Other Languages
Intel Fortran Compiler
Intel's Fortran compiler.
-
Added support for the Fortran 2023 standard.
-
Enhanced support for OpenMP 5.0, 5.1, and 5.2. Compliant with the OpenMp 6.0 standard.
-
[Intel's Developer Guide (
Compiler Command Format
| Language | Command | Execution Format |
|---|---|---|
| Fortran | ifx | ifx [options] file_name |
Intel Distribution for Python
Intel Distribution for Python offers the following features, aiming for performance improvements compared to CPython, which uses the Global Interpreter Lock (GIL):
- Enhanced performance with Intel® MKL and Python* computational packages (NumPy, SciPy, and scikit-learn).
- Streamlined thread scheduling using Intel® TBB library as a Python* package.
- Accelerated code execution through Just-In-Time (JIT) compilation with Numba.
- Improved performance via mpi4py and Cython and Intel® MPI Library.
- Powerful data analytics and neural network functionalities implemented through pyDAAL and Intel® DAAL.
- Easily installed using conda* and PIP.
For more details, visit the following pages:
Environment Setup
This section explains how to install Intel Distribution for Python in your environment. The following steps are based on the information provided on this site:
We will cover two methods of environment setup: using conda and using the offline installer.
Environment Setup with conda
Here, we describe the steps to install the Intel Distribution for Python environment using conda.
Create a virtual environment with conda:
conda create -n intelpython3 python=3.10 -y
conda activate intelpython3
Install the intelpython3_full meta-package by entering the following command:
(intelpython3) yxxxx@igt001:~$ conda install -c intel intelpython3_full
Channels:
- intel
- conda-forge
- defaults
Platform: linux-64
Collecting package metadata (repodata.json): done
Solving environment: done
## Package Plan ##
environment location: /home/yxxxx/miniconda3/envs/intelpython3
added / updated specs:
- intelpython3_full
The following packages will be downloaded:
package | build
---------------------------|-----------------
asn1crypto-1.5.1 | pyhd8ed1ab_0 79 KB intel
brotli-1.1.0 | hd590300_1 19 KB intel
brotli-bin-1.1.0 | hd590300_1 19 KB intel
brotli-python-1.1.0 | py310hc6cd4ac_1 341 KB intel
brotlipy-0.7.0 |py310h2372a71_1006 310 KB intel
c-ares-1.21.0 | hd590300_0 119 KB intel
certifi-2023.7.22 | pyhd8ed1ab_0 150 KB intel
cffi-1.16.0 | py310h2fee648_0 236 KB intel
chardet-5.2.0 | py310hff52083_1 241 KB intel
charset-normalizer-3.3.2 | pyhd8ed1ab_0 46 KB intel
colorama-0.4.6 | pyhd8ed1ab_0 25 KB intel
common_cmplr_lib_rt-2024.0.3| intel_49895 21 KB intel
common_cmplr_lic_rt-2024.0.3| intel_49895 21 KB intel
conda-package-streaming-0.9.0| pyhd8ed1ab_0 19 KB intel
cryptography-41.0.5 | py310h75e40e8_0 1.9 MB intel
cycler-0.12.1 | pyhd8ed1ab_0 13 KB intel
cython-3.0.2 | py310h3fd9d12_0 3.3 MB intel
daal4py-2024.1.0 | py310_intel_299 8.5 MB intel
dal-2024.1.0 | intel_299 49.3 MB intel
dpcpp-llvm-spirv-2024.0.0 |py310haaad0e5_49588 12.5 MB intel
dpctl-0.15.0 | py310h9daa3b9_42 7.7 MB intel
dpnp-0.13.0 |py310ha1a3bb5_171 9.6 MB intel
fortran_rt-2024.0.3 | intel_49895 21 KB intel
freetype-2.12.1 | h267a509_2 620 KB intel
funcsigs-1.0.2 | py_3 14 KB intel
future-0.18.3 | pyhd8ed1ab_0 357 KB intel
gtest-1.14.0 | h00ab1b0_1 395 KB intel
icu-73.2 | h59595ed_0 11.5 MB intel
idna-3.4 | pyhd8ed1ab_0 55 KB intel
impi_rt-2021.11.0 | intel_49493 32.0 MB intel
intel-fortran-rt-2024.0.3 | intel_49895 1022 KB intel
intelpython3_core-2024.0.0 | py310_1 8 KB intel
intelpython3_full-2024.0.0 | py310_0 6 KB intel
ipp-2021.10.0 | intel_653 132.3 MB intel
joblib-1.3.2 | pyhd8ed1ab_0 216 KB intel
jsonpatch-1.33 | pyhd8ed1ab_0 17 KB intel
jsonpointer-2.4 | py310hff52083_3 16 KB intel
kiwisolver-1.4.5 | py310hd41b1e2_1 71 KB intel
level-zero-1.15.1 | h00ab1b0_0 224 KB intel
libabseil-20230802.1 | cxx17_h59595ed_0 1.2 MB intel
libbrotlicommon-1.1.0 | hd590300_1 68 KB intel
libbrotlidec-1.1.0 | hd590300_1 32 KB intel
libbrotlienc-1.1.0 | hd590300_1 276 KB intel
libevent-2.1.12 | hf998b51_1 417 KB intel
libgfortran-ng-13.2.0 | h69a702a_2 23 KB intel
libgfortran5-13.2.0 | ha4646dd_2 1.4 MB intel
libiconv-1.17 | h166bdaf_0 1.4 MB intel
libllvm14-14.0.6 | hcd5def8_4 30.0 MB intel
libpng-1.6.39 | h753d276_0 276 KB intel
libprotobuf-4.24.4 | hf27288f_0 2.4 MB intel
libxml2-2.11.5 | h232c23b_1 689 KB intel
