[QUOTE=Honza;253810]Once there is Win build, I'll try GTX 580.
Any change of CPU usage with those huge numbers?[/QUOTE]
CPU usage is around 90% (of one core) here...

With llrCUDA 0.60:
[code]gary@herford:~/Desktop/gpu-stuff/llrcuda$ ./llrCUDA -d -q237019*2^6100630+1
Starting Proth prime test of 237019*2^6100630+1
Using complex irrational base DWT, FFT length = 2097152, a = 3
237019*2^6100630+1 is not prime. Proth RES64: 381782D8C112D665 Time : 64405.166 sec.
[/code]
And on a CPU:
[code][2010-12-25 21:38:49 WEST] Candidate: 237019*2^6100630+1 Program: llr.exe Residue: 381782D8C112D665 Time: 93141 seconds[/code]
Just two more of ltd's PSP tests left to do...once those are all completed with correct residues I think we can be pretty confident that llrCUDA 0.60 is producing consistently good results. :smile: (Then we can use it for production work! :big grin:)

[QUOTE=Ralf Recker;253812]CPU usage is around 90% (of one core) here...[/QUOTE]
I think It is Linux driver problem.

And it's successful.
19249*2^13018586+1 is prime! Time : 67256.132 sec.
I too am seeing about 90% of a core being used in 64-bit Linux.

[QUOTE=frmky;253881]And it's successful.
19249*2^13018586+1 is prime! Time : 67256.132 sec.
I too am seeing about 90% of a core being used in 64-bit Linux.[/QUOTE]
FYI, I'm getting about 60% of a core being used under 64-bit Ubuntu 10.04.1 (CUDA 3.2). Note however that the number I am testing (237019*2^6101242+1) is rather smaller than yours; might this make a difference in the CPU usage?

Another good result with 0.60:
[code]gary@herford:~/Desktop/gpu-stuff/llrcuda$ time ./llrCUDA -d -q237019*2^6101242+1
Starting Proth prime test of 237019*2^6101242+1
Using complex irrational base DWT, FFT length = 2097152, a = 3
237019*2^6101242+1 is not prime. Proth RES64: 1A0C5A3E8372AFC2 Time : 64425.006 sec.

real 1074m13.369s
user 233m10.970s
sys 467m38.890s[/code]
[code]
[2010-12-27 00:15:56 WEST] Candidate: 237019*2^6101242+1 Program: llr.exe Residue: 1A0C5A3E8372AFC2 Time: 95784 seconds
[/code]

28433*2^7830457+1 is prime! Time : 83787.039 sec.
33661*2^7031232+1 is prime! Time : 76324.166 sec.
44131*2^995972+1 is prime! Time : 1467.252 sec.
46157*2^698207+1 is prime! Time : 618.190 sec.
54767*2^1337287+1 is prime! Time : 1892.311 sec.
65567*2^1013803+1 is prime! Time : 1576.657 sec.
69109*2^1157446+1 is prime! Time : 1767.502 sec.

[i]El numero cuatro[i]:
[code]
gary@herford:~/Desktop/gpu-stuff/llrcuda$ ./llrCUDA -d -q237019*2^6102286+1
Resuming Proth prime test of 237019*2^6102286+1 at bit 684033 [11.20%]
237019*2^6102286+1 is not prime. Proth RES64: 4E2920067DCBCF57 Time : 64448.145 sec.
[/code]
[code]
[2010-12-28 08:59:07 WEST] Candidate: 237019*2^6102286+1 Program: llr.exe Residue: 4E2920067DCBCF57 Time: 91167 seconds
[/code]
With four good n=6M+ residues down, it looks like 0.60 has proven itself sufficiently well to try it out on some production work. I expect this will do decently well on NPLB's megabit mini-drive. :smile: (It's testing numbers in the vicinity of n=1.26M, so the speed boost may not be quite as significant as on the larger numbers I've been recently testing, but it will probably still do better than a CPU.)

3*2^6090515-1 is prime! Time : 15356.340 sec.
3*2^5082306+1 is prime! Time : 12953.410 sec.
3*2^4235414-1 is prime! Time : 10716.365 sec.
3*2^2291610+1 is prime! Time : 3112.366 sec.
59*2^3408416-1 is prime! Time : 8568.866 sec.
5*2^3059698-1 is prime! Time : 4117.422 sec.
7*2^2915954+1 is prime! Time : 3923.360 sec.
17*2^2721830-1 is prime! Time : 3690.609 sec.
5*2^2460482-1 is prime! Time : 3316.242 sec.

3*2^7033641+1 is prime! Time : 36380.626 sec.

Tried the new beta windows driver (267.24) but this made thing even worse.
Now I get a blue screen when I run llrcuda.