[QUOTE=airsquirrels;443858]Caught it. Glad to see we're finally factoring using a recent CUDA version[/QUOTE]

I wonder what the 1080 will do on water?

[QUOTE=Mark Rose;443865]I wonder what the 1080 will do on water?[/QUOTE]

Apparently, even on air the board can be over clocked to around 2Ghz ( and often above ) without resorting to much more than using the MSI or EVGA utilities that set the clocks and target temps. Of course, these are the high-end boards that have better cooling ( dual fans, etc. ) that are designated FTW, SC, etc. And most users report that they are not too loud.
The most telling comment I read was one user who said he felt the card jumped 2 generations rather than just one.

Lovely news, Oliver!

Now, would someone kindly compile a x64 Windows binary? :smile:

[url=https://www.dropbox.com/s/fy8o5d6jnmsgls6/cudalib.7z?dl=1]Here are the new DLLs[/url].

Hi Karl,

I've already asked Jerry if he can built Win/CUDA8 binaries.

Oliver

[QUOTE=TheJudger;444004]Hi Karl,

I've already asked Jerry if he can built Win/CUDA8 binaries.

Oliver[/QUOTE]

All updated :smile:

[QUOTE=flashjh;444025]All updated :smile:[/QUOTE]
Thanks, Jerry!

Cheers!

Here are some freshly submitted results on a GTX 1080 @ 2ish GHz:
[CODE]
got assignment: exp=137990929 bit_min=70 bit_max=72 (5.20 GHz-days)
Starting trial factoring M137990929 from 2^70 to 2^71 (1.73 GHz-days)
k_min = 4277787055080
k_max = 8555574118335
Using GPU kernel "barrett76_mul32_gs"
Date Time | class Pct | time ETA | GHz-d/day Sieve Wait
Oct 02 10:07 | 4619 100.0% | 0.137 n.a. | 1138.42 82485 n.a.%
no factor for M137990929 from 2^70 to 2^71 [mfaktc 0.21 barrett76_mul32_gs]
tf(): total time spent: 2m 12.097s

Starting trial factoring M137990929 from 2^71 to 2^72 (3.47 GHz-days)
k_min = 8555574114780
k_max = 17111148236670
Using GPU kernel "barrett76_mul32_gs"
Date Time | class Pct | time ETA | GHz-d/day Sieve Wait
Oct 02 10:11 | 4619 100.0% | 0.274 0m00s | 1138.42 82485 n.a.%
no factor for M137990929 from 2^71 to 2^72 [mfaktc 0.21 barrett76_mul32_gs]
tf(): total time spent: 4m 24.240s
[/CODE]

Judging by the GPU utilisation (~91%), this is not the limit.

That's [I]almost[/I] useful to me -- if you could tell me the actual core clock speed under load while doing mfaktc then it I could use it for my benchmark data.

Now I know there's a [url=http://www.mersenne.ca/mfaktc.php#]specialised form[/url] for this sort of data submission.
I've sent the standardised data, with comment pointing to your request, James.

My GPUSieveSize parameter was set to '128', though, which improved utilisation to 96% and resulted in ~1190 GHz-d/day value.
A quick look at other parameters didn't shed any useful insight.

Time to feed Pascal
 

Hi,

[B]Thank you, Jerry![/B]

I've uploaded the binaries [URL="http://mersenneforum.org/mfaktc/mfaktc-0.21/mfaktc-0.21.win.cuda80.zip"]here[/URL], too.

Upgrade instructions from mfaktc 0.21 (CUDA 6.5) to mfaktc 0.21 (CUDA 8.0):[LIST=1][*]stop mfaktc[*]replace executables and DLLs[*]start mfaktc again[/LIST]You can replace/restart in the middle of a run easily.

Oliver

P.S. I guess the "extra versions" (Wagstaff mode and LessClasses) will be available soon.
[B]P.P.S. How high is the demand for Wagstaff and/or LessClasses?[/B]

Found my first bug for mfaktc by getting overnight TF work.
It would seem that 128 lines in worktodo file cause some kind of overflow, and mfaktc complains about invalid format in such a case (and doesn't start).
Will report further findings later.