Hi All,
I’m on Windows and have usually installed pymc
with pip
and used my own gcc
. I rarely installed via conda
like is recommended. I’ve been curious if I’ve really been missing out on better performance with MKL
vs OpenBLAS
that is packaged with numpy
on PyPi. I’m comparing a pip
install and conda
install right now (exactly according to instructions). I want to make sure I’m setting the right blas
flags in .pytensorrc
so it’s configured correctly.
I have tried:
[blas]
ldflags = -lmkl -lguide -lpthread
OR
[blas]
ldflags=-lblas
without much difference.
I want to see what you are all using. Could you fill out:
OS: Windows, Mac, Linux flavor
Install method: Conda/Pip
Known/Unknown BLAS used with Numpy:
.pytensorrc file:
```
[blas]
ldflags=-lblas
```
For what it’s worth, this is my numpy.show_config()
for my pip
installation:
> np.show_config()
openblas64__info:
libraries = ['openblas64_', 'openblas64_']
library_dirs = ['openblas\\lib']
language = c
define_macros = [('HAVE_CBLAS', None), ('BLAS_SYMBOL_SUFFIX', '64_'), ('HAVE_BLAS_ILP64', None)]
runtime_library_dirs = ['openblas\\lib']
blas_ilp64_opt_info:
libraries = ['openblas64_', 'openblas64_']
library_dirs = ['openblas\\lib']
language = c
define_macros = [('HAVE_CBLAS', None), ('BLAS_SYMBOL_SUFFIX', '64_'), ('HAVE_BLAS_ILP64', None)]
runtime_library_dirs = ['openblas\\lib']
openblas64__lapack_info:
libraries = ['openblas64_', 'openblas64_']
library_dirs = ['openblas\\lib']
language = c
define_macros = [('HAVE_CBLAS', None), ('BLAS_SYMBOL_SUFFIX', '64_'), ('HAVE_BLAS_ILP64', None), ('HAVE_LAPACKE', None)]
runtime_library_dirs = ['openblas\\lib']
lapack_ilp64_opt_info:
libraries = ['openblas64_', 'openblas64_']
library_dirs = ['openblas\\lib']
language = c
define_macros = [('HAVE_CBLAS', None), ('BLAS_SYMBOL_SUFFIX', '64_'), ('HAVE_BLAS_ILP64', None), ('HAVE_LAPACKE', None)]
runtime_library_dirs = ['openblas\\lib']
Supported SIMD extensions in this NumPy install:
baseline = SSE,SSE2,SSE3
found = SSSE3,SSE41,POPCNT,SSE42,AVX,F16C,FMA3,AVX2
not found = AVX512F,AVX512CD,AVX512_SKX,AVX512_CLX,AVX512_CNL,AVX512_ICL
and for the conda
install:
> np.show_config()
blas_info:
libraries = ['cblas', 'blas', 'cblas', 'blas', 'cblas', 'blas']
library_dirs = ['C:/Users/hectormz/miniconda3/envs/pymc_env\\Library\\lib']
include_dirs = ['C:/Users/hectormz/miniconda3/envs/pymc_env\\Library\\include']
language = f77
define_macros = [('HAVE_CBLAS', None)]
blas_opt_info:
define_macros = [('NO_ATLAS_INFO', 1), ('HAVE_CBLAS', None)]
libraries = ['cblas', 'blas', 'cblas', 'blas', 'cblas', 'blas']
library_dirs = ['C:/Users/hectormz/miniconda3/envs/pymc_env\\Library\\lib']
include_dirs = ['C:/Users/hectormz/miniconda3/envs/pymc_env\\Library\\include']
language = f77
lapack_info:
libraries = ['lapack', 'blas', 'lapack', 'blas']
library_dirs = ['C:/Users/hectormz/miniconda3/envs/pymc_env\\Library\\lib']
language = f77
lapack_opt_info:
libraries = ['lapack', 'blas', 'lapack', 'blas', 'cblas', 'blas', 'cblas', 'blas', 'cblas', 'blas']
library_dirs = ['C:/Users/hectormz/miniconda3/envs/pymc_env\\Library\\lib']
language = f77
define_macros = [('NO_ATLAS_INFO', 1), ('HAVE_CBLAS', None)]
include_dirs = ['C:/Users/hectormz/miniconda3/envs/pymc_env\\Library\\include']
Supported SIMD extensions in this NumPy install:
baseline = SSE,SSE2,SSE3
found = SSSE3,SSE41,POPCNT,SSE42,AVX,F16C,FMA3,AVX2
not found = AVX512F,AVX512CD,AVX512_SKX,AVX512_CLX,AVX512_CNL,AVX512_ICL