llvm-14.0.6 | h32600fe_4 53 KB intel
llvm-spirv-14.0.0 | h2bc3f7f_0 1.4 MB intel
llvmlite-0.40.1 | py310h76e7cf5_0 3.1 MB intel
mkl-dpcpp-2024.0.0 | intel_49656 12 KB intel
numba-0.57.1 | py310h0f6aa51_0 4.0 MB intel
numba-dpex-0.21.4 | py310hce084a9_0 374 KB intel
onemkl-sycl-blas-2024.0.0 | intel_49656 14.6 MB intel
onemkl-sycl-datafitting-2024.0.0| intel_49656 758 KB intel
onemkl-sycl-dft-2024.0.0 | intel_49656 5.9 MB intel
onemkl-sycl-lapack-2024.0.0| intel_49656 8.0 MB intel
onemkl-sycl-rng-2024.0.0 | intel_49656 16.9 MB intel
onemkl-sycl-sparse-2024.0.0| intel_49656 15.9 MB intel
onemkl-sycl-stats-2024.0.0 | intel_49656 3.5 MB intel
onemkl-sycl-vm-2024.0.0 | intel_49656 34.5 MB intel
opencl_rt-2024.0.3 | intel_49895 21 KB intel
packaging-23.2 | pyhd8ed1ab_0 48 KB intel
platformdirs-3.11.0 | pyhd8ed1ab_0 20 KB intel
pluggy-1.3.0 | pyhd8ed1ab_0 22 KB intel
pooch-1.8.0 | pyhd8ed1ab_0 51 KB intel
pycparser-2.21 | pyhd8ed1ab_0 100 KB intel
pyopenssl-23.3.0 | pyhd8ed1ab_0 124 KB intel
pyparsing-3.1.1 | pyhd8ed1ab_0 87 KB intel
pysocks-1.7.1 | pyha2e5f31_6 19 KB intel
python-dateutil-2.8.2 | pyhd8ed1ab_0 240 KB intel
python_abi-3.10 | 2_cp310 4 KB intel
pyyaml-6.0.1 | py310h2372a71_1 167 KB intel
requests-2.31.0 | pyhd8ed1ab_0 55 KB intel
ruamel.yaml-0.17.40 | py310h2372a71_0 196 KB intel
ruamel.yaml.clib-0.2.7 | py310h2372a71_2 132 KB intel
scikit-learn-1.3.1 | py310h1fdf081_1 8.0 MB intel
scikit-learn-intelex-2024.1.0| py310_intel_299 54 KB intel
scipy-1.10.1 | py310h6681978_8 29.4 MB intel
six-1.16.0 | pyh6c4a22f_0 14 KB intel
smp-0.1.5 | py310h8271ca5_21 19 KB intel
snappy-1.1.10 | h9fff704_0 38 KB intel
spirv-tools-2023.2 | h00ab1b0_4 1.9 MB intel
threadpoolctl-3.2.0 | pyha21a80b_0 20 KB intel
toolz-0.12.0 | pyhd8ed1ab_0 48 KB intel
typing-extensions-4.8.0 | hd8ed1ab_0 10 KB intel
typing_extensions-4.8.0 | pyha770c72_0 34 KB intel
urllib3-2.0.7 | pyhd8ed1ab_0 96 KB intel
xgboost-1.7.3 |0_gade498py310_25 6.3 MB intel
yaml-0.2.5 | h7f98852_2 87 KB intel
zlib-1.2.13 | hd590300_5 91 KB intel
zstandard-0.22.0 | py310h1275a96_0 395 KB intel
zstd-1.5.5 | hfc55251_0 532 KB intel
------------------------------------------------------------
Total: 468.5 MB
The following NEW packages will be INSTALLED:
asn1crypto intel/noarch::asn1crypto-1.5.1-pyhd8ed1ab_0
brotli intel/linux-64::brotli-1.1.0-hd590300_1
brotli-bin intel/linux-64::brotli-bin-1.1.0-hd590300_1
brotli-python intel/linux-64::brotli-python-1.1.0-py310hc6cd4ac_1
brotlipy intel/linux-64::brotlipy-0.7.0-py310h2372a71_1006
c-ares intel/linux-64::c-ares-1.21.0-hd590300_0
certifi intel/noarch::certifi-2023.7.22-pyhd8ed1ab_0
cffi intel/linux-64::cffi-1.16.0-py310h2fee648_0
chardet intel/linux-64::chardet-5.2.0-py310hff52083_1
charset-normalizer intel/noarch::charset-normalizer-3.3.2-pyhd8ed1ab_0
colorama intel/noarch::colorama-0.4.6-pyhd8ed1ab_0
common_cmplr_lib_~ intel/linux-64::common_cmplr_lib_rt-2024.0.3-intel_49895
common_cmplr_lic_~ intel/linux-64::common_cmplr_lic_rt-2024.0.3-intel_49895
conda-package-str~ intel/noarch::conda-package-streaming-0.9.0-pyhd8ed1ab_0
cryptography intel/linux-64::cryptography-41.0.5-py310h75e40e8_0
cycler intel/noarch::cycler-0.12.1-pyhd8ed1ab_0
cython intel/linux-64::cython-3.0.2-py310h3fd9d12_0
daal4py intel/linux-64::daal4py-2024.1.0-py310_intel_299
dal intel/linux-64::dal-2024.1.0-intel_299
dpcpp-llvm-spirv intel/linux-64::dpcpp-llvm-spirv-2024.0.0-py310haaad0e5_49588
dpctl intel/linux-64::dpctl-0.15.0-py310h9daa3b9_42
dpnp intel/linux-64::dpnp-0.13.0-py310ha1a3bb5_171
fortran_rt intel/linux-64::fortran_rt-2024.0.3-intel_49895
freetype intel/linux-64::freetype-2.12.1-h267a509_2
funcsigs intel/noarch::funcsigs-1.0.2-py_3
future intel/noarch::future-0.18.3-pyhd8ed1ab_0
gtest intel/linux-64::gtest-1.14.0-h00ab1b0_1
icu intel/linux-64::icu-73.2-h59595ed_0
idna intel/noarch::idna-3.4-pyhd8ed1ab_0
impi_rt intel/linux-64::impi_rt-2021.11.0-intel_49493
intel-fortran-rt intel/linux-64::intel-fortran-rt-2024.0.3-intel_49895
intelpython3_core intel/linux-64::intelpython3_core-2024.0.0-py310_1
intelpython3_full intel/linux-64::intelpython3_full-2024.0.0-py310_0
ipp intel/linux-64::ipp-2021.10.0-intel_653
joblib intel/noarch::joblib-1.3.2-pyhd8ed1ab_0
jsonpatch intel/noarch::jsonpatch-1.33-pyhd8ed1ab_0
jsonpointer intel/linux-64::jsonpointer-2.4-py310hff52083_3
kiwisolver intel/linux-64::kiwisolver-1.4.5-py310hd41b1e2_1
