Help! Magma Test Failed on Multi GPU Setup

Open discussion for MAGMA library (Matrix Algebra on GPU and Multicore Architectures)

Help! Magma Test Failed on Multi GPU Setup

Postby organicchemistry_01 » Tue Jul 04, 2017 11:32 am

This is my configuration

% MAGMA 2.2.0 compiled for CUDA capability >= 2.0, 32-bit magma_int_t, 64-bit pointer.
% CUDA runtime 8000, driver 8000. OpenMP threads 32. MKL 2017.0.0, MKL threads 16.
% device 0: GeForce GTX 1060 6GB, 1784.5 MHz clock, 6070.8 MiB memory, capability 6.1
% device 1: Tesla M2050, 1147.0 MHz clock, 2622.3 MiB memory, capability 2.0

These are sample of failed tests
% side = Left, uplo = Lower, transA = No transpose, diag = Non-unit, ngpu = 2
% M N MAGMA Gflop/s (ms) CUBLAS Gflop/s (ms) CPU Gflop/s (ms) MAGMA CUBLAS LAPACK error
%============================================================================================================
1 1 0.00 ( 335.65) 0.00 ( 0.10) --- ( --- ) 4.84e+00 3.19e-07 --- failed
2 2 0.00 ( 214.03) 0.00 ( 0.20) --- ( --- ) 1.41e+00 1.31e-07 --- failed
3 3 0.00 ( 210.34) 0.00 ( 0.09) --- ( --- ) 1.47e+00 2.55e-07 --- failed

all test fails,

I know I got a weired setup from cuda capable 6.1 and 2.0, But never the less, they should work together nicely since data transfers is from gpu to cpu and back, and not gpu to gpu, so each Gpu will handle its own Cuda code.

Somebody might argue that 2.0 is too old, still many computing hardware, could still be using 2.0, cause its just too expensive to ditch it out.

Please help on fixing the test failed.
organicchemistry_01
 
Posts: 4
Joined: Mon Dec 26, 2016 2:21 am

Re: Help! Magma Test Failed on Multi GPU Setup

Postby mgates3 » Wed Jul 26, 2017 1:04 pm

It's hard to tell exactly what is going on here, because it's lacking the input, i.e., the command line to say what tester this is and what options were used. I guess that it is dtrsm. You can try the --dev option to specify a particular device; otherwise it defaults to device 0.

What is your make.inc file?

Please include the full input (including any relevant environment variables like OMP_NUM_THREADS) & output, e.g.:

Code: Select all
mint magma/testing> echo $OMP_NUM_THREADS
4
mint magma/testing> ./testing_dtrsm -n 1:5:1 --dev 0 -c
% MAGMA 2.2.0 svn compiled for CUDA capability >= 3.0, 32-bit magma_int_t, 64-bit pointer.
% CUDA runtime 8000, driver 8000. OpenMP threads 4.
% device 0: GeForce GT 750M, 925.5 MHz clock, 2047.6 MiB memory, capability 3.0
% Wed Jul 26 12:58:51 2017
% Usage: ./testing_dtrsm [options] [-h|--help]

% side = Left, uplo = Lower, transA = No transpose, diag = Non-unit, ngpu = 1
%   M     N  MAGMA Gflop/s (ms)  CUBLAS Gflop/s (ms)   CPU Gflop/s (ms)      MAGMA     CUBLAS   LAPACK error
%============================================================================================================
    1     1      0.00 (   1.83)      0.00 (   0.68)     ---   (  ---  )   0.00e+00   0.00e+00     ---      ok
    2     2      0.00 (   0.51)      0.00 (   0.08)     ---   (  ---  )   2.44e-16   4.88e-16     ---      ok
    3     3      0.00 (   0.40)      0.00 (   0.07)     ---   (  ---  )   2.37e-16   2.37e-16     ---      ok
    4     4      0.00 (   0.38)      0.00 (   0.07)     ---   (  ---  )   7.54e-17   7.54e-17     ---      ok
    5     5      0.00 (   0.36)      0.00 (   0.11)     ---   (  ---  )   2.48e-16   1.86e-16     ---      ok


Sorry for the delay; these got diverted to my spam folder for some reason.
mgates3
 
Posts: 736
Joined: Fri Jan 06, 2012 2:13 pm


Return to User discussion

Who is online

Users browsing this forum: No registered users and 2 guests

cron