Hi ,TheJudger

Thank you information.

[QUOTE=TheJudger;232032]does this CUDA FFT implementation fit into your code?

NukadaFFT
[url]http://matsu-www.is.titech.ac.jp/~nukada/nufft/[/url]
[/QUOTE]

[QUOTE]
NukadaFFT-1.0/sample/double$ ./a.out 1024b1
Batched 1-D FFT: 1024, batch = 1
Total: 0.101248 msec, 1.011378 GFLOPS.
On-board: 0.040512 msec, 1.263823 GFLOPS.
Max error = 4.024558e-16.

NukadaFFT-1.0/sample/double$ ./a.out 2048b1
Failed to create plan for 1-D FFT.
[/QUOTE]

This version only support < 1024,I guess.

Rolling along:
[URL="http://www.mersenne.org/report_exponent/?exp_lo=35000683&exp_hi=35000683&B1=Get+status"]M35000683[/URL]

Just 7 more to go and the 35M-35.001M range will be completely doublechecked! :smile:

One more down:
[url=http://www.mersenne.org/report_exponent/?exp_lo=35000741&exp_hi=35000741&B1=Get+status]M35000741[/url]

BTW, I ran into a little trouble reporting this result to PrimeNet--see [url=http://www.mersenneforum.org/showthread.php?t=14016]this thread[/url] for more information. It does bring up an interesting point: perhaps the MacLucasFFTW-CUDA software should be renamed to something else (considering that it doesn't really have anything to do with Macs now) so that PrimeNet doesn't see GPU doublechecks as duplicates when the first-pass test was done by MacLucasFFTW-CPU.

Hi ,mdettweiler
Change the output line.:smile:
[QUOTE]
cudalucas.1.0$ ./CUDALucas 216091
Iteration 10000 M( 216091 )C, 0x30247786758b8792, n = 524288, CUDALucas v1.0
...
Iteration 210000 M( 216091 )C, 0xcfe091c8f59f8a7b, n = 524288, CUDALucas v1.0
M( 216091 )P, n = 524288, CUDALucas v1.0
[/QUOTE]

Thank you ,mdetteiler,rouge,Prime95

[QUOTE=msft;232550]Hi ,mdettweiler
Change the output line.:smile:


Thank you ,mdetteiler,rouge,Prime95[/QUOTE]
Thanks! :smile:

George, are the CUDALucas results close enough to the MacLucasFFTW format to be accepted by PrimeNet as is, or will that need to be added as a distinct category?

[QUOTE=mdettweiler;232551]George, are the CUDALucas results close enough to the MacLucasFFTW format to be accepted by PrimeNet as is, or will that need to be added as a distinct category?[/QUOTE]

I need to modify the manual results page. Should have it ready in a few hours.

Meanwhile, I just tested M74000053 with MacLucasFFTW. :smile:

I'm working on double-check of M67239937 with a precompiled version of MacLucas, hope it will be enough for double-checking.

Next one to try: M123999973...

Luigi

Thank you ET_ ,:smile:

Hi ,
[URL="http://www.mersenne.org/report_exponent/?exp_lo=100000039&exp_hi=&B1=Get+status"]M100000039[/URL].

[QUOTE=msft;236326]Hi ,
[URL="http://www.mersenne.org/report_exponent/?exp_lo=100000039&exp_hi=&B1=Get+status"]M100000039[/URL].[/QUOTE]
Hmm...looks like the residue doesn't match. Anyone up for a third check? (I can't, Gary's GPU is offline at the moment and when it comes back it will be doing NPLB sieving work.)

Hi, mdettweiler
[QUOTE=mdettweiler;236374]Hmm...looks like the residue doesn't match. Anyone up for a third check? (I can't, Gary's GPU is offline at the moment and when it comes back it will be doing NPLB sieving work.)[/QUOTE]
I think third check need Prime95.