level-zero intel/linux-64::level-zero-1.15.1-h00ab1b0_0
libabseil intel/linux-64::libabseil-20230802.1-cxx17_h59595ed_0
libbrotlicommon intel/linux-64::libbrotlicommon-1.1.0-hd590300_1
libbrotlidec intel/linux-64::libbrotlidec-1.1.0-hd590300_1
libbrotlienc intel/linux-64::libbrotlienc-1.1.0-hd590300_1
libevent intel/linux-64::libevent-2.1.12-hf998b51_1
libgfortran-ng intel/linux-64::libgfortran-ng-13.2.0-h69a702a_2
libgfortran5 intel/linux-64::libgfortran5-13.2.0-ha4646dd_2
libiconv intel/linux-64::libiconv-1.17-h166bdaf_0
libllvm14 intel/linux-64::libllvm14-14.0.6-hcd5def8_4
libpng intel/linux-64::libpng-1.6.39-h753d276_0
libprotobuf intel/linux-64::libprotobuf-4.24.4-hf27288f_0
libxml2 intel/linux-64::libxml2-2.11.5-h232c23b_1
llvm intel/linux-64::llvm-14.0.6-h32600fe_4
llvm-spirv intel/linux-64::llvm-spirv-14.0.0-h2bc3f7f_0
llvmlite intel/linux-64::llvmlite-0.40.1-py310h76e7cf5_0
mkl-dpcpp intel/linux-64::mkl-dpcpp-2024.0.0-intel_49656
numba intel/linux-64::numba-0.57.1-py310h0f6aa51_0
numba-dpex intel/linux-64::numba-dpex-0.21.4-py310hce084a9_0
onemkl-sycl-blas intel/linux-64::onemkl-sycl-blas-2024.0.0-intel_49656
onemkl-sycl-dataf~ intel/linux-64::onemkl-sycl-datafitting-2024.0.0-intel_49656
onemkl-sycl-dft intel/linux-64::onemkl-sycl-dft-2024.0.0-intel_49656
onemkl-sycl-lapack intel/linux-64::onemkl-sycl-lapack-2024.0.0-intel_49656
onemkl-sycl-rng intel/linux-64::onemkl-sycl-rng-2024.0.0-intel_49656
onemkl-sycl-sparse intel/linux-64::onemkl-sycl-sparse-2024.0.0-intel_49656
onemkl-sycl-stats intel/linux-64::onemkl-sycl-stats-2024.0.0-intel_49656
onemkl-sycl-vm intel/linux-64::onemkl-sycl-vm-2024.0.0-intel_49656
opencl_rt intel/linux-64::opencl_rt-2024.0.3-intel_49895
packaging intel/noarch::packaging-23.2-pyhd8ed1ab_0
platformdirs intel/noarch::platformdirs-3.11.0-pyhd8ed1ab_0
pluggy intel/noarch::pluggy-1.3.0-pyhd8ed1ab_0
pooch intel/noarch::pooch-1.8.0-pyhd8ed1ab_0
pycparser intel/noarch::pycparser-2.21-pyhd8ed1ab_0
pyopenssl intel/noarch::pyopenssl-23.3.0-pyhd8ed1ab_0
pyparsing intel/noarch::pyparsing-3.1.1-pyhd8ed1ab_0
pysocks intel/noarch::pysocks-1.7.1-pyha2e5f31_6
python_abi intel/linux-64::python_abi-3.10-2_cp310
pyyaml intel/linux-64::pyyaml-6.0.1-py310h2372a71_1
requests intel/noarch::requests-2.31.0-pyhd8ed1ab_0
ruamel.yaml intel/linux-64::ruamel.yaml-0.17.40-py310h2372a71_0
ruamel.yaml.clib intel/linux-64::ruamel.yaml.clib-0.2.7-py310h2372a71_2
scikit-learn intel/linux-64::scikit-learn-1.3.1-py310h1fdf081_1
scikit-learn-inte~ intel/linux-64::scikit-learn-intelex-2024.1.0-py310_intel_299
scipy intel/linux-64::scipy-1.10.1-py310h6681978_8
smp intel/linux-64::smp-0.1.5-py310h8271ca5_21
snappy intel/linux-64::snappy-1.1.10-h9fff704_0
spirv-tools intel/linux-64::spirv-tools-2023.2-h00ab1b0_4
threadpoolctl intel/noarch::threadpoolctl-3.2.0-pyha21a80b_0
toolz intel/noarch::toolz-0.12.0-pyhd8ed1ab_0
typing-extensions intel/noarch::typing-extensions-4.8.0-hd8ed1ab_0
typing_extensions intel/noarch::typing_extensions-4.8.0-pyha770c72_0
urllib3 intel/noarch::urllib3-2.0.7-pyhd8ed1ab_0
xgboost intel/linux-64::xgboost-1.7.3-0_gade498py310_25
yaml intel/linux-64::yaml-0.2.5-h7f98852_2
zlib intel/linux-64::zlib-1.2.13-hd590300_5
zstandard intel/linux-64::zstandard-0.22.0-py310h1275a96_0
zstd intel/linux-64::zstd-1.5.5-hfc55251_0
The following packages will be SUPERSEDED by a higher-priority channel:
ca-certificates conda-forge::ca-certificates-2024.2.2~ --> intel::ca-certificates-2023.7.22-hbcca054_0
libgcc-ng conda-forge::libgcc-ng-13.2.0-h807b86~ --> intel::libgcc-ng-13.2.0-h807b86a_2
libgomp conda-forge::libgomp-13.2.0-h807b86a_5 --> intel::libgomp-13.2.0-h807b86a_2
libsqlite conda-forge::libsqlite-3.45.1-h279700~ --> intel::libsqlite-3.44.0-h2797004_0
openssl conda-forge::openssl-3.2.1-hd590300_0 --> intel::openssl-3.1.4-hd590300_0
pip conda-forge::pip-24.0-pyhd8ed1ab_0 --> intel::pip-23.3.1-pyhd8ed1ab_0
setuptools conda-forge::setuptools-69.1.1-pyhd8e~ --> intel::setuptools-68.2.2-pyhd8ed1ab_0
tzdata conda-forge::tzdata-2024a-h0c530f3_0 --> intel::tzdata-2023c-h71feb2d_0
wheel conda-forge::wheel-0.42.0-pyhd8ed1ab_0 --> intel::wheel-0.41.3-pyhd8ed1ab_0
The following packages will be DOWNGRADED:
python-dateutil 2.8.2-pyhd3eb1b0_0 --> 2.8.2-pyhd8ed1ab_0
six 1.16.0-pyhd3eb1b0_1 --> 1.16.0-pyh6c4a22f_0
Proceed ([y]/n)? y
When prompted with Proceed([y]/n)?, respond with y to proceed with the installation.
