9:34 p.m.
Here are my results (Fedora Core 19 x64):
first the shell output:
load_function_package_from_file:
'/home/helio/gretl/functions/matrix_perf.gfn' is already loaded
detect blas: confused, found too many libs!
-----
[helio@localhost gretl]$ ldd /usr/local/bin/gretl_x11
linux-vdso.so.1 => (0x00007fffb9bfe000)
libgtksourceview-3.0.so.1 => /lib64/libgtksourceview-3.0.so.1
(0x00000033d9200000)
libgtk-3.so.0 => /lib64/libgtk-3.so.0 (0x00000033d7800000)
libgdk-3.so.0 => /lib64/libgdk-3.so.0 (0x00000033d7400000)
libatk-1.0.so.0 => /lib64/libatk-1.0.so.0 (0x00000035a7e00000)
libgio-2.0.so.0 => /lib64/libgio-2.0.so.0 (0x00000035a6600000)
libpangocairo-1.0.so.0 => /lib64/libpangocairo-1.0.so.0
(0x00000033d8600000)
libcairo-gobject.so.2 => /lib64/libcairo-gobject.so.2
(0x00000033d7000000)
libpango-1.0.so.0 => /lib64/libpango-1.0.so.0 (0x0000003098c00000)
libcairo.so.2 => /lib64/libcairo.so.2 (0x00000033d8a00000)
libgretl-1.0.so.10 => /usr/local/lib/libgretl-1.0.so.10
(0x00007f87cce3e000)
liblapack.so.3 => /usr/lib64/atlas/liblapack.so.3
(0x00007f87cc5da000)
libblas.so.3 => /lib64/libblas.so.3 (0x000000316e600000)
libgfortran.so.3 => /lib64/libgfortran.so.3 (0x000000327f600000)
libm.so.6 => /lib64/libm.so.6 (0x000000327d600000)
libdl.so.2 => /lib64/libdl.so.2 (0x000000327ca00000)
libz.so.1 => /lib64/libz.so.1 (0x000000327d200000)
libxml2.so.2 => /lib64/libxml2.so.2 (0x0000003283200000)
libgmp.so.10 => /lib64/libgmp.so.10 (0x0000003e4b000000)
libfftw3.so.3 => /lib64/libfftw3.so.3 (0x00000036ca600000)
libcurl.so.4 => /lib64/libcurl.so.4 (0x00000036be600000)
libgdk_pixbuf-2.0.so.0 => /lib64/libgdk_pixbuf-2.0.so.0
(0x00000033d8200000)
libgobject-2.0.so.0 => /lib64/libgobject-2.0.so.0
(0x00000035a5e00000)
libglib-2.0.so.0 => /lib64/libglib-2.0.so.0 (0x00000035a5a00000)
libgomp.so.1 => /lib64/libgomp.so.1 (0x0000003285e00000)
libpthread.so.0 => /lib64/libpthread.so.0 (0x000000327ce00000)
libc.so.6 => /lib64/libc.so.6 (0x000000327c600000)
libgmodule-2.0.so.0 => /lib64/libgmodule-2.0.so.0
(0x00000035a6200000)
libX11.so.6 => /lib64/libX11.so.6 (0x0000003280a00000)
libXi.so.6 => /lib64/libXi.so.6 (0x0000003284600000)
libXfixes.so.3 => /lib64/libXfixes.so.3 (0x0000003284e00000)
libatk-bridge-2.0.so.0 => /lib64/libatk-bridge-2.0.so.0
(0x00000035a9e00000)
libpangoft2-1.0.so.0 => /lib64/libpangoft2-1.0.so.0
(0x0000003099000000)
libfontconfig.so.1 => /lib64/libfontconfig.so.1 (0x0000003284200000)
libXinerama.so.1 => /lib64/libXinerama.so.1 (0x0000003282a00000)
libXrandr.so.2 => /lib64/libXrandr.so.2 (0x0000003282e00000)
libXcursor.so.1 => /lib64/libXcursor.so.1 (0x0000003286200000)
libXcomposite.so.1 => /lib64/libXcomposite.so.1 (0x000000328b200000)
libXdamage.so.1 => /lib64/libXdamage.so.1 (0x0000003289a00000)
libwayland-client.so.0 => /lib64/libwayland-client.so.0
(0x000000328a600000)
libxkbcommon.so.0 => /lib64/libxkbcommon.so.0 (0x000000328ce00000)
libwayland-cursor.so.0 => /lib64/libwayland-cursor.so.0
(0x000000328e200000)
libXext.so.6 => /lib64/libXext.so.6 (0x0000003280e00000)
libffi.so.6 => /lib64/libffi.so.6 (0x000000327f200000)
