On 22.01.2020 17:04, Sven S wrote:
Am 22.01.20 um 14:47 schrieb Marcin Błażejowski:
>
> Hi Sven,
>
> my machine is: 4 Hyper-Threaded Core i7-8550U CPU @ 1.80GHz.
>
> My results:
>
> OMP... OPENBLAS... best of 3 runs
> <unset/default> <unset/default> 7.54
> 4 4 7.24
> 4 1 8.36
> 1 4 12.66
> 1 1 12.51
Thanks Marcin, that's interesting! Actually I have the same CPU in a
laptop, so I could cross-check how the Windows package influences the
whole thing. (I'm assuming you were on Linux with that test?)
So as I suspected, it's not so easy in general. BTW, if you're really
on Linux, do you know what your openblas version is?
That is something which bothers me: I have OpenBLAS 0.3.7+ds-7 taken
from Debian repos (libopenblas-base:amd64 +
libopenblas-openmp-dev:amd64) and gretl links (after a long alternatives
chain) against '/usr/lib/x86_64-linux-gnu/openblas-openmp/libblas.so.3'
(and in case of LAPACK against
'/usr/lib/x86_64-linux-gnu/openblas-openmp/liblapack.so.3'). Looks good:
OpenBLAS witn OpenMP. But 'eval $sysinfo' produces following results:
bundle anonymous:
ncores = 4
mpi = 1
omp = 1
hostname = "marcin-note"
os = "linux"
omp_num_threads = 4
mpimax = 8
blas = "netlib"
nproc = 8
wordlen = 64
I remember that previous version of OpenBLAS either taken from repos or
compiled by myself was giving (among others):
blas = openblas
blas_parallel = OpenMP
I'm stuck.
Cheers,
Marcin
--
Marcin Błażejowski