As an Amazon Associate I earn from qualifying purchases @AMAZON
OWC 128GB Memory Kit: Photoshop CC diglloyd Benchmarks
There are really two points of interest in considering whether installing 128GB is preferential to 64GB:
- Does the slower memory speed of the 128GB memory kit significantly degrade performance for those tasks not involving large memory requirements?
- Does the 128GB option substantially improve performance for really big jobs?
These questions are addressed here with Photoshop performance metrics run on the 3.5 GHz 6-core Mac Pro with D500 GPUs.
diglloyd Photoshop benchmarks
Note that specific Photoshop minor version and system version numbers produce different test results in early 2014 due to short term changes in the code forced by graphics driver bugs. Hence these numbers cannot be compared to other results with slightly different version of Photoshop or OS X.
The diglloydMedium and diglloydSpeed1 benchmarks easily fit into a 64GB memory configuration.
Since the 64GB config has greater memory bandwidth than the 128GB config, it is to be expected that the 64GB config will be faster. But the actual speed difference is quite small, amounting to 3.4% - 3.9%. Which shows that the CPU on-chip caches are highly effective.
Commentary on diglloydHuge benchmark follows below.
The diglloydHuge benchmark eats up about 55GB of memory just for Photoshop, which is just a bit too much for a 64GB configuration (OS X and the virtual memory system and caching and other things eat up memory also).
The 128GB configuration shows a significant gain in performance of 31%. That advantage would grow were the job to become larger.
The disadvantage of the 64GB memory configuration is obscured by the phenomenal performance of the Mac Pro internal SSD, which can write at around 930 MB/sec and read at around 1200 MB/sec (e.g. the Photoshop scratch file). Placing the Photoshop scratch file on a slower volume or any hard-drive based solution would severely degrade performance with the 64GB config.