Looks good on most trash cans.
Bad timing — this morning I woke up to find that Adobe Photoshop CC has been updated to version 14.2. So total redo on all Photoshop benchmarks to be sure the number reflect the current version. Version number in graphs should read “v14.2” for the latest results.
The good news is it might address some Filter menu bugs (dimmed, unavailable). The bad news it now crashes in the Shake Reduction and Lens Correction filter 100% of the time (which is good news in that a fix should be easy). But it means that I have to drop it out of my sharpening test suite.
Update: crashes seemed to be confined to the 2013 Mac Pro and do not occur on the 2010 Mac Pro; maybe it is a GPU support code issue.
Could it be related to the new D500 GPUs with 3GB VRAM each?
Update 2: also crashes in Lens Correction filter (maybe others too). Photoshop CC 14.2 appears to be an unstable release to be avoided by 2013 Mac Pro users.
Update 3: CONFIRMED—this is a GPU-related bug specific to the 2013 Mac Pro. With the GPU disabled on the 2013 Mac pro there are no crashes. No crashes either way with 2010 Mac Pro.
Update 4: all benchmarks now updated to Photoshop CC 14.2. A few things had to be removed due to the crashes.
Many thanks to Other World Computing for loaning the 8-core 3.3 GHz Mac Pro for testing. I have confirmed with OWC that a CPU upgrade program for the 2013 Mac Pro is in the works, and that pricing will be competitive on higher performance CPUs than Apple offers, indeed there might be savings.
For my own needs, it looks like I’ll be buying and upgrading a 4-core Mac Pro to the 8-core 3.33 GHz (cancelling my prior 6-core 3.5 GHz order most likely, but this is not 100% decided until tomorrow’s testing). If you plan to upgrade the CPU, then it makes sense to get the 4-core with 1TB SSD and D500 or D700 GPUs. OWC will rebate back whatever CPU you start with however.
My thanks also to authorized Apple dealer B&H Photo for loaning the 6-core 3.5 GHz Mac Pro which arrives tomorrow.
Hmmm... buy a Mac Pro at B&H, then ship it to OWC for a CPU upgrade and 64GB memory upgrade and some backup drives and add-ons. Sounds like a smart move for those looking for the strongest performance possible. Stay tuned for an updated MPG Pro Workstation program.
Thank you also for subscribing to my photography publications.
Choose the best number of CPU cores for your actual work: 2013 Mac Pro: How Many CPU Cores for Your Workflow?
Yesterday I wrote on the mediocre scalability of Adobe Photoshop CC, which means that more than 6 CPU cores amounts to negligible gains, which is why I’ve been recommending 6 cores for Photoshop for years now—sporadic improvements by Adobe have emerged, but scalability remains modest.
While the disappointing scalability of Photoshop CC might see some influence from memory bandwidth, it almost certainly has a lot more to do with structural design issues in the code.
But does perfect or near perfect scaling to 12 cores exist in practice? In other words, will some programs get the job done with 12 core in half the time as 6 cores? Absolutely—I’ve written such code myself in Integrity Checker, though it’s throughput is usually I/O limited.
So here is an existence proof with PhotoZoom Pro 4 showing that perfect scalability to 12 cores* exists. One wonders what dual CPUs with 24 cores might offer for such tasks, which take a long time to run.
* 12 cores meaning 12 cores with hyperthreading = 24 virtual cores.
Thanks to authorized Mac dealer B&H Photo, a 6-core 3.5 GHz Mac Pro is coming to MPG for testing on Jan 16, as shown below. While the SSD (flash) drive is 256GB, it is otherwise my “sweet spot” model for most uses and users*, unless price is no object for squeezing out performance.
Note that B&H Photo has substantial savings on AppleCare versus the Apple price.
Choose the best number of CPU cores for your actual work: 2013 Mac Pro: How Many CPU Cores for Your Workflow?
Don’t decide on CPU cores based on Photoshop CC alone unless that is your (only) usage. Some programs offer better scalability.
It appeared that Adobe Photoshop CC was using my 12-core Mac Pro really well based on CPU usage, which was exciting. I thought that perhaps Adobe had made some progress and/or that OS X Mavericks had improved matters. But I was fooled by CPU usage, which is indicative but not definitive of getting useful work done.
I spent some careful hours measuring the actual performance, and it turns out that the CPU usage is mostly overhead busy-work (thread contention), not useful computation.
What matters is not CPU utilization, but useful work finished: do more cores get the work done faster?
For example, if a task takes 30 seconds with 6 CPU cores, we would like it to finish in ~15 seconds with 12 cores (perfect scalability), or at least something like ~18 seconds which would still be a large improvement. But Photoshop CC falls short of even that reduced goal, as the asymptotic curves make plain.
What is the cause? It could be memory bandwidth, in which case the 2013 Mac Pro will offer better relative performance in 8/10/12 core range versus 6 cores. How much better is unclear until testing can be done. Or it could be synchronization overhead, which is a function of both the problem space and the code design quality (efficiency)—unfixable except by redesign by Adobe if that is the case.
UPDATE: added is a graph with normalized times and theoretical best results.
Update: my enthusiasm was misplaced because I was fooled by CPU usage. It turns out that the CPU usage I was seeing was mostly overhead busy work (thread contention), not useful computation.
CPU utilization is not the same as useful work accomplished. See the real story.
In the next few weeks I’ll be doing an all-new investigation of the use of CPU cores in Photoshop in particular, because things have changed (not sure when exactly), and much for the better.
Today I was researching how well CPU cores are used in the latest Adobe Photoshop CC and I’m pleased to report that some old rules can now be forgotten: I am seeing that 12 cores handily outperform 8 cores or 6 cores even on common tasks like Smart Sharpen.
I’ve pre-tested my current 12-core 3.33 GHz 2010 Mac Pro by disabling CPU cores from scheduling, in order to directly compare on the same machine. Those CPU cores all run at their full speed of 3.33 GHz, so the actual behavior remains to be seen on the 2013 Mac Pro, where use of more CPU cores ramps down the clock speed for all of them.
Choose the best number of CPU cores for your actual work:
What ends up being my favored choice for a drive enclosure will necessarily be driven in part by whether the auditory characteristics are acceptable to me (I have no nearby closet or other space to run a long Thunderbolt cable into).
Who cares about fan noise?
It took me decades to understand that my sensitivity to sensory stimuli such as fan noise was heightened relative to most people: things that bothered me seemed to not even be noticed by people I worked with.
I share this here because for a minority of the population like me, it can be baffling until understood, and even when understood, it can be a frustrating barrier to communication (a very different perceptual basis: to use an analogy, how would you explain red and green to a person partially r/g colorblind?). That is what took me so long to figure out.
Over a period of years I gradually came to understand that an innate difference existed, but I never understood it explicitly until I came across the epistemological explanation: I am a highly sensitive personality or HSP), and I know one of my daughters is—anyone who has multiple children knows how different they can be right from day one (well, I suppose they could all be very similar by chance).
The personality difference creates barriers to communication; people of normal sensitivity seem completely baffled by what seems an unreasonable emphasis (by me) on things like fan noise (though it is but one example).
