[QUOTE=henryzz;252462]Is raising the voltage permanently not an option?[/QUOTE]

I guess I could raise it above some more points, and let it sag during operation.

-- Craig

[QUOTE=sdbardwick;252446]Here's one (oops it's actually a 1090T OC'd to 3400 [17x200] instead of 3200[16x200]).[/QUOTE]Thanks, benchmark added. Now you can see a direct comparison between a (single core of an) [url=http://mersenne-aries.sili.net/throughput.php?scale=1.2795650924844&cpu1=AMD%20Phenom%28tm%29%20II%20X6%201090T%20Processor|512|6144&mhz1=3500&cpu2=Intel%28R%29%20Core%28TM%29%20i7-2600K%20CPU%20%40%203.40GHz|256|8192&mhz2=3500]X6-1090T and i7-2600K[/url], at identical clockspeeds the AMD only does 70% of the FFT work of the Intel. Of course, the AMD does have more cores available. But the Intel can almost certainly run faster. But the AMD is cheaper. And so it goes :smile:

@KingKurly: your benchmark(s) would also be welcome, you could post here, or in the [url=http://www.mersenneforum.org/showthread.php?t=59]Perpetual Benchmark Thread[/url] or just [url=http://mersenne-aries.sili.net/throughput.php]email to me[/url].

I think I made a mistake when running that benchmark. There must have been a background process running, as it is notably slower than when it was clocked at 3.2gHz. If I get access later I'll re-run it.
This is a partial result from before @ 3.2:
[CODE]Prime95 64-bit version 26.3, RdtscTiming=1
Best time for 768K FFT length: 9.842 ms., avg: 9.963 ms.
Best time for 896K FFT length: 12.088 ms., avg: 12.261 ms.
Best time for 1024K FFT length: 13.349 ms., avg: 13.444 ms.
Best time for 1280K FFT length: 17.199 ms., avg: 17.278 ms.
Best time for 1536K FFT length: 21.077 ms., avg: 21.171 ms.
Best time for 1792K FFT length: 25.878 ms., avg: 25.984 ms.
Best time for 2048K FFT length: 28.287 ms., avg: 28.429 ms.
Best time for 2560K FFT length: 36.407 ms., avg: 36.546 ms.
Best time for 3072K FFT length: 45.043 ms., avg: 45.168 ms.
Best time for 3584K FFT length: 54.607 ms., avg: 54.738 ms.
Best time for 4096K FFT length: 60.257 ms., avg: 60.452 ms.
Best time for 5120K FFT length: 78.992 ms., avg: 79.270 ms.
Best time for 6144K FFT length: 99.608 ms., avg: 99.875 ms.
Best time for 7168K FFT length: 123.402 ms., avg: 123.595 ms.
Best time for 8192K FFT length: 136.919 ms., avg: 137.173 ms.[/CODE]Sorry!

[QUOTE=sdbardwick;252490]I think I made a mistake when running that benchmark.[/QUOTE]I've purged your old benchmark data and uploaded the new one. Let me know if you get a proper full benchmark since you haven't supplied updated TF timing.

[QUOTE=James Heinrich;252494]I've purged your old benchmark data and uploaded the new one. Let me know if you get a proper full benchmark since you haven't supplied updated TF timing.[/QUOTE]
RDP to the rescue!
3.2 TF Results:
[CODE]Best time for 58 bit trial factors: 1.954 ms.
Best time for 59 bit trial factors: 2.000 ms.
Best time for 60 bit trial factors: 1.998 ms.
Best time for 61 bit trial factors: 2.175 ms.
Best time for 62 bit trial factors: 2.215 ms.
Best time for 63 bit trial factors: 2.636 ms.
Best time for 64 bit trial factors: 2.958 ms.
Best time for 65 bit trial factors: 3.672 ms.
Best time for 66 bit trial factors: 4.389 ms.
Best time for 67 bit trial factors: 4.355 ms.[/CODE]

Re-purged, re-updated :smile:

Is there a final or recommended inventory list for the SB system?

Just an update from me, my 2133 ram now has unrecoverable read errors.

I did some more reading, and sandy bridge cpus are only rated for 1.5V ram. It can do 1.65V ram, but your technically over-volting the memory parts of the cpu.

This is what i'm running with atm:

[url]http://www.gskill.com/products.php?index=349[/url]

1600MHz @1.5V I haven't found anything faster that is still at 1.5V.

I do notice the difference. In my pseudo-scientific testing I've found that faster ram does yield better performance for prime hunting (P-1 saw the biggest improvement). Sorry I don't have proper figures, but it takes so damn long to do tests properly :)

Also I found too the biggest improvement in stability came from increasing the "QPI/Vtt voltage" from 1.05 to 1.16V.


-- Craig