GPU LL result rejected

mdettweiler · 2010-10-04, 17:34

Over the last few weeks I have been running LLs on a friend's GPU and reserving/reporting them via a PrimeNet account under his name. Since MacLucasFFTW-CUDA only supports power-of-2 FFT lengths, it's most efficient in the ranges right below a power-of-2 FFT change, and thus I've been working on doublechecks starting from 35M. At this point I've already reserved and later submitted two such results without issue.

This time, however, things didn't go quite so smoothly. I reserved M35000741 (under the username gd_barnes, a.k.a. Gary Barnes), ran it on the GPU (which took a few days) then attempted to report the result via the manual results page just now. However, I got this message (I've blanked out the last two digits of the residual here, but they were all there in the error):

Code:

MacLucasFFTW lines found: 1
Processing result: M( 35000741 )C, 0x8a9ff8fb5658d4**, n = 2097152, MacLucasFFTW v8.1  Ballester
Error code: 40, error text: Another computer has already reported this LL result for M35000741
Success

Looking at the Exponent Status page, I see that this exponent had a first-time result returned way back on PrimeNet v4 (as is usual for this range); no other LL results were listed. My reservation was also not listed, so I tried reserving it again via Manual Assignments. That done, I again tried to submit the result via Manual Results and got the same error. The reservation also disappeared from Exponent Status page, probably because the server somehow saw the result as not needed.

Anybody have an idea of what's going on here?

Prime95 · 2010-10-04, 18:40

The previous result was computed using MacLucas. Primenet thinks you are re-reporting that old result.

I've temporarily modified the old result so you can report the new result.

Long term I need to change the manual reporting scripts so that it recognizes your result as being from the CUDA application. Perhaps msft can change the output line from "MacLucasFFTW" to "MacLucasCUDA".

mdettweiler · 2010-10-04, 18:45

Quote:

Originally Posted by Prime95

The previous result was computed using MacLucas. Primenet thinks you are re-reporting that old result.

I've temporarily modified the old result so you can report the new result.

Long term I need to change the manual reporting scripts so that it recognizes your result as being from the CUDA application. Perhaps msft can change the output line from "MacLucasFFTW" to "MacLucasCUDA".

Thanks, it worked now.

rogue · 2010-10-04, 18:45

Quote:

Originally Posted by Prime95

The previous result was computed using MacLucas. Primenet thinks you are re-reporting that old result.

I've temporarily modified the old result so you can report the new result.

Long term I need to change the manual reporting scripts so that it recognizes your result as being from the CUDA application. Perhaps msft can change the output line from "MacLucasFFTW" to "MacLucasCUDA".

Or maybe it should just be CUDALucas. Wasn't MacLucas originally written for Mac users who couldn't hook up to GIMPS through a client? This software is really CUDA specific rather than specific to an OS.

Prime95 · 2010-10-04, 20:03

Quote:

Originally Posted by rogue

Or maybe it should just be CUDALucas.

Sounds good to me

2010-10-04, 17:34	#1
mdettweiler A Sunny Moo Aug 2007 USA (GMT-5) 14151₈ Posts	GPU LL result rejected Over the last few weeks I have been running LLs on a friend's GPU and reserving/reporting them via a PrimeNet account under his name. Since MacLucasFFTW-CUDA only supports power-of-2 FFT lengths, it's most efficient in the ranges right below a power-of-2 FFT change, and thus I've been working on doublechecks starting from 35M. At this point I've already reserved and later submitted two such results without issue. This time, however, things didn't go quite so smoothly. I reserved M35000741 (under the username gd_barnes, a.k.a. Gary Barnes), ran it on the GPU (which took a few days) then attempted to report the result via the manual results page just now. However, I got this message (I've blanked out the last two digits of the residual here, but they were all there in the error): Code: MacLucasFFTW lines found: 1 Processing result: M( 35000741 )C, 0x8a9ff8fb5658d4**, n = 2097152, MacLucasFFTW v8.1 Ballester Error code: 40, error text: Another computer has already reported this LL result for M35000741 Success Looking at the Exponent Status page, I see that this exponent had a first-time result returned way back on PrimeNet v4 (as is usual for this range); no other LL results were listed. My reservation was also not listed, so I tried reserving it again via Manual Assignments. That done, I again tried to submit the result via Manual Results and got the same error. The reservation also disappeared from Exponent Status page, probably because the server somehow saw the result as not needed. Anybody have an idea of what's going on here?

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Why was this rejected?	petrw1	PrimeNet	2	2015-10-01 17:11
What does my result mean?	Unregistered	Information & Answers	6	2013-04-16 21:52
Odd result	1997rj7	PrimeNet	2	2009-12-04 08:48
Rejected k/n pairs	em99010pepe	No Prime Left Behind	18	2008-12-06 12:50
Tony Reix, I'm feeling rejected :)	Prime95	Lounge	2	2007-02-26 08:05

2010-10-04, 18:40	#2
Prime95 P90 years forever! Aug 2002 Yeehaw, FL 7526₁₀ Posts	The previous result was computed using MacLucas. Primenet thinks you are re-reporting that old result. I've temporarily modified the old result so you can report the new result. Long term I need to change the manual reporting scripts so that it recognizes your result as being from the CUDA application. Perhaps msft can change the output line from "MacLucasFFTW" to "MacLucasCUDA".