A personal note especially for parents and teachers...
My lifetime of experience tells me that non-HSP people have seriously difficulty understanding the difference, let alone empathizing with it; the superficial and erroneous characterization is typically “shyness”. Both my daughter and I have been characterized that way by teachers over the years, which reveals a serious gap in their professional training that impairs their effectiveness, or worse. After all, teaching is first and foremost about appropriate communication. But neither of us is shy or insecure—far from it. How many HSP children are labeled this way and so acquire the label and are stunted by it by its repetition in assembly-line schools? I don’t know, but the very idea is troubling. I never was shy, but was labeled as such for years. It wasn’t until high school that the school counselor took note. Finally, the personality tests I took revealed a clear picture in sharp contrast to the prior 9 years of misunderstanding.
I’ve been doing some research on how well CPU cores are used in the latest Adobe Photoshop CC and I’m pleased to report that some old rules can now be forgotten: I am seeing that 12 cores handily outperform 8 cores or 6 cores even on common tasks like Smart Sharpen.
Update: this turns out to be incorrect: higher CPU utilization can also mean synchronization overhead or memory bandwidth limitations. Photoshop quickly runs into non-scalability issues beyond 6 cores.
I’ve pre-tested on my current 12-core 3.33 GHz Mac Pro by disabling CPU cores from scheduling, in order to directly compare on the same machine.
In the next few weeks I’ll be doing an all-new investigation of the use of CPU cores in Photoshop in particular.
Choose the best number of CPU cores for your actual work.
Maybe some day Apple will let us show real CPU cores (12 here) instead of virtual cores (24 here), because it’s a darn mess. There ought to be several views: all virtual cores (24 rows), real cores + virtual together (12 rows), and similar. I'd also like to see an average and peak figure next to each core.
I have a fine art reproduction business, and we live in Photoshop all day long. It's all we do. Our 4x5 digital scan back camera produces 1GB files, and we crunch those files in Photoshop all day long. It's all we do.
So as I'm looking for the right configuration to purchase for the new 2013 Mac Pro, I ran across your article. Money is no object. It's all about work flow and efficiency. We open and close files that are 500MB to 1.5BG all day long. So no matter the cost, I want the fastest machine I can buy for Photoshop.
Your article helped me immensely. It's the best read I've found so far. I'm still not sure if I'll go 6 core or 8, but it appears 6. And I'm not sure which GPU, but it would appear dual AMD D500's make more sense than the D700, which may not be fully utilized for PS.
Certainly you've convinced me that the 12 core is a waste of time.
Thanks for a fantastically written article.
MPG: I appreciate thanks—always welcome. This site is time intensive and costly to stay current (MPG pays for its gear), so I appreciate support for our advertisers such as OWC, and please subscribe to my publications or use my buying links for Macs and photography gear. I am also available for consulting to help decide on a system design.
Photoshop is one application but it has many functions: CPU usage and GPU usage can vary for each function, so there is no entirely right answer. In short, the fastest machine ideally will be the one that handles the peak loads notably faster while being no slower on common operations.
Sometimes there are freebies no matter what machine is used: , simple optimizations that can speed up saving and opening files by up to 20X, that is, assuming a fast SSD.
Based on the description of intensive usage over long periods of time seen above, I would advise a 3.33 GHz 8-core (upgrade the CPU from a 4-core), the 1TB SSD and the D700 GPUS. However, the 6-core with D500 CPUS should run at very similar speeds, unless GPU usage is intensive, in which case D700 GPUs might be a nice bump up. Again, the actual workload might make some tasks noticeably faster and others might show no difference.
The unknown factor: Adobe is not likely to use the 2nd GPU initially, but we can hope that later this year both Photoshop and Lightroom will be updated to support use of the dual GPUs.
Paul I writes:
I appreciate your answer about not partitioning my 1 TB SSD, so there is only one volume on the computer.
In Windows, documents are by default put into various user folders. I found it more convenient to have a separate folder for photos (I use Lightroom to organize them), and sub-folders for other data, such as Word docs, PDF’s, etc.. Would that make sense for the MacBook Pro, or would you suggest using the default locations for documents.
MPG: I prefer to locate my data in a place of my choosing on a separate drive*, so that if the system install goes wonky I can boot off something else, clone back over it and resume, all without worrying that my main data is unaffected. Fast, easy, no-brainer solution, unlike Apple’s restore partition**.
On my desktop systems, the startup volume is always Boot and all of my important data is on Master. Those two volumes are backed-up via cloning to BootClone and MasterClone, whose volumes are show below. Note that Boot and Master could be two volumes (partitions) on the same drive, or two volumes on two different drives.
* The “Documents” and “Pictures” folders are nuisances that assume a single volume. They become useless and indeed confusing as soon as any real needs evolve for more storage.
** The Apple restore partition is fine for minor repairs, but a time wasting exercise with data loss of various kinds, if only calendar and preferences and re-entering serial numbers for soft (but often a lot more). A bootable clone is vastly superior in every case, because the system can boot right up and be back in business. Then just wipe the internal system and clone over it.
For my travel laptop with a relatively small SSD of 500GB (usable), the wasted space is not acceptable; I’ll bump up against limits on some trips if I bifurcate the space. So I do not partition for my travel laptop (though I might if it had a 1TB flash drive.
The core issue is that SSDs are expensive on a per GB basis and that partitioning results in some wasted space: one partition might fill up while the other has plenty of space. There is no quick and easy way to fix this other than redoing the partitioning, which means a backup/restore cycle. Alternatively, migrating data from one to the other—both things are time-wasters that I don’t want on my radar.
With SSDs (flash drives) of 512GB and smaller, the wasted space is too much in my view. With a 1TB SSD, the issue diminishes.
The trick is figuring (in advance) how much to allocate to each partition.
Usually the system and quite a few applications need no more than about 50GB if one properly separates data onto Master, but various caches and so on can consume some space, so anything in the 80 to 120GB range is a good choice
My recommendation for a 1TB SSD and for those following good “data hygiene” of properly separating system and applications from data is to use an ~100GB Boot volume and a 900GB Master volume. This is likely what I’ll do for my new 2013 Mac Pro.
One of the benefits of Thunderbolt 2 / Thunderbolt is the ability to place noisy gear a good distance away from the working area—up to 30 meters / ~100 feet.
OWC now offers Thunderbolt 2 / Thunderbolt cables up to 30 meters in length.
At 10 meters and longer optical fiber is used instead of copper, and this makes the 10 meter and longer cables much for expensive.
See Us at CES Booth #30472
Other World Computing Thunderbolt™ Connectivity High-Performance Copper and Optical Cables Now Available up to 30 Meter Length
Starting at $27.99, wide choice of OWC top quality Thunderbolt cables enables maximum flexibility for placement of Thunderbolt devices in home or office
Las Vegas, NV—January 10, 2014—Other World Computing, (OWC®) http://www.macsales.com, a leading zero emissions Mac® upgrade and storage technology company, today announced at CES 2014 that it has rounded out its array of OWC Thunderbolt cable offerings with three new optical cable lengths of 10M, 20M, and 30M. Longer length option allows noisy storage appliances to be moved to equipment closet for a quiet and productive workplace.
