Tests that follow compare the speed using MemoryTester 'compute' on the 2.5 GHz MacBook Pro, using either the 4GB Apple memory, or the 16GB OWC memory.
Update: the diglloydSpeed1 benchmark runs in 22 seconds with 16GB, instead of 26 seconds with 4GB memory. That’s probably because 4GB is just a bit on the low side (not enough memory to run efficiently).
The test shows that the Apple 4GB (2 X 2GB) memory configuration has lower speed than the OWC 16GB kit, by about 9%. This is unlikely to have any real world effect as indeed the SHA1 test shows no difference (see graph further below). However, I have not yet run my other suite of tests, such as Photoshop speed.
I also tested my 13" MacBook Pro with the 8GB OWC kit (2 X 4GB), and memory speed is also testing faster on the 13" model. I have no explanation for the slower memory speed of the Apple 4GB configuration in the 17" MacBook Pro; it must be some type of module design issue.
Compute intensive benchmarks
The SHA1 hash has moderate memory access. The memory configuration exerts no effect, presumably processor caching is hiding the small memory speed difference.
Pure CPU speed test which does not involve memory— as expected there is no difference in speed between memory configurations.