mfaktc: a CUDA program for Mersenne prefactoring

[Xyzzy]

Of course, you can also directly upload the entire results.txt file, instead of cut/paste into the browser 4k-limit window. The direct file upload can handle a far larger chunk of information (though TheJudger did generate even a results.txt file that caused problems not too long ago).

We had a 78K file fail to complete a few days ago.

[nucleon]

Originally Posted by Christenson

No problem. I'm not proposing that manual assignments be dropped or changed; there are too many advantages to having them available, including it being an important simplification when new methods are being developed.

I'm more into the idea that we have a standard, automatic web interface for internet-connected machines, and not everyone writes bash scripts the way you can, or even has bash available! (Yes, I can run mingwin, but that's beyond the willingness of many). Although, getting my card up under xubuntu wasn't what I'd call trivial, but this is not the place to scratch that particular itch.

Agreed - in time mfaktc needs to get automated submit/get work. Or better yet - integrated into prime95. Prime95 handles the submit/get work, and has a generic option to run gpu code as separate process.

But Oliver is but one person. I think we can handle the inconvenience of manual submit/get work for the moment while Oliver works on his priorities. Getting version 0.17 out and getting the code optimal/valid.

Another priority is I'd like to see someone take up the challenge to get CUDA P-1 code out :) I think that's way more important than automated submissions.

-- Craig

[Xyzzy]

Random, late night questions:

Our experience on our setups (i5/570) is that the GPU is the limiter. The i7/580 is an unknown for us, but it is $250 more.

1 - How do AMD processors compare, clock to clock and/or price to performance, for this particular task?

2 - Given most of us run 2 instances, would an i3 be enough? One of them has the clockspeed and HT. Does HT make a difference? Does reduced L3 cache make a difference?

3 - Can you run 2 GPU cards and use them both? Is it SLI or separate?

4 - How much more productive would a 580 be over a 570, noting it is ~$150 more?

Thanks!

[nucleon]

Originally Posted by Xyzzy

Random, late night questions:

Our experience on our setups (i5/570) is that the GPU is the limiter. The i7/580 is an unknown for us, but it is $250 more.

Got any stats on GHZ-days/day on your i5/570 combo?

-- Craig

[Ungelovende]

Originally Posted by Xyzzy

Random, late night questions:

I'm running 2X GTX570 - 2+2 instances on a Sandy 2600k getting 4X 195 M/s. Win7 64bits, no SLI. I have also tested them on a i7 920/1366, but it was not fast enough.

[Xyzzy]

Got any stats on GHZ-days/day on your i5/570 combo?

We think ~540 a day, for two boxes. (We hope that is a sane number.)

We have posted (probably too much) about our setup over the last few pages.

We are tempted to build a very "neato" box to use for real work, like games and crap, since all of our other boxes are dedicated. We are not sure what kind of "neato" box we can build but it is fun trying to figure out! If we can run 2 GPUs then that is a good excuse!

We could build 2 more of what we have now but that would be boring.

Our 4 non-GPU quads will be retired within the next month.

[Xyzzy]

I'm running 2X GTX570 - 2+2 instances on a Sandy 2600k getting 4X 195 M/s.

Is it difficult (heat, physical space, power, 16x slots) to get them in the case?

Win7 64bits, no SLI.

Is running them SLI an option?

I have also tested them on a i7 920/1366, but it was not fast enough.

The CPU or the GPU?

Thanks!

[Christenson]

Originally Posted by Xyzzy

We think ~540 a day, for two boxes. (We hope that is a sane number.)

We have posted (probably too much) about our setup over the last few pages.

My GTX440 will do about 90 GHz days/day if I have my numbers correct, so 540 GHz days/day isn't insane. Keep talking, you have to keep up those Aleph-0 posts snake1 gave you!

If you are going to do a super hot system, AMD bulldozer is supposed to be out soon, and it should be quite a multicore CPU.

[Uncwilly]

Originally Posted by Xyzzy

Our 4 non-GPU quads will be retired within the next month.

Are they going to be wished away into the cornfield? Would you be willing to allow them to be adopted and not euthanized?

[Xyzzy]

Are they going to be wished away into the cornfield? Would you be willing to allow them to be adopted and not euthanized?

We have a plan to have them "adopted". We have not finalized the details yet.

Snake1: We are sure the adoption process involves the forum.
Fish1: V yvxr gb cvpgher Wrfhf nf n svther fxngre. Ur jrnef yvxr n juvgr bhgsvg, naq Ur qbrf vagrecergvir vpr qnaprf bs zl yvsr'f wbhearl.

[mdettweiler]

Hi guys,

Recently, I've been playing around a bit with mfaktc on a GTX 460 (768MB) running on a system with a stock Q6600 CPU. I'm currently taking M332228447 from 75 to 81 bits (79>80 in progress right now). With the default SievePrimes=25000, I'm getting speeds of about 109M/s. The CPU appears to be the bottleneck, with mfaktc using 100% of one core and 82% of the GPU. Does this sound like what I should be optimally getting on this GPU/CPU/exponent combination, or might adjusting SievePrimes help reduce the bottleneck?

Also: I've noticed that whenever I have to stop and restart mfaktc, when it first comes back it starts at only ~75M/s, and takes about 1-2 hours to work its way back up to the usual ~109M/s. Is this normal behavior? Does anyone know why it does this?

Thanks,
Max