Perfect cables for Thunderbolt-based production workflows and data needs for today and beyond
• Utilize latest Thunderbolt chipsets for high-speed 10Gb/s Thunderbolt and 20Gb/s Thunderbolt 2 devices
• Enhance video workflow with support for faster 4K video transfers + 4K display capabilities via DisplayPort 1.2
• Able to Daisy-chain up to six Thunderbolt devices
• Can be used for Target Disk Mode between two Macs that support Thunderbolt
• Enables new iMac® as a display for a MacBook Pro® equipped with Thunderbolt
• OWC One-Year Warranty
• Enhances video workflows with support for faster 4K video transfers + 4K display capabilities via DisplayPort 1.2
OWC Thunderbolt cables are built to the highest standards for use with 10Gb/s Thunderbolt and 20Gb/s Thunderbolt 2 devices supporting the very latest Thunderbolt and Displayport capabilities.
Available immediately in six different length choices; click here for the entire line of OWC Thunderbolt Cables, starting at $27.99:
OWC .5 Meter Thunderbolt Cable $ 27.99
OWC 1.0 Meter Thunderbolt Cable $ 34.99
OWC 2.0 Meter Thunderbolt Cable $ 39.99
OWC 3.0 Meter Thunderbolt Cable $ 44.99
OWC 10 Meter Thunderbolt Cable $319.00
OWC 20 Meter Thunderbolt Cable $619.00
OWC 30 Meter Thunderbolt Cable $899.00
- Terminology: Volume vs Drive
- How to Partition Drives Into Multiple Volumes
- How to Initialize or Erase a New or Existing Drive
- How to Create a Software RAID-1
- How to Create a Software RAID-0
See also the chronological index.
Mouse over the buttons to see the sequence below.
The prior OWC post on CPU upgrades had an error in one case, and I let OWC know.
Now the latest tests with upgraded CPUs match exactly what I’d expect from each in terms of theoretical best results:
- A single thread runs 2.5% faster on a 3.7 vs 3.5 GHz CPU (4.0/3.9 = 2.56%). That’s TurboBoost at 4.0 vs 3.9 GHz.
- The 8-core 3.3 GHz CPU is definitely the very best choice; it gives up almost nothing on single-core execution and yet offers a ~21% boost over the 6-core 3.7 GHz CPU when 8 cores are used. In the 3/4/5 core range, it’s going to be a wash, depending on actual task.
- The 10-core CPU 0ffers a modest advantage over the 8-core only if all 10 cores are used. Given software overhead and less than 10 cores at 100% usage it offers essentially zero value over the 8 core for real world usage. With 9 cores in use it will scarcely differ from the 8-core.
A key point is that these are the BEST possible improvements. Real world software has implementation inefficiences, involves the disk at times, etc. Hence the real question is what each CPU does for a particular workflow. That will be my focus when I obtain a Mac Pro for testing—real world metrics for real workflow challenges, including how these CPUs stack up against my 3.33 GHz 12-core Mac Pro (2010 model).
I ordered the MPG 2013 Mac Pro at 4:00 AM Pacific time Dec 19 (“opening day”), but as of Jan 8, status of “January” is there like some fixed graphic.
My mistake was ordering the 1TB SSD, which at the time of order moved it out from “Dec 30” to “January”.
It would be helpful to have it say “about a week”, “about 2 weeks”, or similar.
But “January” for 19 days comes across as “we have no clue what our production department is doing”. Maybe there is some production glitch and it’s not yet predictable what deliveries will be.
See discussion of upgrading the CPU in the 2013 Mac Pro.
The 8-core 3.3 GHz CPU is highly attractive for my photography work, offering an improved blend of high performance under both lightly-threaded and highly-threaded computing tasks such as my photography. Apple does not offer this CPU as an option.
Discussion here is premised on two key ideas:
- That OWC will in fact offer CPU upgrades for the 2013 Mac Pro. As this was written, that confirmation has yet to be officially stated.
- That the incremental gains are of value to the user; there really is no point to upgrading for mainstream use; know your own particulars or consult with me to discuss the value (or lack thereof).
Shown below is the MPG recommendation for the Mac Pro configuration to order for upgrade purposes. Recommendation is premised on a realistic need for the highest performance! It is not necessary for most mainstream uses.
- Upgrade the memory to 64GB at OWC.
- Order the D700 GPUs, because the D700 GPUS might be helpful in time.
- The 1TB SSD (flash) is not easily upgradeable and is the only internal storage available, so it should be ordered also.
Bottom line: the CPU upgrade will be expensive*, so skimping on the other components makes little sense in the context of the total system cost and a real need for high compute power. If your own usage can benefit from a modestly faster CPU, then top-end versions of the other components should also come along.
* OWC does not have CPU purchase pricing power like Apple does. Discounted CPU prices are highly unlikely.
I hope to have a chance to test a Mac Pro modified to use the 8-core 3.3 GHz CPU sometime soon.
For my uses, an 8-core running at 3.3 GHz (Turbo Boost 4.0 GHz) is better than even a 6-core at 3.7 GHz, because there is one task in my workflow that makes heavy use of all 12 CPU cores in my 2010 Mac Pro.
The extra expense of the 3.3 GHz 8 core CPU is a modest improvement that is clearly not worth it for most uses and users, but since I slog away all year long on my Mac Pro, anything that gives me a bit more grunt is worth it; my 12-core 3.33 GHz machine has served me well for 3+ years.
As shown below, the top green line is the 8-core 3.3 GHz CPU. It offers performance as good as any of the other CPUs up to 4 cores, a tad less than the 3.7 GHz 6-core for 5/6 cores, but it extends its performance out to 7 or 8 cores thus providing about 1/3 more computing power over the 6-core CPU for tasks that can use 8 cores. Plus it has 25MB cache memory, versus 15MB for the 3.7 GHz 6-core vs 12MB for the 3.5 GHz stock 6-core.
The 10-core CPU looks like an interesting alternative to the 12-core, offering notably higher performance up to 10 cores in use.
The good news is that OWC has found that both the 3.7 GHz 6-core and the 3.3 GHz 8-core are viable upgrades, and the 10-core 3.0 GHz also.
Backup is a critical consideration for all computer users.
Two articles have been updated to discuss the value and how-to of cloning:
With the advent of multi-drive Thunderbolt storage systems and software-based RAID-5, RAID of various kinds especially the fault tolerant aspects become more interesing than ever before due to easy of connectivity.
See the discussion of RAID.
I always validate my drives before putting them into my 'production' use because I don’t like surprises down the line: most hard drives are shipped without the media being tested. Moreover, a good hard drop off the delivery truck can damage a drive.
Recently MPG purchased five Hitachi 7K4000 hard drives as test drives (as shown, 6th drive was a replacement for one bad one). It was especially important that they perform similarly becuase they might be tested together in RAID-0, RAID-1 and RAID-5 configurations. To have one slow drive slowing down the others would be an issue in general, and particularly so for testing accurately.
How does the FirmTek ThunderTek/PX Q6G over Thunderbolt compare to native eSATA on the Mac Pro?
Review of the FirmTek ThunderTek/PX Q6G updated with results.
