TensorFlow 2.8.0 No AVX, No GPU, Python 3.7, 3.8, 3.9, 3.10, Ubuntu 18.04, multiple Archs
Novaal opened this issue · 14 comments
Built on Ubuntu 18.04. Builds are not tested and provided as is.
Example configuration for Python 3.7 and westmere:
PYTHON_VERSION=python3.7
PYTHON_BIN_PATH=$(which $PYTHON_VERSION) \
PYTHON_LIB_PATH=$($PYTHON_VERSION -c "import pip; print(pip.__path__[0].rstrip('/pip'))") \
TF_NEED_CUDA=0 \
TF_NEED_ROCM=0 \
TF_DOWNLOAD_CLANG=0 \
CC_OPT_FLAGS="-march=westmere -Wno-sign-compare" \
TF_SET_ANDROID_WORKSPACE=0 \
TF_ENABLE_XLA=1 \
TF_NEED_OPENCL_SYCL=0 \
TF_NEED_MPI=0 \
./configureThis produces the following .tf_configure.bazelrc for Python 3.7.
build --action_env PYTHON_BIN_PATH="/usr/bin/python3.7"
build --action_env PYTHON_LIB_PATH="/usr/local/lib/python3.7/dist-packages"
build --python_path="/usr/bin/python3.7"
build --define=with_xla_support=true
build:opt --copt=-march=westmere
build:opt --host_copt=-march=westmere
build:opt --copt=-Wno-sign-compare
build:opt --host_copt=-Wno-sign-compare
test --flaky_test_attempts=3
test --test_size_filters=small,medium
test:v1 --test_tag_filters=-benchmark-test,-no_oss,-gpu,-oss_serial
test:v1 --build_tag_filters=-benchmark-test,-no_oss,-gpu
test:v2 --test_tag_filters=-benchmark-test,-no_oss,-gpu,-oss_serial,-v1only
test:v2 --build_tag_filters=-benchmark-test,-no_oss,-gpu,-v1only
| GCC Compiler Option (-march) |
Python 3.7 | Python 3.8 | Python 3.9 | Python 3.10 | All Builds |
|---|---|---|---|---|---|
| barcelona | Download | Download | Download | Download | Link |
| btver1 | Download | Download | Download | Download | Link |
| core2 | Download | Download | Download | Download | Link |
| nehalem | Download | Download | Download | Download | Link |
| westmere | Download | Download | Download | Download | Link |
Install with:
pip install --ignore-installed --upgrade tensorflow-2.8.0-cp37-cp37m-linux_x86_64.whlPlease let me know here if it works for you.
Hi Yaroslav,
The pip3 install https://tf.novaal.de/core2/tensorflow-2.8.0-cp39-cp39-linux_x86_64.whl
work perfectly on Debian os Sid with cpu I7 K875 .
Just take care to add to the PATH set like : HOME/.local/bin
Thank a lot ;)
Thank you! tensorflow-2.8.0-cp310-cp310-linux_x86_64.whl works like a charm on debian:sid with E5649, that surprisingly doesn't have AVX :(
Works great on M1 Mac running X86_64 QEMU guest with Debian 11.3.0
Command: "pip install --ignore-installed --upgrade tensorflow-2.8.0-cp39-cp39-linux_x86_64.whl"
Defaulting to user installation because normal site-packages is not writeable
"ERROR: tensorflow-2.8.0-cp39-cp39-linux_x86_64.whl is not a supported wheel on this platform."
How can i solve this?
Hi @MathewsJosh
i tried all Python 3.9 wheels in a fresh docker container and could not verify any installation problems.
Please make sure you are using the correct python version and that you are not installing via pip install https://... . If you are installing multiple times e.g. when building a Docker image the http-call might fail, because the CDN is flagging you as a bot and blocking your requests. So please download the wheel once and install it from the local file system.
Command: "pip install --ignore-installed --upgrade tensorflow-2.8.0-cp39-cp39-linux_x86_64.whl"
Defaulting to user installation because normal site-packages is not writeable "ERROR: tensorflow-2.8.0-cp39-cp39-linux_x86_64.whl is not a supported wheel on this platform."
How can i solve this?
You may try one of the other ones, depending in your CPU (I guess). There are only 4 versions for each python version.
Hello Yaroslav,
I just try to use 3.10 python whl to update my install ... it was ok with python 3.9 , and I get this output :
numba 0.56.0 requires numpy<1.23,>=1.18, but you have numpy 1.23.1 which is incompatible
when I use : pip3 install --ignore-installed --upgrade tensorflow-2.8.0-cp310-cp310-linux_x86_64.whl
any ideas?
regards
MacOS Tensorflow compiled without AVX, Python 3.8-3.11, MPS GPU/CPU, - https://github.com/alessandro893/tensorflow-macos-noavx
thank u buddy. can u please write a few words about the selection of the GCC Compiler Option? what should i choose?
Works for me on
OS: Linux Mint 21 (based on Ubuntu 22.04 LTS)
CPU: Intel i5 M 430 (march: westmere)
Python: 3.8, 3.10
Hi Guys it is the possibility to have TensorFlow 2.8 with GPU support and No AVX. Where I can find the WHL file? Thanks in advance.
Thank you very much, working perfectly on my Linux-based 2009 Mac Pro (dual Xeon X5690 Westmere)
Hello, do anyone know why the tensorflow is stuck even if it just simply import(import tensorflow on python shell and nothing happen, cant ctrl+c, it looks like loading but nothing happen) ?
i install: tensorflow-2.8.0-cp38-cp38m-linux_x86_64.whl
heres quick spec of my system:
OS: Linux Mint 20.1 x86_64
Host: 80MH Lenovo ideapad 100-14IBY
Kernel: 5.4.0-58-generic
CPU: Intel Celeron N2840 (2) @ 2.582GHz
GPU: Intel Atom Processor Z36xxx/Z37xxx Series Graphics & Di
Memory: 1227MiB / 1869MiB
Thanks
core2 variant seems to have installed on Intel(R) Pentium(R) CPU U5400
Though you may have to downgrade protobuf to 3.20 via pip (not conda or mamba !!!)