After completion, you can check the installed packages with conda list. The installation includes a comprehensive set of packages enhanced for performance, including Intel® MKL, Intel® TBB, Numba, mpi4py, Cython, pyDAAL, and more, ensuring an optimized Python environment for computational tasks.
Steps for Environment Setup Using the Offline Installer
Next, we'll explain how to set up the Intel Distribution for Python environment in your user environment using the offline installer.
Download the offline installer using wget:
wget https://registrationcenter-download.intel.com/akdlm/IRC_NAS/eb21f0d0-bb7a-44b3-84d1-6489b0138b2e-2/l_AITools.2024.0.2.1.sh
Verify that the file has been downloaded:
$ ls -l l_AITools.2024.0.2.1.sh
-rw-r--r-- 1 yxxxx xxxxx 2068849631 Dec 20 21:28 l_AITools.2024.0.2.1.sh
Execute the following command with regular user permissions:
(base) yxxxx@igt003:~$ sh ./l_AITools.2024.0.2.1.sh
The following message will be displayed. The End User Agreement will appear, so please take a moment to read through it. Press the Enter key and the space key several times to scroll through the text:
Welcome to AITools 2024.0.2
In order to continue the installation process, please review the license
agreement.
Please, press ENTER to continue
>>>
Intel End User License Agreement for Developer Tools (Version October 2021)
IMPORTANT NOTICE - PLEASE READ AND AGREE BEFORE DOWNLOADING, INSTALLING, COPYING
OR USING
This Agreement is between you, or the company or other legal entity that you
represent and warrant you have the legal authority to bind, (each, "You" or
"Your") and Intel Corporation and its subsidiaries (collectively, "Intel")
regarding Your use of the Materials. By downloading, installing, copying or
using the Materials, You agree to be bound by the terms of this Agreement. If
You do not agree to the terms of this Agreement, or do not have legal authority
or required age to agree to them, do not download, install, copy or use the
Materials.
(omitted for brevity)
12.4 SEVERABILITY. The parties intend that if a court holds that any provision
or part of this Agreement is invalid or unenforceable under applicable law,
the court will modify the provision to the minimum extent necessary to make
it valid and enforceable, or if it cannot be made valid and enforceable,
the parties intend that the court will sever and delete the provision or
part from this Agreement. Any change to or deletion of a provision or part
of this Agreement under this Section will not affect the validity or
enforceability of the remainder of this Agreement, which will continue in
full force and effect.
You will be asked to accept the license terms. Please answer with yes:
Do you accept the license terms? [yes|no]
>>> yes
You will be prompted to confirm the installation directory. If there's no special reason to change it, simply press Enter to proceed:
AITools will now be installed into this location:
/home/yxxxx/intel/oneapi/intelpython
- Press ENTER to confirm the location
- Press CTRL-C to abort the installation
- Or specify a different location below
[/home/yxxxx/intel/oneapi/intelpython] >>>
The installation will begin and take some time to complete:
PREFIX=/home/yxxxx/intel/oneapi/intelpython
Unpacking payload ...
Installing base environment...
Downloading and Extracting Packages:
(omitted)
Installing pytorch-gpu environment...
(omitted)
Installing pytorch environment...
Downloading and Extracting Packages:
(omitted)
Installing tensorflow-gpu environment...
Downloading and Extracting Packages:
(omitted)
Installing tensorflow environment...
Downloading and Extracting Packages:
Downloading and Extracting Packages:
Preparing transaction: done
Executing transaction: done
Thanks for installing the Intel(R) AI Tools Bundle, the default path is <home_path>/intel/oneapi/intelpython. If you set up the installation path with -p tag, we recomend to add /intel/oneapi/intelpython, i.e. <path>/intel/oneapi/intelpython.
1. After installing, initialize your installed AI Tool with the following command:
$ source <home_path>/intel/oneapi/intelpython/bin/activate
2. Next conda environments are included:
- intelpython (base)
- tensorflow for CPU and GPU
- pytorch for CPU and GPU
- modin
You can then activate different conda environments as needed via the following command:
$ conda activate <conda environment>
3. AI Reference Models for Intel(R) Architecture can be
found in your installation of Intel(R) AI Tool, typically found at:
<home_path>/intel/oneapi/ai_reference_models
After the installation is complete, you will be prompted whether you want to modify your shell environment to automatically initialize conda. This is a matter of personal preference, but here we choose not to by selecting no:
Do you wish to update your shell profile to automatically initialize conda?
This will activate conda on startup and change the command prompt when activated.
If you'd prefer that conda's base environment not be activated on startup,
run the following command when conda is activated:
conda config --set auto_activate_base false
You can undo this by running `conda init --reverse $SHELL`? [yes|no]
[no] >>>
You have chosen to not have conda modify your shell scripts at all.
To activate conda's base environment in your current shell session:
eval "$(/home/yxxxx/intel/oneapi/intelpython/bin/conda shell.YOUR_SHELL_NAME hook)"
To install conda's shell functions for easier access, first activate, then:
conda init
Thank you for installing AITools!
The installation is complete. Enter the command displayed in the message to load the environment. In this example, the login shell is bash:
eval "$(/home/yxxxx/intel/oneapi/intelpython/bin/conda shell.bash hook)"
You can verify the installed environment with conda, where you'll see Intel's packages have been installed, providing a fully optimized Python environment for your computational needs.
