With the R.S.T. Pro2 in place, I proceeded to hammer away at the TWINX PC4400. We tested the TWINX at seven different speeds and timings... 250MHz @ 2.5-4-3-5, 250MHz @ 2.5-4-3-6, 250MHz @ 3-4-4-8, 275MHz @ 2.5-4-3-6, 275MHz @ 3-4-4-8, 277MHz @ 3-4-4-8, and 288MHz @ 3-4-4-8. At 277MHz and 288MHz, we needed to bump the voltage from the default 2.75v to 2.8v. These were our results:
|250MHz @ 2.5-4-3-5
|250MHz @ 2.5-4-3-6
|250MHz @ 3-4-4-8
|275MHz @ 2.5-4-3-6
|275MHz @ 3-4-4-8
|277MHz @ 3-4-4-8
|288MHz @ 3-4-4-8
I should point out that we tried different speed and timing variations outside of what you see above, but the numbers we've presented were FSB speeds Windows was able to POST at. However, as you can see, R.S.T. Pro had a thing or two to say when it came down to stability.
For fun, I grabbed our Corsair PC4000 ProSeries, and bumped it up to 290FSB, which it had handled in the past. Given what we've learned today, I guess I wasn't too surprised to see that it failed the R.S.T. Pro tests, despite it being fairly stable in Windows. We also tested several other ram modules using this test method, but I'll sum it up later on.
ASUS P4C800-E: Pentium 4 2.4C, 2 x 512MB Corsair TWINX1024-4400, ATI AIW Radeon 9800 Pro, 80GB Western Digital, Windows XP SP1, ATI Catalyst 3.10.
Test Software for both platforms will be:
Unreal Tournament 2003
Unfortunately, I do not have any 512MB/PC4400 sticks of competing products. We will be presenting benchmarks at 250MHz @ 2.5-4-3-6, 275MHz @ 3-4-4-8 and 277MHz @ 3-4-4-8 on the P4C800-E, all in Dual Channel mode. I will not be displaying 200MHz numbers, which I will explain why at the end of the review. The CPU and Memory ratio will be 1:1.
SiSoft Sandra 2004 - Memory Benchmark
Although a synthetic benchmark, it's a popular one, freely available if you wish to make comparison benchmarks. We will be testing the memory speeds, using the 32-bit 2004 version.
Corsair claims a 10% improvement compared to their PC4000 XMS modules, and our Sisoft Sandra results reflect that more or less. There is an improvement of about 300 points when moving from 250MHz to 275Mhz.
A good indicator of CPU/Motherboard performance is version 4.2, by Xavier Gourdon. We used a computation of 10000000 digits of Pi, Chudnovsky method, 1024 K FFT, and no disk memory. Note that lower scores are better, and times are in seconds.
We shave about two seconds with the 25MHz jump. At 277MHz, there is still an improvement, but it's very small.
TMPGEnc MPEG Encoding
Video editing is a taxing chore, and we'll be testing the TWINX using TMPGEnc 2.512 to encode a 7.78MB, 1:30 movie trailer to a 24FPS MPG file. Note that lower scores are better.
Not much of a change here, but the MPEG-1 file is probably too small to show any major results. The move from 250MHz to 275MHz still shaves a second off the encoding time.