libselinux.so.1 => /lib64/libselinux.so.1 (0x000000337d200000)
libresolv.so.2 => /lib64/libresolv.so.2 (0x000000327ee00000)
libgthread-2.0.so.0 => /lib64/libgthread-2.0.so.0
(0x00000035a6a00000)
libharfbuzz.so.0 => /lib64/libharfbuzz.so.0 (0x0000003179a00000)
libfreetype.so.6 => /lib64/libfreetype.so.6 (0x0000003281e00000)
libpixman-1.so.0 => /lib64/libpixman-1.so.0 (0x0000003287200000)
libEGL.so.1 => /lib64/libEGL.so.1 (0x000000337e600000)
libpng15.so.15 => /lib64/libpng15.so.15 (0x00000033d6c00000)
libxcb-shm.so.0 => /lib64/libxcb-shm.so.0 (0x0000003288600000)
libxcb-render.so.0 => /lib64/libxcb-render.so.0 (0x0000003288200000)
libxcb.so.1 => /lib64/libxcb.so.1 (0x0000003280600000)
libXrender.so.1 => /lib64/libXrender.so.1 (0x0000003282200000)
libGL.so.1 => /lib64/libGL.so.1 (0x000000337de00000)
librt.so.1 => /lib64/librt.so.1 (0x000000327da00000)
/lib64/ld-linux-x86-64.so.2 (0x000000327c200000)
libf77blas.so.3 => /usr/lib64/atlas/libf77blas.so.3
(0x00007f87cc3ac000)
libcblas.so.3 => /usr/lib64/atlas/libcblas.so.3 (0x000000364d000000)
libquadmath.so.0 => /lib64/libquadmath.so.0 (0x00007f87cc16f000)
libgcc_s.so.1 => /lib64/libgcc_s.so.1 (0x000000327ea00000)
liblzma.so.5 => /lib64/liblzma.so.5 (0x0000003281600000)
libidn.so.11 => /lib64/libidn.so.11 (0x000000329d000000)
libssh2.so.1 => /lib64/libssh2.so.1 (0x00000036be200000)
libssl3.so => /lib64/libssl3.so (0x0000003c22c00000)
libsmime3.so => /lib64/libsmime3.so (0x0000003c22800000)
libnss3.so => /lib64/libnss3.so (0x0000003c22000000)
libnssutil3.so => /lib64/libnssutil3.so (0x0000003c21c00000)
libplds4.so => /lib64/libplds4.so (0x0000003c21400000)
libplc4.so => /lib64/libplc4.so (0x0000003c21000000)
libnspr4.so => /lib64/libnspr4.so (0x0000003c21800000)
libgssapi_krb5.so.2 => /lib64/libgssapi_krb5.so.2
(0x00000036bd200000)
libkrb5.so.3 => /lib64/libkrb5.so.3 (0x00000036bca00000)
libk5crypto.so.3 => /lib64/libk5crypto.so.3 (0x00000036bda00000)
libcom_err.so.2 => /lib64/libcom_err.so.2 (0x0000003290200000)
liblber-2.4.so.2 => /lib64/liblber-2.4.so.2 (0x000000398c800000)
libldap-2.4.so.2 => /lib64/libldap-2.4.so.2 (0x0000003c24000000)
libatspi.so.0 => /lib64/libatspi.so.0 (0x00000035a9600000)
libdbus-1.so.3 => /lib64/libdbus-1.so.3 (0x0000003281200000)
libexpat.so.1 => /lib64/libexpat.so.1 (0x0000003283600000)
libpcre.so.1 => /lib64/libpcre.so.1 (0x000000337ce00000)
libgraphite2.so.3 => /lib64/libgraphite2.so.3 (0x000000328aa00000)
libX11-xcb.so.1 => /lib64/libX11-xcb.so.1 (0x0000003285600000)
libxcb-dri2.so.0 => /lib64/libxcb-dri2.so.0 (0x0000003289600000)
libxcb-xfixes.so.0 => /lib64/libxcb-xfixes.so.0 (0x0000003286600000)
libxcb-shape.so.0 => /lib64/libxcb-shape.so.0 (0x000000328be00000)
libgbm.so.1 => /lib64/libgbm.so.1 (0x000000337e200000)
libwayland-server.so.0 => /lib64/libwayland-server.so.0
(0x0000003287e00000)
libglapi.so.0 => /lib64/libglapi.so.0 (0x000000337ee00000)
libudev.so.1 => /lib64/libudev.so.1 (0x0000003281a00000)
libdrm.so.2 => /lib64/libdrm.so.2 (0x0000003286e00000)
libXau.so.6 => /lib64/libXau.so.6 (0x0000003280200000)