(base) yxxxx@at138:~$ conda list
# packages in environment at /home/yxxxx/intel/oneapi/intelpython:
#
# Name Version Build Channel
_libgcc_mutex 0.1 conda_forge intel
_openmp_mutex 4.5 2_gnu intel
archspec 0.2.2 pyhd8ed1ab_0 conda-forge
boltons 23.0.0 pyhd8ed1ab_0 conda-forge
brotli 1.1.0 hd590300_1 intel
brotli-bin 1.1.0 hd590300_1 intel
brotli-python 1.1.0 py39h3d6467e_1 intel
bzip2 1.0.8 h7f98852_4 intel
c-ares 1.20.1 hd590300_0 intel
ca-certificates 2023.7.22 hbcca054_0 intel
certifi 2023.7.22 pyhd8ed1ab_0 intel
cffi 1.16.0 py39h7a31438_0 intel
charset-normalizer 3.3.0 pyhd8ed1ab_0 intel
colorama 0.4.6 pyhd8ed1ab_0 intel
conda 23.9.0 py39hf3d152e_2 conda-forge
conda-libmamba-solver 23.11.1 pyhd8ed1ab_0 conda-forge
conda-package-handling 2.2.0 pyh38be061_0 intel
conda-package-streaming 0.9.0 pyhd8ed1ab_0 intel
contourpy 1.0.7 py39h4b4f3f3_0 intel
cryptography 41.0.4 py39hd4f0224_0 intel
cycler 0.12.1 pyhd8ed1ab_0 intel
daal4py 2024.0.0 py39_intel_49530 intel
dal 2024.0.0 intel_49530 intel
dpcpp-cpp-rt 2024.0.0 intel_49795 intel
dpcpp-llvm-spirv 2024.0.0 py39haaad0e5_49588 intel
dpcpp_cpp_rt 2024.0.0 intel_49795 intel
dpctl 0.15.0 py39h9daa3b9_42 intel
dpnp 0.13.0 py39ha1a3bb5_170 intel
fmt 10.1.1 h00ab1b0_1 conda-forge
fonttools 4.39.4 py39hd1e30aa_0 intel
fortran_rt 2024.0.0 intel_49795 intel
freetype 2.12.1 h267a509_2 intel
icc_rt 2024.0.0 intel_49795 intel
icu 73.2 h59595ed_0 intel
idna 3.4 pyhd8ed1ab_0 intel
importlib-resources 5.12.0 pyhd8ed1ab_0 intel
importlib_resources 5.12.0 pyhd8ed1ab_0 intel
intel-cmplr-lib-rt 2024.0.0 intel_49795 intel
intel-cmplr-lic-rt 2024.0.0 intel_49795 intel
intel-fortran-rt 2024.0.0 intel_49795 intel
intel-opencl-rt 2024.0.0 intel_49795 intel
intel-openmp 2024.0.0 intel_49795 intel
intelpython 2024.0.0 0 intel
ipp 2021.10.0 intel_555 intel
joblib 1.3.2 pyhd8ed1ab_0 intel
jsonpatch 1.33 pyhd8ed1ab_0 intel
jsonpointer 2.4 py39hf3d152e_3 intel
keyutils 1.6.1 h166bdaf_1 intel
kiwisolver 1.4.5 py39h7633fee_1 intel
krb5 1.21.2 h659d440_0 intel
lcms2 2.16 hb7c19ff_0 conda-forge
lerc 4.0.0 h27087fc_1 intel
level-zero 1.14.0 h00ab1b0_2 intel
libarchive 3.7.2 h039dbb9_0 conda-forge
libbrotlicommon 1.1.0 hd590300_1 intel
libbrotlidec 1.1.0 hd590300_1 intel
libbrotlienc 1.1.0 hd590300_1 intel
libcurl 8.4.0 hca28451_0 intel
libdeflate 1.19 hd590300_0 intel
libedit 3.1.20221030 h5eee18b_1 intel
libev 4.33 h516909a_2 intel
libffi 3.4.2 h7f98852_5 intel
libgcc-ng 13.2.0 h807b86a_2 intel
libgomp 13.2.0 h807b86a_2 intel
libiconv 1.17 h166bdaf_0 intel
libjpeg-turbo 3.0.0 hd590300_1 intel
libllvm14 14.0.6 hcd5def8_4 intel
libmamba 1.5.4 had39da4_0 conda-forge
libmambapy 1.5.4 py39h10defb6_0 conda-forge
libnghttp2 1.57.0 h47da74e_0 intel
libnsl 2.0.1 hd590300_0 intel
libpng 1.6.39 h753d276_0 intel
libsolv 0.7.27 hfc55251_0 conda-forge
libsqlite 3.43.2 h2797004_0 intel
libssh2 1.11.0 h0841786_1 intel
libstdcxx-ng 13.2.0 h7e041cc_2 intel
libtiff 4.6.0 ha9c0a0a_2 intel
libuuid 2.38.1 h0b41bf4_0 intel
libwebp-base 1.3.2 hd590300_0 intel
libxcb 1.15 h0b41bf4_1 intel
libxml2 2.11.5 h232c23b_1 intel
libzlib 1.2.13 hd590300_5 intel
llvmlite 0.40.1 py39h76e7cf5_0 intel
lz4-c 1.9.4 hcb278e6_0 intel
lzo 2.10 h516909a_1000 intel
matplotlib-base 3.7.1 py39he190548_0 intel
mkl 2024.0.0 intel_49630 intel
mkl-dpcpp 2024.0.0 intel_49630 intel
mkl-service 2.4.0 py39h3539a15_40 intel
mkl_fft 1.3.6 py39h1d81ff8_61 intel
mkl_random 1.2.2 py39h5a378b4_81 intel
mkl_umath 0.1.1 py39h2b1685c_91 intel
munkres 1.1.4 pyh9f0ad1d_0 intel
ncurses 6.4 hcb278e6_0 intel
numba 0.57.1 py39hb75a051_0 intel
numba-dpex 0.21.4 py39hce084a9_0 intel
numpy 1.24.3 py39ha320b8e_5 intel
numpy-base 1.24.3 py39hbac2b65_5 intel
onemkl-sycl-blas 2024.0.0 intel_49630 intel
onemkl-sycl-datafitting 2024.0.0 intel_49630 intel
onemkl-sycl-dft 2024.0.0 intel_49630 intel
onemkl-sycl-lapack 2024.0.0 intel_49630 intel
onemkl-sycl-rng 2024.0.0 intel_49630 intel
onemkl-sycl-sparse 2024.0.0 intel_49630 intel
onemkl-sycl-stats 2024.0.0 intel_49630 intel
onemkl-sycl-vm 2024.0.0 intel_49630 intel
openjpeg 2.5.0 h488ebb8_3 intel
openssl 3.2.0 hd590300_1 conda-forge
packaging 23.2 pyhd8ed1ab_0 intel
pandas 2.1.3 py39hddac248_0 intel
pillow 10.0.1 py39had0adad_2 intel
pip 23.3 pyhd8ed1ab_0 intel
platformdirs 3.11.0 pyhd8ed1ab_0 intel
pluggy 1.3.0 pyhd8ed1ab_0 intel
pooch 1.7.0 pyhd8ed1ab_4 intel
pthread-stubs 0.4 h36c2ea0_1002 intel
pybind11-abi 4 hd8ed1ab_3 conda-forge
pycosat 0.6.6 py39hd1e30aa_0 intel
pycparser 2.21 pyhd8ed1ab_0 intel
pyopenssl 23.2.0 pyhd8ed1ab_1 intel
pyparsing 3.1.1 pyhd8ed1ab_0 intel