I was pleasantly surprised to see significant speed gains with the 2013 MacBook Pro Retina 'Crystal Well'.
Saving energy is always a plus (well, not always), and the reduced electricity use here at $.42/KWh or so here in California is highly attractive, but energy usage does not exist in a vacuum. I am speaking of the need for external expansion with everything. Is automobile mileage calculated with all seats and interior ripped out to save weight?
To use a 2013 Mac Pro as a workhorse like the older “less efficient” 2010-2012 models, drives have to be connected, compatibility and expansion boxes connected, etc. These devices have their own power draw, and chances are high that they are substantially less efficient for the same functionality.
For example, it is not efficient to have 3/4/5 power bricks powering a gaggle of devices. It is efficient to have one very high-efficiency power supply making power for the computer, various internal drives, etc.
Moreover, even assuming that the external devices themselves are as power efficient as the Mac Pro (highly unlikely), power bricks can be “vampire” devices that continue to suck power even when the device is not in use.
The only intellectually defensible way to assess power usage is to add up all the power costs for an equivalent configuration of the prior model vs new model, e.g., with all the SSDs and drives and so on needed.
Of course, there is variation in setup. Presumably a few users can get by with only the internal SSD in the 2013 Mac Pro, and can use storage on a NAS and have their IT staff do backups. Or something like that. But most Mac Pro users will need Stuff attached.
In my 2010 Mac Pro, I have 2 internal hard drives, 2 PCIe SSDs, 3 SATA SSDs, and an eSATA card. These are all powered by the 2010 Mac Pro tower’s power supply. I use them all—heavily.
In the 2013 Mac Pro, all of my devices now have to be relocated into external enclosures. The power draw will rise for those devices (all those enclosures and fans and power bricks), canceling out some of the savings of the 2013 model. I expect lower total power draw (I hope).
I have a WattsUp power meter, so I can make some estimation of the differences, but my 2010 Mac Pro is also a 12 core with 80GB and so comparing that to my planned 6-core with 64GB itself isn’t entirely fair. Still, the idle power usage is of some interest, more so if the performance is similar.
I decided to run some Photoshop speed tests on the new MPG MacBook Pro, but I was having a devil of a time getting consistent results; I was seeing times literally twice or half as fast as the same test! Test results have to be highly consistent to be of any value.
I was intrigued: how could Photoshop run twice as fast some of the time?
The culprit turned out to be the App Nap feature in OS X Mavericks. As soon as the box was checked, all tests ran at full speed and with great consistency.
Apparently App Nap must sense user activity (mouse or keyboard) and if not seen, then nap. Or maybe it’s just a bug, since OS X Mavericks is still alpha quality months after its release—rife with new bugs.
Running a script in Photoshop doesn’t need user interaction (and must not), hence the system takes naps in the middle of the job, in this case benchmark tests!
This is a serious bug for anyone who regularly runs a job-based workflow (think Lightroom or Photoshop batch processing of all sorts and many other applications that might require some time to render or merge or whatever).
Check thebox for each such application to rule out such huge performance losses (2X).
This site is free for all. But the gear isn’t free (I have to buy it), and incredible time and effort are required.
A good way to say “thank you” is by subscribing to my photographic publications (and our advertisers will appreciate your business also).
See the quick primer on fault tolerance below the graph.
Formerly the province of hardware solutions, SoftRAID 5 is an impressive achievement in bringing RAID-5 (and and RAID 1+0) to the desktop with high performance, and using modest CPU resources.
SoftRAID 5 is in beta test. Users interested in a beta test should send an email to:
Faster speeds for writes known to occur with faster drives (e.g., 500+ MB/sec for writes). This graph for a 2TB volume generated from mismatched drives (two pairs of two drives) of different capacity and type, yet one could hardly complain about the speed.
With five fast drives (one parity drive), the performance quite possibly could reach 650 MB/sec, which is approaching the real-world limits of Thunderbolt v1 performance.
Quick primer on fault tolerance
Fault tolerance means that a part (e.g. drive) can fail without loss of data or functionality (replace the failed part ASAP).
The simplest fault-tolerance solution is a hardware or software RAID-1 mirror. For example, the OWC Mercury Elite Pro Dual Thunderbolt/USB3 solution can be configured as a RAID-1 mirror. One of the two drives can fail with no loss of data or functionality.
RAID 1 + 0 is a striped pair of mirrors, e.g., striping dual OWC Mercury Elite Pro Dual units. SoftRAID 5 can do this in software.
RAID-5 is striping (high performance) with one parity drive. For example, 4 X 4TB drives (16TB), yielding 12TB of usable capacity, with 4TB used as parity information (one drive failure can be tolerated).
RAID-6 (not supported by SoftRAID) is striping (high performance) with two or more parity drives. For example, 6 X 4TB drives (24TB), yielding 16TB of usable capacity with 8TB used for parity information (two drive failures can be tolerated).
Apple met its 2013 shipping deadline for the 2013 Mac Pro. It would be a lot easier to call it the 2014 Mac Pro!
I eagerly await my Mac Pro for testing as I continue with my assessment of Thunderbolt peripherals, at least a few of which are mandatory for the new Mac Pro in order to make a working system for more than basic use, and for many, absolutely essentially for transitioning expensive gear to the new Mac Pro.
B&H Photo carries the two MBP.Retina models that I recommend.
How fast is the internal PCIe SSD in the last 2013 MacBook Pro Retina 'Crystal Well? Fast.
The spikes are interesting behavior which might or might not mean something significant, such as performance when mostly full or after long term use.
A backup drive for travel is essential, but store it separately in case of theft of the laptop! The beauty of the OWC Mercury Envoy Pro EX is that it fits into the palm of my hand with or without its supplied soft cloth bag with short USB cable (powered from data cable alone). Which means it can go into most any pocket also or nook or cranny while traveling.
The computer-generated images of the 2013 Mac Pro don’t look anything like the real thing—seen in person at the Apple Store, the diminutive unit has a burnished look so smooth that it looks more like a glass sculpture than metal. In high-key lighting it picks up its surroundings, looking more gray than black.
I felt like a microfiber cloth should be shipped with the unit to keep it shiny, and I even felt hesitation in touching it for fear of leaving fingerprints. Very elegant, but time will tell what it looks like when scratched or what happens if one has the habit of eating lunch near it.
My aging Apple Airport Extremes (the flat MacMini-like units) were duds from the get-go several years ago, never performing well even 10 feet distant from the computer! But they had been getting more and more flaky and unable to deliver acceptable performance.
So today I visited the Apple Store today, and bought two of the newest model thinking to replace the two older ones.
Boy was I wrong (in a way)— a single unit of the new model covers all the range of the two older ones, and does so delivering 20 Mbps to a certain key location versus 1-2 Mbps for the old model. That’s 10-15X faster in my real world scenario. WOW!
B&H Photo has the Airport Extreme for about $185 (Apple Store price is $199).
Stopping by the Apple Store today, I took the time to view a variety of my site’s HD and UltraHD images* on the Sharp 4K display attached to the new Mac Pro (images are in the subscriber publications).
