View Single Post
Old 2021-08-27, 15:46   #10
chris2be8
 
chris2be8's Avatar
 
Sep 2009

91F16 Posts
Default

I got that error too:
Code:
$ echo "((2^499-1)/20959)" | ./ecm -gpu -cgbn -sigma 3:1000 20000 0
GMP-ECM 7.0.5-dev [configured with GMP 6.1.2, --enable-asm-redc, --enable-gpu, --enable-assert] [ECM]
Input number is ((2^499-1)/20959) (146 digits)
Using B1=20000, B2=0, sigma=3:1000-3:1383 (384 curves)
CUDA error occurred: cannot set while device is active in this process
While running cudaSetDeviceFlags (cudaDeviceScheduleBlockingSync)   (file cgbn_stage1.cu, line 601)
Though the different line number makes me suspect I may have an old version of your code. What command should I run to download the latest version?

After commenting out line 601 it works:
Code:
$ echo "((2^499-1)/20959)" | ./ecm -gpu -cgbn -sigma 3:1000 20000 0
GMP-ECM 7.0.5-dev [configured with GMP 6.1.2, --enable-asm-redc, --enable-gpu, --enable-assert] [ECM]
Input number is ((2^499-1)/20959) (146 digits)
Using B1=20000, B2=0, sigma=3:1000-3:1383 (384 curves)
Running GPU kernel<24,128> ...
Copying results back to CPU ...
Computing 384 Step 1 took 1653ms of CPU time / 2928ms of GPU time
But after raising B1 to 200000 it fails:
Code:
$ echo "((2^499-1)/20959)" | ./ecm -gpu -cgbn -sigma 3:1000 200000 0
GMP-ECM 7.0.5-dev [configured with GMP 6.1.2, --enable-asm-redc, --enable-gpu, --enable-assert] [ECM]
Input number is ((2^499-1)/20959) (146 digits)
Using B1=200000, B2=0, sigma=3:1000-3:1383 (384 curves)
Running GPU kernel<24,128> ...
CUDA error occurred: the launch timed out and was terminated
While running cudaDeviceSynchronize()   (file cgbn_stage1.cu, line 632)
@bsquared, what happens if you try raising B1 by another factor of 10? How high a B1 will it take?

This GPU has CUDA arch 3.0, is that new enough for CGBN? I also have a newer GPU with CUDA arch 5.2, I'll try installing on that next.

NB. msieve says what CUDA arch the card is when used for GPU based poly selection. I'm using that to check what they really support.

Chris

PS. Should discussion of ecm with CGBN be split off into another thread?
chris2be8 is offline   Reply With Quote