pysocks 1.7.1 pyha2e5f31_6 intel
python 3.9.18 h4499717_0 intel
python-dateutil 2.8.2 pyhd8ed1ab_0 intel
python-tzdata 2023.3 pyhd8ed1ab_0 intel
python_abi 3.9 2_cp39 intel
pytz 2023.3.post1 pyhd8ed1ab_0 intel
pyyaml 6.0.1 py39hd1e30aa_1 intel
readline 8.2 h8228510_1 intel
reproc 14.2.4.post0 hd590300_1 conda-forge
reproc-cpp 14.2.4.post0 h59595ed_1 conda-forge
requests 2.31.0 pyhd8ed1ab_0 intel
ruamel.yaml 0.17.36 py39hd1e30aa_0 intel
ruamel.yaml.clib 0.2.7 py39hd1e30aa_2 intel
scikit-learn 1.3.1 py39ha22ef79_1 intel
scikit-learn-intelex 2024.0.0 py39_intel_49530 intel
scipy 1.10.1 py39h6681978_6 intel
setuptools 68.2.2 pyhd8ed1ab_0 intel
six 1.16.0 pyh6c4a22f_0 intel
tbb 2021.11.0 intel_49403 intel
tbb4py 2021.11.0 py39_intel_49403 intel
threadpoolctl 3.2.0 pyha21a80b_0 intel
tk 8.6.13 h2797004_0 intel
tqdm 4.66.1 pyhd8ed1ab_0 intel
typing-extensions 4.8.0 hd8ed1ab_0 intel
typing_extensions 4.8.0 pyha770c72_0 intel
tzdata 2023c h71feb2d_0 intel
unicodedata2 15.0.0 py39hb9d737c_0 intel
urllib3 2.0.7 pyhd8ed1ab_0 intel
wheel 0.41.2 pyhd8ed1ab_0 intel
xgboost 1.7.3 0_gade498py39_25 intel
xorg-libxau 1.0.11 hd590300_1 intel
xorg-libxdmcp 1.1.3 h7f98852_1 intel
xz 5.2.6 h166bdaf_0 intel
yaml 0.2.5 h7f98852_2 intel
yaml-cpp 0.8.0 h59595ed_0 conda-forge
zipp 3.17.0 pyhd8ed1ab_0 intel
zstandard 0.21.0 py39h6e5214e_1 intel
zstd 1.5.5 hfc55251_0 intel
It is evident that packages such as numpy, scipy, and scikit-learn have also been introduced from the Intel channel.
Basic Operation Check with Sample Programs
Let's use the sample programs distributed by Intel to check the environment.
Within this collection of samples, try running the following sample:
This sample utilizes the Intel Extension for PyTorch, an open-source software developed by Intel. For reference, see:
First, create a virtual environment with conda:
$ conda create -n pytorch python=3.10
Channels:
- conda-forge
- defaults
Platform: linux-64
Collecting package metadata (repodata.json): done
Solving environment: done
## Package Plan ##
environment location: /home/yxxxx/miniconda3/envs/pytorch
added / updated specs:
- python=3.10
(The rest of the package download and installation process is omitted for brevity.)
Proceed ([y]/n)? y
Activate the newly created virtual environment and install the Intel Distribution for Python on it:
$ conda activate pytorch
$ mamba install -c intel intelpython3_full
(Looking for 'intelpython3_full' and the subsequent download and installation process is omitted for brevity.)
Install PyTorch using pip. This step is necessary because the Intel Extension for PyTorch does not have a conda package, and managing version compatibility through conda can be cumbersome:
$ python -m pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cpu
(Assuming installation success, details of packages installed via pip are shown.)
Next, install the intel-extension-for-pytorch using pip:
(pytorch) yxxxx@igt003:~/src/oneAPI-samples/AI-and-Analytics/Getting-Started-Samples/Intel_Extension_For_PyTorch_GettingStarted$ python -m pip install intel-extension-for-pytorch oneccl_bind_pt --extra-index-url https://pytorch-extension.intel.com/release-whl/stable/cpu/us/
(Assuming installation success, details of packages installed via pip are shown.)
You can verify the installation state. You will notice packages installed from PyPI, conda forge, and the Intel channel.
(pytorch)$ mamba list
# packages in environment at /home/yxxxx/miniconda3/envs/pytorch:
#
# Name Version Build Channel
_libgcc_mutex 0.1 conda_forge conda-forge
_openmp_mutex 4.5 2_gnu conda-forge
asn1crypto 1.5.1 pyhd8ed1ab_0 intel
brotli 1.1.0 hd590300_1 intel
brotli-bin 1.1.0 hd590300_1 intel
brotli-python 1.1.0 py310hc6cd4ac_1 intel
brotlipy 0.7.0 py310h2372a71_1006 intel
bzip2 1.0.8 hd590300_5 conda-forge
c-ares 1.21.0 hd590300_0 intel
ca-certificates 2023.7.22 hbcca054_0 intel
certifi 2023.7.22 pyhd8ed1ab_0 intel
cffi 1.16.0 py310h2fee648_0 intel
chardet 5.2.0 py310hff52083_1 intel
charset-normalizer 3.3.2 pyhd8ed1ab_0 intel
colorama 0.4.6 pyhd8ed1ab_0 intel
common_cmplr_lib_rt 2024.0.3 intel_49895 intel
common_cmplr_lic_rt 2024.0.3 intel_49895 intel
conda-package-streaming 0.9.0 pyhd8ed1ab_0 intel
cryptography 41.0.5 py310h75e40e8_0 intel
cycler 0.12.1 pyhd8ed1ab_0 intel
cython 3.0.2 py310h3fd9d12_0 intel
daal4py 2024.1.0 py310_intel_299 intel
dal 2024.1.0 intel_299 intel
dpcpp-cpp-rt 2024.0.3 intel_49895 intel
dpcpp-llvm-spirv 2024.0.0 py310haaad0e5_49588 intel
dpcpp_cpp_rt 2024.0.3 intel_49895 intel
dpctl 0.15.0 py310h9daa3b9_42 intel
dpnp 0.13.0 py310ha1a3bb5_171 intel