The Sharp UltraHD 4K display offers an expansive, almost immersive viewing experience, meaning it’s not just large (31.5-inch diagonal) but that entire space contains so much detail that the visual impact is compelling beyond any traditional display.
UltraHD images fill the screen with a wondrous level of detail that otherwise can only be experienced on a MacBook Pro with Retina Display, except that the Sharp has 3840 X 2160 pixels instead of 2880 X 1800, so you can see the entire UltraHD image. I did not assess the size of type or user interface elements.
The Sharp display has a somewhat lower pixel density and so doesn’t quite have the lustrous deeply rich vibrancy of the Retina display—close but not quite the same level. Then again, maybe the too-bright Apple Store interfered with my perception.
It’s not so sharp of Sharp to offer a low-resolution image of its display on its product page (even the View Larger size!). It’s tiny and looks blurred on a Retina display. Huh?
* The diglloyd.com publications include HD and UltraHD images for material added in the last 9 months or so, including examples and full aperture series. Subscribers also enjoy Retina-grade images in the diglloyd photography blog.
Better to wait
I’m going to wait on a 4K display because my needs involve true calibration (not faux calibration) and accurate color and grayscale over time. So I’ll be sticking with my NEC PA302W wide-gamut calibrated display for a good while more.
Moreover, there is no good pixel-doubled solution at 3840 X 2160 (I’m not too keen on 1920 X 1080 effective size and it’s not clear if this is properly supported as yet in any case).
The truly hassle-free solution is 5120 X 3200 or 5120 X 2880 which is 2X the current 30/27 inch displays. Those sizes allow for pixel doubling so that text and user interface is not too small, while retaining the working space of the current 30/27 inch displays for non-image usage.
End of the year, time to consider next year’s needs, and the Dec 31 “in service” deadline for tax deductibility.
And so I ended up unexpectedly paying for and receiving two MacBook Pro with Retina Display 'Crystal Well' laptops (I had thought that one would miss the end of year cutoff and ordered another to get it in time, intending to return the other). The new model is little different except for its GPU and lightning-fast SSD, but I have plans for my 18-month-old MBPR as a test mule. See the two MBP.Retina models I recommend.
As it turned out, fortune both frowned and smiled. Smiled in that both arrived Dec 31 and frowned in that one was defective.
The computer I had thought to keep turned out to have an entirely flaky problem with any internet connection, particularly a wired (via Thunderbolt to Ethernet adapter). So much so that I drove to the Apple Store to get a new Airport Extreme, thinking it was my 5-year-old unit, but that’s another (very good) experience.
Both units were spec'd identically and treated identically in all respects (setup, connectivity, etc), but the bad unit could not make an internet connection over the Thunderbolt-to-Ethernet adapter, or it would make a connection then lose it. I spent an hour or so trying to make it work—heck I do this all the time and I concluded it was defective (I swapped cables, adapters, etc doing all the single-variable things one should do).
Then I popped its cohort out of its box, and lo—it worked perfectly from the get-go and after numerous back and forth switching between two internet connections—zero problems with 2nd unit—it worked the first time and every time. So the bad unit goes back.
With a pair of the OWC Mercury Elite Pro Dual units on hand (for a short time), I was able to satisfy my interest as to some interesting configurations.
The review of the OWC Mercury Elite Pro Dual has been expanded to show:
- RAID-0 striping (high performance) using 4 drives (2 X 2 Independent).
- RAID 1+0 fault-tolerant stripe of mirrors.
MPG has five FirmTek multi-bay eSATA enclosures to contend with in the transition to the 2013 Mac Pro. These are high quality enclosures with hot swap trays and thus a big investment to preserve, not to mention many drives bolted into drive trays for those enclosures.
The FirmTek ThunderTek/PX Q6G makes the transitioning easy.
If you’re not looking to spend a bundle on a 2013 Mac Pro (and don’t forget transition costs), B&H Photo still has the 3.2 GHz quad-core for $2199.
The 2012 Mac Pro is a workhorse, and if you later want to take it further, OWC an upgrade it to a 6-core 3.46 GHz CPU.
It’s easy to upgrade the 2012 Mac Pro at reasonable cost, and you have a $1000 or so head start over the base-model 2013 Mac Pro: head over to OWC for Mac Pro memory and fast hard drives like the Hitachi 7K4000 or Toshiba HGST 3TB and you’re set for internals. Unless you also want the very best speed of all: the OWC Mercury Accelsior PCIe SSD, for things like optimizing Lightroom.
Conrad S writes:
After reading the reviews and weighing the options I decided to opt for upgrading my 2010 Mac Pro rather than buying a new one.
The numbers I was seeing from the new benchmarks, coupled with the incredibly high total cost of ownership, just didn't seem like a good fit for me. I still look forward to reading your review of the new machine once you've got one.
You're spot on there I think unless usage is specialized.
DIGLLOYD: the 2013 Mac Pro is a video-oriented machine. But its CPU speed is likely to be only marginally faster than the current model on many tasks. You get USB3 and Thunderbolt and fast GPUS and 4K video support and a measly 4 memory slots.
The machine-only cost is very high even by itself, let alone if you need to transition from the 2010-2012 Mac Pro to the 2013 Mac Pro. Including the Mac Pro itself, think $7K to $12K or more, depending). External storage, AppleCare, display, etc. It is very easy to spend $10K or more on a new Mac Pro system.
But wait, it’s not necessarily better.
Any new Mac Pro might be a downgrade for me
My current workhorse is a 2010 Mac Pro 12-core 3.33 GHz system with a gaggle of fast SSDs and 80GB memory and dual internal 4TB hard drives and various externals. It’s a serious expense (and mess) to transition all that gear.
While most of those 12 real CPU cores go unused most of the time in Photoshop, my most frequently-performed operation is sharpening with Topaz InFocus, which uses all of those real CPU cores and then some (virtual cores too).
Update: it turns out that Topaz InFocus does not scale. I was shocked to find (with testing) that 2/4/6/8/10/12 cores all run at the same speed. That is not a mis-print: high CPU utilization seems to be huge synchronization overhead beyond two cores. All that red in the graph at right is apparently spin lock or simlar thread contention overhead. The code in Topaz InFocus chews up CPU cores without getting more real work done.
And this usage is interactive: select a layer, sharpen, repeat for 6 to 10 layers. I do this sometimes hundreds of times a day if I’m pushing out aperture series. It is reasonably responsive on my existing 12-core Mac Pro , but it is the slowest part of my workflow.
So what happens to my workflow when I switch to a 6 core 3.5 GHz 2013 Mac Pro?
The new 6-core has a nominally faster clock speed, but Turbo Boost will drop down to 3.6 GHz when 6 cores in use. That’s 6 cores at 3.6 GHz versus 12 cores at 3.33 GHz.
Put another way, 21.6 GHz vs 40 GHz of 'grunt' on a compute-intensive task. Which machine will be faster? Memory bandwidth is better on the new model, so that might help a little, and there are some other processor efficiencies, but it’s a big gap to make up—the fact remains that my old model has nearly twice the GHz grunt of the new 6-core. So I’m nervous that the new 6-core could in fact seriously DEGRADE my present workflow. Unless those Topaz guys can eventually harness the dual GPUs for sharpening. Possibly an 8-core 3.4 GHz CPU would offer parity.
