Bug in DGETRF2_OOC?

Open discussion for MAGMA

Bug in DGETRF2_OOC?

Postby keitat » Fri May 04, 2012 5:19 pm

Hi,

I am testing dgetrf_ooc routine using testing_dgetrf.cpp. Then, I got inaccurate answer. See the output below. In the tester source, I call dgetrf_ooc and dgetrf_piv in sequence.
Is there a bug in dgetrf_ooc?

#if 0
magma_dgetrf( M, N, h_R, lda, ipiv, &info);
#else
magma_dgetrf_ooc( M, N, h_R, lda, ipiv, &info);
magma_dgetrf_piv( M, N, h_R, lda, ipiv, &info);
#endif

(The test code for dgetrf_ooc)
M N CPU GFlop/s GPU GFlop/s ||PA-LU||/(||A||*N)
============================================================
1024 1024 22.93 19.62 3.939029e-18
2048 2048 42.29 62.54 3.352813e-18
3072 3072 54.05 116.79 3.850485e-18
4032 4032 63.26 161.97 3.650564e-18
5184 5184 70.39 202.22 3.442783e-18
6016 6016 75.18 198.69 1.785890e-04
7040 7040 79.02 223.66 2.093331e-04
8064 8064 82.24 244.17 1.859847e-04
9088 9088 84.87 256.97 1.651123e-04
10112 10112 87.17 260.84 1.469614e-04
12112 12112 90.51 267.34 9.770538e-05
14112 14112 93.43 267.54 5.281521e-05

(The original test code)
M N CPU GFlop/s GPU GFlop/s ||PA-LU||/(||A||*N)
============================================================
1024 1024 22.86 20.61 3.939029e-18
2048 2048 42.38 60.72 3.352813e-18
3072 3072 54.01 110.46 3.850485e-18
4032 4032 64.17 155.78 3.650564e-18
5184 5184 71.36 194.95 3.442783e-18
6016 6016 75.10 216.85 3.320306e-18
7040 7040 78.92 238.00 3.232373e-18
8064 8064 82.12 254.61 3.273794e-18
9088 9088 84.94 268.94 3.191273e-18
10112 10112 87.09 276.30 3.143110e-18
12112 12112 90.58 269.87 3.059934e-18
keitat
 
Posts: 8
Joined: Tue Jan 24, 2012 2:19 pm

Return to User discussion

Who is online

Users browsing this forum: Google [Bot] and 3 guests

cron