filelock 3.9.0 pypi_0 pypi
fortran_rt 2024.0.3 intel_49895 intel
freetype 2.12.1 h267a509_2 intel
fsspec 2023.4.0 pypi_0 pypi
funcsigs 1.0.2 py_3 intel
future 0.18.3 pyhd8ed1ab_0 intel
gtest 1.14.0 h00ab1b0_1 intel
icc_rt 2024.0.3 intel_49895 intel
icu 73.2 h59595ed_0 intel
idna 3.4 pyhd8ed1ab_0 intel
impi_rt 2021.11.0 intel_49493 intel
intel-cmplr-lib-rt 2024.0.3 intel_49895 intel
intel-cmplr-lic-rt 2024.0.3 intel_49895 intel
intel-extension-for-pytorch 2.2.0+cpu pypi_0 pypi
intel-fortran-rt 2024.0.3 intel_49895 intel
intel-opencl-rt 2024.0.3 intel_49895 intel
intel-openmp 2024.0.3 intel_49895 intel
intelpython 2024.0.0 0 intel
intelpython3_core 2024.0.0 py310_1 intel
intelpython3_full 2024.0.0 py310_0 intel
ipp 2021.10.0 intel_653 intel
jinja2 3.1.2 pypi_0 pypi
joblib 1.3.2 pyhd8ed1ab_0 intel
jsonpatch 1.33 pyhd8ed1ab_0 intel
jsonpointer 2.4 py310hff52083_3 intel
kiwisolver 1.4.5 py310hd41b1e2_1 intel
ld_impl_linux-64 2.40 h41732ed_0 conda-forge
level-zero 1.15.1 h00ab1b0_0 intel
libabseil 20230802.1 cxx17_h59595ed_0 intel
libbrotlicommon 1.1.0 hd590300_1 intel
libbrotlidec 1.1.0 hd590300_1 intel
libbrotlienc 1.1.0 hd590300_1 intel
libevent 2.1.12 hf998b51_1 intel
libffi 3.4.2 h7f98852_5 conda-forge
libgcc-ng 13.2.0 h807b86a_2 intel
libgfortran-ng 13.2.0 h69a702a_2 intel
libgfortran5 13.2.0 ha4646dd_2 intel
libgomp 13.2.0 h807b86a_2 intel
libiconv 1.17 h166bdaf_0 intel
libllvm14 14.0.6 hcd5def8_4 intel
libnsl 2.0.1 hd590300_0 conda-forge
libpng 1.6.39 h753d276_0 intel
libprotobuf 4.24.4 hf27288f_0 intel
libsqlite 3.44.0 h2797004_0 intel
libstdcxx-ng 13.2.0 h7e041cc_2 intel
libuuid 2.38.1 h0b41bf4_0 conda-forge
libxcrypt 4.4.36 hd590300_1 conda-forge
libxml2 2.11.5 h232c23b_1 intel
libzlib 1.2.13 hd590300_5 conda-forge
llvm 14.0.6 h32600fe_4 intel
llvm-spirv 14.0.0 h2bc3f7f_0 intel
llvmlite 0.40.1 py310h76e7cf5_0 intel
markupsafe 2.1.3 pypi_0 pypi
mkl 2024.0.0 intel_49656 intel
mkl-dpcpp 2024.0.0 intel_49656 intel
mkl-service 2.4.0 py310h3539a15_41 intel
mkl_fft 1.3.6 py310h1d81ff8_61 intel
mkl_random 1.2.2 py310h5a378b4_81 intel
mkl_umath 0.1.1 py310h2b1685c_91 intel
mpmath 1.3.0 pypi_0 pypi
ncurses 6.4 h59595ed_2 conda-forge
networkx 3.2.1 pypi_0 pypi
numba 0.57.1 py310h0f6aa51_0 intel
numba-dpex 0.21.4 py310hce084a9_0 intel
numpy 1.24.3 py310ha320b8e_5 intel
numpy-base 1.24.3 py310hbac2b65_5 intel
oneccl-bind-pt 2.2.0+cpu pypi_0 pypi
onemkl-sycl-blas 2024.0.0 intel_49656 intel
onemkl-sycl-datafitting 2024.0.0 intel_49656 intel
onemkl-sycl-dft 2024.0.0 intel_49656 intel
onemkl-sycl-lapack 2024.0.0 intel_49656 intel
onemkl-sycl-rng 2024.0.0 intel_49656 intel
onemkl-sycl-sparse 2024.0.0 intel_49656 intel
onemkl-sycl-stats 2024.0.0 intel_49656 intel
onemkl-sycl-vm 2024.0.0 intel_49656 intel
opencl_rt 2024.0.3 intel_49895 intel
openssl 3.1.4 hd590300_0 intel
packaging 23.2 pyhd8ed1ab_0 intel
pillow 10.2.0 pypi_0 pypi
pip 23.3.1 pyhd8ed1ab_0 intel
platformdirs 3.11.0 pyhd8ed1ab_0 intel
pluggy 1.3.0 pyhd8ed1ab_0 intel
pooch 1.8.0 pyhd8ed1ab_0 intel
psutil 5.9.8 pypi_0 pypi
pycparser 2.21 pyhd8ed1ab_0 intel
pyopenssl 23.3.0 pyhd8ed1ab_0 intel
pyparsing 3.1.1 pyhd8ed1ab_0 intel
pysocks 1.7.1 pyha2e5f31_6 intel
python 3.10.13 h4499717_0 intel
python-dateutil 2.8.2 pyhd8ed1ab_0 intel
python_abi 3.10 2_cp310 intel
pytz 2023.3.post1 pyhd8ed1ab_0 intel
pyyaml 6.0.1 py310h2372a71_1 intel
readline 8.2 h8228510_1 conda-forge
requests 2.31.0 pyhd8ed1ab_0 intel
ruamel.yaml 0.17.40 py310h2372a71_0 intel
ruamel.yaml.clib 0.2.7 py310h2372a71_2 intel
scikit-learn 1.3.1 py310h1fdf081_1 intel
scikit-learn-intelex 2024.1.0 py310_intel_299 intel
scipy 1.10.1 py310h6681978_8 intel
setuptools 68.2.2 pyhd8ed1ab_0 intel
six 1.16.0 pyh6c4a22f_0 intel
smp 0.1.5 py310h8271ca5_21 intel
snappy 1.1.10 h9fff704_0 intel
spirv-tools 2023.2 h00ab1b0_4 intel
sympy 1.12 pypi_0 pypi
tbb 2021.11.0 intel_49513 intel
tbb4py 2021.11.0 py310_intel_49513 intel
threadpoolctl 3.2.0 pyha21a80b_0 intel
tk 8.6.13 noxft_h4845f30_101 conda-forge
toolz 0.12.0 pyhd8ed1ab_0 intel
torch 2.2.1+cpu pypi_0 pypi
torchaudio 2.2.1+cpu pypi_0 pypi
torchvision 0.17.1+cpu pypi_0 pypi
typing-extensions 4.8.0 hd8ed1ab_0 intel
typing_extensions 4.8.0 pyha770c72_0 intel
tzdata 2023c h71feb2d_0 intel
urllib3 2.0.7 pyhd8ed1ab_0 intel
wheel 0.41.3 pyhd8ed1ab_0 intel
xgboost 1.7.3 0_gade498py310_25 intel
xz 5.2.6 h166bdaf_0 conda-forge
yaml 0.2.5 h7f98852_2 intel
zlib 1.2.13 hd590300_5 intel
zstandard 0.22.0 py310h1275a96_0 intel
zstd 1.5.5 hfc55251_0 intel
Let's try running a sample program using the virtual environment we created.