The 2013 Mac Pro does have a hugely expensive 12-core CPU option, but that model is 2.7 GHz though with Turbo Boost to 3.0 GHz—slower than my current Mac Pro in clock speed. It is not likely to outperform by much with a 10% handicap built in.
Users of existing 2010-2012 Mac Pros can upgrade to a 12-core 3.46 GHz for less money and a sub-minute CPU-tray turnkey swap-out.
Kudos to Apple for offering user-accessible flash storage (SSD), see Mac Pro (Late 2013): Removing and installing flash storage.
See updated discussion on 2013 Mac Pro: Choosing PCIe Flash Drive Capacity.
Upgrade solutions are likely to emerge by mid 2014 or thereabouts. Device quality, speed, longevity and capacity are all buying factors, and buyers are advised not to make assumptions on those points when upgrading.
Apple does not offer the peak speed CPU options in the Mac Pro. Shown in the table below is a slightly faster 6-core and a slightly faster 8-core CPUs. Due to Turbo Boost, the actual differences in speed are small.
These are not large bumps up over the stock CPUs, but the Mac Pro is such a big investment that eking out another 6-10% makes a lot of sense in the context of total system cost, particularly if the starting model is the 4-core and a down-the-road desire for 6 or 8 cores develops.
If the new Mac Pro has a socketed CPU (and Apple has not “locked down” the system somehow in EFI firmware), perhaps the low-end 4-core can be upgraded to a fast 8-core, which would be an appealing path down the road as CPU prices decline.
One never knows what those hard core elves over at OWC might be up to, given their CPU upgrade program for the prior model. But they’re taking one apart, which is the first step to investigating possible enhancements.
TPD of 150 watts vs 130 watts should be a non issue for any reasonable usage, and probably a non issue in any case with Apple likely over-engineering the cooling system.
The answer is not obvious for most users:
See also Algorithms and Performance.
The review of the OWC Mercury Elite Pro Dual has been expanded to show performance used with Thunderbolt as a hardware RAID-1 mirror.
My oldest teenage daughter got an iPhone for Christmas. Within a few hours: $8.40 in text messaging charges, giving me a heart attack, so to speak. The Apple iMessage app is simply “data” like any other app, but not SMS messaging. She now knows the difference, and had to inform all her friends not to SMS her. But that is not a fix, it’s a case-by-case headache and does not stop random junk or new friends, or plain old mistakes by friends.
We disabled SMS for outgoing, but it doesn’t stop incoming SMS at $0.20 each with messages like “hey”. It turns out that there is NO WAY to disable incoming SMS text messaging on an iPhone.
Note that if 5 million unwary users text at $8 for one day (or a few) as above, that’s $40M. That is quite a compelling business model.
I couldn’t call or chat with AT&T on Christmas day, but an online search found this aparently disturbing policy on the AT&T web site.
You cannot disable text messaging on your wireless account, as this service comes bundled with your wireless service from AT&T. All of our messaging-capable wireless phones come pre-activated to send and receive text messages at pay-per-use rates. It is not possible to remove text messaging from your wireless account or turn it off on your wireless device.
As stated: AT&T sets up phones to charge extra for the data you already pay for (by the language fiat of calling data “text”). Even an astute consumer has no way to disable it, by design. Wow. But see the update below—the statement is misleading or erroneous at the least.
But is it actually true that AT&T will not disable SMS messaging for a smartphone? I called AT&T the next day and learned some things, see Update below.
The choice is either an additional $20 a month for unlimited texting or the risk of an unbounded texting charge. Strong arm tactics. That is in addition to $30 a month for “data” (what is text if not data?!) and that’s on top of $9.99 a month for the phone in a family plan ($40 total). So AT&T really wants you to pony up $60 a month (plus fees and taxes). There is some kind of sharing plan alternative. But looking into it, it’s a one way street (cannot be reversed), and it raises my total costs for all my family phones. I feel like a steer in a cattle chute heading to slaughter.
There is a parental controls feature at AT&T. That will set you back $4.99 a month. In other words, you get charged so you can control stuff you don’t want. And then you still have to monitor things. Joy.
Few companies can get away with this kind of behavior (minimal competition is a real problem in the phone industry, for consumers that is). If the US congress wants to fix something everyone could agree on, this is surely a fertile area, referring to the steer noted previously.
But let’s hold a powerful player accountable: Apple makes enormous profits from its iPhone, and Apple has in its power the ability to include a “total SMS disable” feature on the iPhone. And Apple surely has the power to pressure AT&T to offer consumers a choice. Shame on Apple and Tim Cook (and predecessor) for being enablers. For a company seeking to build the best product on the market, this is surely rationalized-away cognitive dissonance (total customer experience), and a slap in the face to customers buying Apple phones. Silence is a moral sanction—a brief press release stating that Apple “disapproves” of such policies would be a powerful lever, and it would satisfy me.
I called AT&T on the 26th. In spite of the “unusually heavy call delay” advisory, I got through in two minutes or so. The phone representative was courteous and professional—absolutely no complaints there, I complimented her accordingly (I never a confuse a company policy with the person on the line!). The representative got the change made quickly, and even credited the pay-per-use texting charges.
My phone bill now under my control again, I felt relieved. But this kind of experience need not exist at all in the first place; customers have a reasonable expectation of being able to control their costs without jumping through hoops. It ought to be a control right on the phone, or at least online in the customer account. Having to call in is a hurdle, presumably by intent.
Account details with Test Message Opt-Out
YES, you can add text block and other blocks to your phone plan. The 'MobileShare' plans include unlimited texting, but other plans are pay-per-use, which was my issue above.
An account might look like this once the Opt-Outs are added:
The online opt-out support page
I read the opt-out text to the AT&T representative, who told me that this statement applies to a MobileShare plan, though the web page does not/did not say that when I visited it. Of course, she could have been as confused as I was.
MobileShare plans have unlimited texting, so why would pay-per-use be mentioned?
You cannot disable text messaging on your wireless account, as this service comes bundled with your wireless service from AT&T.
All of our messaging-capable wireless phones come pre-activated to send and receive text messages at pay-per-use rates. It is not possible to remove text messaging from your wireless account or turn it off on your wireless device.
This makes no sense: pay-per-user rates for unlimited texting? It’s a contradiction. One has to conclude that the statement is for pay-per-use non MobileShare plans.
Regarding MobileShare, it has its own implications (if-then-but-else-and stuff one has to look at carefully), and so is not a no-brainer for savings. For example, phones on MobileShare that are out of contract carry a $20 per month surcharge, and I have two. So one pays that surcharge, or upgrades phones. It is possible to sidegrade to the same model and take on the 2 year contract at less cost than, say, upgrading to an iPhone 5s. Point is, “savings” are not necessarily there when the carefully crafted (by AT&T) plan is examined. That analysis will vary for everyone. Nor is switching providers viable as a practical matter; my own remote travel needs for coverage cannot be determined by any coverage map.
The OWC Mercury Elite Pro Dual houses two hard drives in a high-grade aluminum case and offers Thunderbolt + USB3 connectivity along with user-selectable RAID modes or independent drives, and is bootable.
