[QUOTE=nucleon;332199]After running it for a few days, I'm not overly happy with the product. I've switched it to TF to see how it went. Even mfaktc crashes every so often.[/QUOTE]

When I bought my EVGA 2GB 560 (factory overclocked) card, I had 24 hours to return it to the store if it was not "deterministic" (they had no idea what I was talking about, but agreed I could return it if the memory was bad...).

One of the tests I put it through was the [URL="http://wili.cc/blog/gpu-burn.html"]GPU-burn[/URL]. I can't vouch for this test beyond it was one of many I did against the card and the card passed.

I'm happy to say my 560 passed every test I applied (including image SIFTing and matching).

It might be worth our while to develop a test suite for GPUs....

[QUOTE=ixfd64;332188]FYI: a full list of opportunities to win a Titan is available here: [URL]http://www.overclock.net/t/1365601/facebook-giveaways-6-gtx-titans-asus-ares-ii-gtx-670-gtx-660-ti-7970-fx-8350-facebook-required[/URL][/QUOTE]

Must enter on Facebook.....:rant:

running now 57885161 with factory clocks and dp activated


best results with mfactc and oc was 520ghz-d/d

[QUOTE=chalsall;332202]When I bought my EVGA 2GB 560 (factory overclocked) card, I had 24 hours to return it to the store if it was not "deterministic" (they had no idea what I was talking about, but agreed I could return it if the memory was bad...).

One of the tests I put it through was the [URL="http://wili.cc/blog/gpu-burn.html"]GPU-burn[/URL]. I can't vouch for this test beyond it was one of many I did against the card and the card passed.

I'm happy to say my 560 passed every test I applied (including image SIFTing and matching).

It might be worth our while to develop a test suite for GPUs....[/QUOTE]

There are many nice gpu-stress programs, mainly furmark, folding@h [URL="http://folding.stanford.edu/English/DownloadUtils"]utils[/URL].

Hmm, by the way, is compute stress and "gaming" stress different? What I mean is, is it possible to be rock solid on "game" type tests but fail on compute?

Our GPU draws more watts playing a game than it does totally maxxed out trial factoring.

:confused:

[QUOTE=kracker;332234]Hmm, by the way, is compute stress and "gaming" stress different?[/QUOTE]
"Compute stress", for example, does not use ROPs of the GPU.
Furmark was good for stress testing until AMD and Nvidia implemented protection from it either on driver or hardware level.
The last gpu from Nvidia to have no Furmark protection was gtx 480.
OCCT should be limited too, according how [URL="http://www.techpowerup.com/134460/Disable-GeForce-GTX-580-Power-Throttling-using-GPU-Z.html?cp=2"]GTX 580 was limited in the past[/URL].
Alternatively, Heaven benchmark from Unigine is a good stability tester.

GTX Titan, I am disappoint ([I]sic[/I]). Considering its FP64 throughput, I was expecting [I]at least[/I] 1,000-1,500 GHz-days/day for mfaktc. I know GFLOPS isn't everything, but still...

As has been stated before, mfaktc in pretty much all integer code. FP64 throughput won't help an ounce. The CuLu performance, on the other hand, is decent, but perhaps not what we were hoping it to be.

I've heard AMD cards are better optimized for integer performance. Does this mean they will generally outperform Nvidia cards (like in the case of Bitcoin mining) once mfakto supports GPU sieving?

Doing further testing...

Furmark - no issues
MemtestG80 - no issues
occt - no issues
oclhashcat - doesn't run
gpuburn - linux only (don't have linux GPU-capable machine)

mfaktc - fails after extended times
cudalucas - fails at random points (even one test the app went thru all FFT sizes and failed on all of them)

mboard bios - latest non beta version.
Ram at reasonable timings (1600).

I still have it in a machine with both GTX580 and Titan. I could remove the GTX580, and see how it goes.

I think I might run mfaktc and restart when required until further notice. I really wanted to do some DC-LL testing. Oh well, maybe driver release might fix things.

-- Craig

[QUOTE=ixfd64;332307]I've heard AMD cards are better optimized for integer performance. Does this mean they will generally outperform Nvidia cards (like in the case of Bitcoin mining) once mfakto supports GPU sieving?[/QUOTE]

Where did you hear that?
I'm doing my best, but I would not expect the result of GPU sieving be much different than with mfaktc: the additional effort of the sieving is barely offset by the higher (read: better) SievePrimes.
AMD's integer performance is really great with shifts, logical, add and sub operations. Multiplication is only good for 24bit numbers.