$ python ./Intel_Extension_For_PyTorch_Hello_World.py
onednn_verbose,info,oneDNN v3.3.4 (commit f240e12a29cff8ec1f37e5907971aa4315d30f1e)
onednn_verbose,info,cpu,runtime:OpenMP,nthr:24
onednn_verbose,info,cpu,isa:Intel AVX-512 with AVX512BW, AVX512VL, and AVX512DQ extensions
onednn_verbose,info,gpu,runtime:none
onednn_verbose,info,graph,backend,0:compiler_backend
onednn_verbose,info,graph,backend,1:dnnl_backend
onednn_verbose,primitive,info,template:operation,engine,primitive,implementation,prop_kind,memory_descriptors,attributes,auxiliary,problem_desc,exec_time
onednn_verbose,graph,info,template:operation,engine,partition_id,partition_kind,op_names,data_formats,logical_tensors,fpmath_mode,backend,exec_time
onednn_verbose,primitive,exec,cpu,reorder,jit:uni,undef,src_f32::blocked:acdb::f0 dst_f32:p:blocked:Acdb16a::f0,attr-scratchpad:user ,,6x3x3x3,0.00488281
onednn_verbose,primitive,exec,cpu,convolution,brgconv:avx512_core,forward_training,src_f32::blocked:acdb::f0 wei_f32:ap:blocked:Acdb16a::f0 bia_f32:a:blocked:a::f0 dst_f32::blocked:acdb::f0,attr-scratchpad:user ,alg:convolution_direct,mb50_ic3oc6_ih112oh110kh3sh1dh0ph0_iw112ow110kw3sw1dw0pw0,1.32788
onednn_verbose,primitive,exec,cpu,convolution,jit:avx512_core,backward_weights,src_f32::blocked:acdb::f0 wei_f32:ap:blocked:Acdb16a::f0 bia_f32:a:blocked:a::f0 dst_f32::blocked:acdb::f0,attr-scratchpad:user ,alg:convolution_direct,mb50_ic3oc6_ih112oh110kh3sh1dh0ph0_iw112ow110kw3sw1dw0pw0,0.819092
onednn_verbose,primitive,exec,cpu,convolution,brgconv:avx512_core,forward_training,src_f32::blocked:acdb::f0 wei_f32:ap:blocked:Acdb16a::f0 bia_f32:a:blocked:a::f0 dst_f32::blocked:acdb::f0,attr-scratchpad:user ,alg:convolution_direct,mb50_ic3oc6_ih112oh110kh3sh1dh0ph0_iw112ow110kw3sw1dw0pw0,0.50708
onednn_verbose,primitive,exec,cpu,convolution,jit:avx512_core,backward_weights,src_f32::blocked:acdb::f0 wei_f32:ap:blocked:Acdb16a::f0 bia_f32:a:blocked:a::f0 dst_f32::blocked:acdb::f0,attr-scratchpad:user ,alg:convolution_direct,mb50_ic3oc6_ih112oh110kh3sh1dh0ph0_iw112ow110kw3sw1dw0pw0,0.781006
onednn_verbose,primitive,exec,cpu,convolution,brgconv:avx512_core,forward_training,src_f32::blocked:acdb::f0 wei_f32:ap:blocked:Acdb16a::f0 bia_f32:a:blocked:a::f0 dst_f32::blocked:acdb::f0,attr-scratchpad:user ,alg:convolution_direct,mb10_ic3oc6_ih112oh110kh3sh1dh0ph0_iw112ow110kw3sw1dw0pw0,0.155029
onednn_verbose,primitive,exec,cpu,convolution,brgconv:avx512_core,forward_training,src_f32::blocked:acdb::f0 wei_f32:ap:blocked:Acdb16a::f0 bia_f32:a:blocked:a::f0 dst_f32::blocked:acdb::f0,attr-scratchpad:user ,alg:convolution_direct,mb10_ic3oc6_ih112oh110kh3sh1dh
0ph0_iw112ow110kw3sw1dw0pw0,0.141846
[CODE_SAMPLE_COMPLETED_SUCCESFULLY]
This is a basic operation, but it demonstrates that the sample program is recognizing and operating in a multi-threaded environment and with AVX512 using Intel Extension for Pytorch -> Intel OneDNN.
Next, let's input a slightly more CPU-intensive example as a batch job to a compute node for computation.
To switch environments and run the example using conda, load the conda.sh and activate the PyTorch environment you created.
#!/bin/bash
#$ -cwd
#$ -V
#$ -l short
#$ -l d_rt=00:50:00
#$ -l s_rt=00:50:00
#$ -l s_vmem=50G
#$ -l mem_req=50G
#$ -N short
#$ -S /bin/bash
export OMP_NUM_THREADS=4
source /home/ymuna/miniconda3/etc/profile.d/conda.sh
conda activate pytorch
python ./IntelPytorch_Quantization.py
The script limits the number of CPU cores used by setting OMP_NUM_THREADS because Python modules utilizing MKL operate in a multi-threaded manner.
Files already downloaded and verified
Inference with FP32
Inference ...
Inference Time Avg: 0.38231764554977415
Inference with Static INT8
Inference ...
Inference Time Avg: 0.07147334575653076
Inference with Dynamic INT8
Inference ...
Inference Time Avg: 0.07396594524383544
Summary
FP32 inference time: 0.382
INT8 static quantization inference time: 0.071
INT8 dynamic quantization inference time: 0.074
Static INT8 5.35X faster than FP32
Dynamic INT8 5.17X faster than FP32
[CODE_SAMPLE_COMPLETED_SUCCESFULLY]
As shown above, it is evident that the compute node recognizes and operates the PyTorch environment.