RAID made easy: no technical knowledge required.
Thunderbolt vs USB3
The graph shows that USB3 can handle dual fast hard drives very well, certainly no less well than Thunderbolt. This graph is for DiskTester fill-volume on the the first 1TB of a 6TB RAID-0 stripe of Toshiba 3TB hard drives.
Note that Thunderbolt writes are actually less consistent than USB3. The test was run twice and this is actually a little better for TB than the other run. But it should have no real world bearing.
USB logic steps:
Drive > SATA to USB controller > USB cable > USB host/chipset
TB logic steps:
Drive > SATA to PCIe controller > TB controller > TB cable logic > cable > TB cable logic > TB Host controller
Added to the 2013 Mac Pro coverage is a new section on transitioning the existing 12-core Mac Pro with all its SSDs and hard drives to the 2013 Mac Pro.
A transition overview is followed by 5 problem-solving pages. Total cost for hardware to compensate for the lack of no internal expansion approaches $2000 in my case.
My system is a little more expansive than most. It poses additional transition challenges to get two PCIe SSDs, three 6G SSDs, one 3G SSD and two 4TB hard drives—all internal now—connected to the new Mac Pro.
Make Lightroom or Photoshop FlyMaking an Old Dog of a Laptop Run Like a Young Puppy,
Make Lightroom or Photoshop Fly
Fast reliable SSD for laptop or other system
Or make Lightroom fly (put the catalogs onto the SSD).
I have used OWC SSDs for years now (3 of these puppies striped). I also use a pair of the even faster OWC Mercury Accelsior PCIe SSD in my Mac Pro (any Thunderbolt Mac can also use the PCIe version in the Helios case). The Mercury Envoy Pro EX SSD is fantastic for travel.
I do wonder why Apple won’t let us have three appealing CPU options.
- E5-2687WV2: 3.4 GHz 8-core (TPD 150 W), Turbo Boost to 4.0 GHz, 25MB cache
- E5-2667V2: 3.3 GHz 8-core (TPD 130 W), Turbo Boost to 4.0 GHz, 25MB cache
- E5-1660V2: 3.7 GHz 6-core (TPD 130 W), Turbo Boost to 4.0 GHz, 15MB cache
The 8-core chips in particular look appealing, but a 6-core 3.7 GHz / 15MB cache is appealing over 3.5 GHz / 12MB cache as well.
If the new Mac Pro has a socketed CPU (and Apple has not “locked down” the system somehow), perhaps the low-end 4-core can be upgraded to a fast 8-core.
TPD of 150 watts vs 130 watts should be a non issue for any reasonable usage, and probably a non issue in any case with Apple likely over-engineering the cooling system.
These are not large bumps up, but the Mac Pro is such a big investment that eking out another 6-10% makes a lot of sense in the context of total system cost—and slightly faster CPU options exist in MacBook Pro and iMac offerings.
Apple’s memory is always expensive to go to the higher config ($1200 for 64GB).
If you bought the 6-core Mac Pro as I did, then that 16GB trades in for $100 at OWC, effectively bringing the cost down to $795 for 64GB memory. Which in context of the system cost makes a lot of sense, though most users will find 32GB just fine*.
After you upgrade with OWC, save even more with OWC’s Trade-In Rebate for the original factory memory. Receive $75 for an Apple factory 12GB (4GB x 3) set, $100 for factory 16GB (4GB x 4) set removed from any Mac Pro 2013 model.
* If you want Lightroom to run lickety-split, get those catalog/preview folders onto a fast SSD, e.g. get at least the 512GB option, or add the OWC Accelsior PCIe SSD in the Helios enclosure via Thunderbolt, and host the catalogs there.
B&H Photo has the new Mac Pro available for pre-order.
B&H is our trusted photography vendor, but is also is a Mac dealer and generally has a much better price on AppleCare. Thanks for using this site’s links for photography or Macs, and see the gear lists.
Bill W writes:
As I mentioned, I ordered my 6-core Mac Pro from B&H yesterday morning.
This afternoon I received an order update.
It says: Status: Drop Ship order submitted to vendor
Sounds like B&H is handing their pre-orders over to Apple to fill directly. I like that! I prefer being in the "Apple queue" instead of the "B&H queue".
OS X Mavericks is the most bug-riddled release yet; every day various bugs bedevil me, like black flies in the high North. After a while, one mostly forgets about them, “scratching” without thinking. But they don’t ever go away; they hover patiently and stealthily move in. Black flies leave ugly red welts; Mavericks ingrains time-wasting habits to work around bugs*.
OS X 10.9.1 Mavericks makes one “fix” to Apple Mail (think “fix” as in corruption): shortly after updating, Apple Mail deleted my entire “VIP” list. Gone, never to be seen again. Not one entry. Start from scratch. This is a data loss bug, the most offensive kind. I can reconstruct that list (all 30 entries or so), since the original emails are not deleted (I hope), but in the context of the various other bugs that bedevil me every single day in Mavericks, I find myself wishing for the good old days of Mountain Lion.
So what do I do about it? Get my next fix with a new Mac Pro. Irritating is still better than awful.
Early signs of a company in decline include complete lack of awareness of deteriorating quality control along with construction of a corporate temple. And Tim Cook ain’t Steve Jobs. Big trains have lots of inertia in their favor.
* Finder list view and file open and save dialogs are badly broken in some views, for starters. They waste my time every single day.
I’m seeing some reader confusion on USB3 vs Thunderbolt.
For an external hard drive, Thunderbolt has no speed advantage over USB3.
Using Thunderbolt for a single hard drive is fine and indeed preferable (cable strength and durability), but to date Thunderbolt entails a cost disadvantage both for the enclosure and the cable itself (expensive). And when three or more drives are used in a RAID configuration, then Thunderbolt becomes valuable for its higher throughput and more robust cabling.
One thing that makes me nervous is using USB3 over a splitter or hub device—if that is needed (shortage of ports), better to add drives via Thunderbolt. Ditto for those narrow (thin) USB connectors.
Bottom line: the most robust and more complex systems should useThunderbolt, but USB3 with appropriate connectors remains an excellent choice for many uses (e.g. backup and single drive setups, travel, etc).
- B&H Photo has the new Mac Pro available for pre-order.
- 64GB memory for Mac Pro from OWC.
- Landing page at OWC for Thunderbolt products.
- My favorite USB3 external for main storage or backup.
- Enclosure for two hard drives.
- Wide gamut display.
- Dual-slot Helios PCIe expansion chassis.
- Simple eSATA to USB3 adapter.
USB3 is plenty fast for hard drives.
What I ordered
Ordered at 4:00 AM Pacific Time as shown below.
Apple could be a little more helpful than a 31 day shipping window, but I’m glad I groggily ordered early in the morning because shipping estimates quickly moved to February by later in the morning.
The 6-core CPU looks like the best choice for Photoshop and Lightroom. I rejected the 8-core for that purpose, because if 7 or 8 cores are actually used, they run at 3.0 GHz. But taking GHz into account, 6 cores on the 6-core at 3.5 GHz is equivalent to 7 cores at 3.0 GHz. Not much of a win for 8 cores by the math.