libxcb-glx.so.0 => /lib64/libxcb-glx.so.0 (0x0000003287600000)
libXxf86vm.so.1 => /lib64/libXxf86vm.so.1 (0x0000003283e00000)
libatlas.so.3 => /usr/lib64/atlas/libatlas.so.3 (0x000000364c200000)
libssl.so.10 => /lib64/libssl.so.10 (0x00000036bde00000)
libcrypto.so.10 => /lib64/libcrypto.so.10 (0x00000039a7a00000)
libkrb5support.so.0 => /lib64/libkrb5support.so.0
(0x00000036bd600000)
libkeyutils.so.1 => /lib64/libkeyutils.so.1 (0x00000036bce00000)
libsasl2.so.3 => /lib64/libsasl2.so.3 (0x0000003c23c00000)
libcrypt.so.1 => /lib64/libcrypt.so.1 (0x0000003c22400000)
libfreebl3.so => /lib64/libfreebl3.so (0x0000003c20c00000)
----
gretl versão 1.9.91cvs
Sessão atual: 2014-06-16 22:25
? include matrix_perf.gfn
/home/helio/gretl/functions/matrix_perf.gfn
? matrix_perf(1234)
dgemm experiment 1, variant 1, speed in Gflops
m n k vanilla openmp sysblas
128 128 128 0,76280 3,8224 1,2408
128 128 256 1,4744 4,4971 2,3319
128 128 512 1,5409 4,6394 2,2827
128 128 1024 1,6307 4,7384 2,4370
128 128 2048 1,6366 4,6429 2,3655
result: openmp dominates
dgemm experiment 1, variant 2, speed in Gflops
m n k vanilla openmp sysblas
128 128 128 1,2369 4,2107 2,3447
256 256 128 1,7027 4,2914 2,4852
512 512 128 1,7378 4,3442 2,4599
1024 1024 128 1,7530 4,5483 2,3357
2048 2048 128 1,6727 4,3896 2,4288
result: openmp dominates
dgemm experiment 1, variant 3, speed in Gflops
m n k vanilla openmp sysblas
128 128 128 1,3331 1,6034 1,0660
256 256 256 1,7022 4,6465 2,2175
512 512 512 1,7230 3,6227 2,6052
1024 1024 1024 1,5251 4,3004 2,0975
2048 2048 2048 1,5605 4,1371 2,0563
result: openmp dominates
dgemm experiment 2, variant 1, speed in Gflops
m n k vanilla openmp sysblas
8 8 8 0,54613 0,19646 0,48416
16 8 8 0,81994 0,38813 0,68170
32 8 8 1,0980 0,62199 0,85589
64 8 8 1,3154 0,82638 1,0962
128 8 8 1,4332 1,0507 1,2646
256 8 8 1,4836 1,2024 1,4424
512 8 8 1,5283 1,2608 1,5017
1024 8 8 1,5241 1,3197 1,4720
2048 8 8 1,5281 1,3507 1,5226
4096 8 8 1,5382 1,3684 1,5155
result: vanilla dominates
dgemm experiment 2, variant 2, speed in Gflops
m n k vanilla openmp sysblas
10 2 1000 1,2576 1,1964 1,6339
20 2 1000 1,3761 1,9025 1,8154
40 2 1000 1,2545 1,9819 1,8016
80 2 1000 1,6691 2,2706 2,1917
160 2 1000 1,5171 2,3899 2,4151
320 2 1000 1,6823 2,5837 2,3722
640 2 1000 1,5027 2,8140 2,1026
1280 2 1000 1,4856 2,8152 2,1165
2560 2 1000 1,5764 2,3430 2,0931
5120 2 1000 1,5695 3,4026 2,1107
result: openmp dominates for mnk >= 640000
dgemm experiment 2, variant 3, speed in Gflops
m n k vanilla openmp sysblas
10 10 1000 1,1356 2,6000 1,7212
20 10 1000 1,3239 3,8221 1,5229
40 10 1000 1,2920 3,7631 1,8438
80 10 1000 1,4216 4,1526 2,2323
160 10 1000 1,6549 4,3422 2,4932
320 10 1000 1,6679 4,3360 2,4248
result: openmp dominates
Operating system: Linux (64-bit)
BLAS library: sysblas
Number of processors: 4
OpenMP enabled: yes
Performance summary:
vanilla -
dominates outright in 1 out of 6 tests
openmp -
dominates outright in 4 out of 6 tests
dominates in 1 test(s) for mnk >= 640000
sysblas -
dominates outright in 0 out of 6 tests