Moreover, the chances of all 8 cores being fully utilized for any meaningful duration are about nil, and then there is software overhead and efficiency, which increases with the number of cores and thus exerts a “core management tax”. It all depends.
The 8-core CPU has double the cache memory, and that could be interesting when 2/3/4/5/6 of eight cores are used. But it’s impossible to say other than for a specific task (and could vary task by task). Given the big upcharge, the 8-core holds out slim hope for actual benefits for what I do. What I really want is the 6-core CPU at 3.7 GHz, but Apple declines to offer that faster CPU choice.
I’ll be ordering 64GB from OWC.
SSD / flash
I was a bit frustrated to see shipping status change from “Dec 30” to “January” when I chose the 512GB or 1TB SSD options.
But a 256GB SSD is too cramped to use as both a boot drive and for my other purposes, so I gritted my teeth and ordered the 1TB option. After all, there is only one slot for an SSD, it’s the only internal storage, and there is no other option, at least for a while.
Based on past testing, I felt there was no value in paying for the D700 GPUs with 6GB memory each. If the dual D500 GPUs don’t provide a massive speedup over the prior Mac Pro 5870 video card, then something is seriously wrong. Or they are just not used, which is usually the case. And they are OpenCL, which is not necessarily supported by much software.
But the 6-core includes the D500 GPUs as standard (over the 4 core), and so there they are. One way to save money is to start with the 4-core CPU, upgrade it to the 6-core, and this then saves about $400 on the GPUs by sticking to the D300 models. But at 4:00 AM I forgot about this, and besides, the fractional cost when all other costs are considered drops well below 10% of the system price. And there is no way to know for sure if the GPUs have any benefit short of testing identical configs on specific tasks. None of the GPU options are slow.
Keyboard and mouse
I ordered a wired keyboard and standard mouse.
Wireless has always given me trouble sooner or later: “grain” feel, batteries die at inopportune times, problems at boot under some circumstances, distance from the CPU make for flaky behavior, dropouts of fractional seconds cause erratic movement. Or it can work well. But wired always works.
Don’t overspend (or worse), underspend:
- Custom-tailored advice for your own workflow and needs.
- Advice on transitioning from prior Mac to new Mac Pro.
- Backup strategy.
- Which new model to get without overspending for your particular usage.
An excerpt from my Mac Pro 2013 Choosing the CPU / GPU page:
Not sure? Consult with Lloyd Chambers.
Here’s a specification that caught my eye, just in case I feel like computing in Nepal:
Maximum altitude: 16,400 feet (5000 meters)
And 12 dBA at idle acoustics is simply amazing for a workstation-grade computer.
Choosing the right CPU for your needs
See the discussion.
|CPU Cores||Clock Speed||Cache Memory||Mainstream Task Speed*||Core-Friendly Speed**||Comments|
|Higher numbers are faster|
+37% / +5.7%
||Fastest for general use due to highest clock speed, but certain operations in programs like Adobe Photoshop Lightroom will be slower than with 6-core or 8-core options—and others will be faster! It all depends.|
~ 5.6 ±
+30% / + 16%
|14.0||18.9||About 5% slower than the 4-core in clock speed, but the two extra CPU cores are WELL worth it for programs like Adobe Photoshop Lightroom. Best all-around solution.|
|12.0||19.2||With a 14% drop in clock speed, the 8-core model is not likely to outperform the 6-core model for most tasks, but it has more cache memory and this might mitigate the clock speed losses. And it’s a good middle ground for workflows which mix video with other tasks.
Still, Photoshop hardly ever uses even 4 cores for common tasks. The 8-core is really for video processing or other specialty tasks which can use all the cores.
|10.8||22.7||Appropriate only for video users, unless big changes accrue, Photoshop and Lightroom and all productivity software will run slowest on this machine.|
* Ordinary Task Speed = expected speed with mainstream tasks which typically use four CPU cores or fewer and rarely more except for brief spikes.
** Core-Friendly = Estimated real-world best-case performance taking into account clock speed and CPU cores, application multi-threading efficiency, memory contention.
± Taking clock speed into account, the equivalent number of 3.7 GHz CPU cores (multiplier of # of cores times the clock speed). This does not take the inevitable multi-core overhead into account (hardware and software factor), which degrades performance as the number of CPU cores increases.
The 2013 Mac Pro goes on sale tomorrow, December 19. I’ll be ordering one, and with a little luck, will also have a 2nd model loaned from B&H Photo. But when Apple takes orders and when Apple ships to the first lucky few remains to be seen. It also could pose a conundrum: if the features I want are BTO (build to order) and if that delays shipping BTO are a week or two or longer, then it’s “buy and return” for me, so that later I can buy what I want. Because I need it immediately for review.
Maybe it should just be renamed “2014 Mac Pro” as more appropriate to timing.
“Sweet spot” for photographers: the 6-core model. Stay tuned for discussion later today for what exactly I’ll be getting and why.
See my Thunderbolt area, to be greatly expanded in the coming months.
Loads of Thunderbolt products are coming out, but here is a nice landing page at OWC for Thunderbolt products.
I’ll be putting my Accelsior PCIe SSDs into a Helios enclosure (the new dual-slot Helios PCIe chassis) and moving various hard drives into Thunderbolt cases (in time) but mainly into my favorite USB3 hard drive enclosure. For those without eSATA enclosures, a simple eSATA to USB3 adapter might do the trick.
USB3 is plenty fast to house two hard drives.
You’ll want memory. Apple’s memory is always expensive to go beyond the lower amounts. I plan to put 64GB of OWC memory in my 2013 Mac Pro. I want 128GB, but that is not yet feasible with current tech at 1866 MHz.
My kids go bonkers when they see me drop my iPhone on to the hard tile floor—it makes a good demo! I have no fear there, because the NewerTech NuGard KX ballistic case protects the phone (or iPod or iPad). And it makes a much better grip and adds little weight or bulk.
This is a no brainer. OWC memory is a fraction of the price that Apple charges. Go to 16GB (the max) in laptops and MacMini, and 32GB in the iMac, and 32GB or more in the Mac Pro.
Backup or lose it
Backup isn’t sexy, but it decidely bites to lose all your Stuff.
For the Photoshop and Lightroom user
The OWC Mercury Accelsior E2 is an ultra-fast PCIe SSD with dual eSATA ports. Install it into a Mac Pro (tower style) directly, or connect to any Thunderbolt-equipped Mac via the Helios enclosure. It makes Photoshop and Lightroom rock.
I’ve been using a PowerRocks MagicStick 2800 mAh with my iPhone. It keeps it nicely topped up while on the go. The supplied cable is used to charge the battery; you supply your own phone cable (e.g., Lightning cable for recent iPhone models).
Get 20% off PowerRocks using code .
This extended battery is small enough to easily fit in a purse, briefcase, backpack...even in the palm of your hand. This lightweight but powerful battery is perfect for work, vacation and everyday on-the-go. It has a modern slim, sleek, and stylish design. Its compact design makes it convenient to carry in your pocket or purse. You could even leave it plugged in to your phone. Micro USB cable and velvet bag are included.