Frequently Asked Questions¶
How to check whether GPU drivers are installed successfully?
Run import tensorflow and it will show which platform you are running on: Intel Level-Zero(default) or Intel OpenCL.
And the high level API of TensorFlow tf.config.experimental.list_physical_devices() will tell you the device types that are registered to TensorFlow core.
$ python
>>> import tensorflow as tf
2021-07-01 06:40:55.510076: I itex/core/devices/gpu/dpcpp_runtime.cc:116] Selected platform: Intel(R) Level-Zero.
>>> tf.config.experimental.list_physical_devices()
[PhysicalDevice(name='/physical_device:CPU:0', device_type='CPU'), PhysicalDevice(name='/physical_device:XPU:0', device_type='XPU')]
How to know the configurations and rate of utilization of local GPU devices?
System Monitoring Utility tool can be used to show the capability (clock frequency, EU count, amount of device memory, and so on) of your devices and usage of each sub-module (device memory, GPU engines, and so on).
What’s the relationship of
TensorFlow*,Intel® Optimization of TensorFlow*andIntel® Extension for TensorFlow*?
Intel® Optimization of TensorFlow* is designed to optimize for Intel CPU. It could replace stock TensorFlow* (Google TensorFlow*) for Intel CPU. All Intel optimizations are available in both Intel® Optimization for TensorFlow* and stock TensorFlow* (since 2.9) for Intel CPU. That means you only need to install one of them. DO NOT install them in same time, the impact is unknown.
Intel® Extension for TensorFlow* is an extension of stock TensorFlow* and help extend to accelerate on Intel CPU or support Intel GPU.
Intel® Extension for TensorFlow* only co-works with stock TensorFlow*. Please DO NOT install Intel® Extension for TensorFlow* with Intel® Optimization for TensorFlow*.
Starting in Q1 2024, Intel® Optimization for TensorFlow* will be discontinued. Intel optimization will be available only through Intel® Extension for TensorFlow* co-works with the stock TensorFlow*. Intel will continue to upstream advanced optimization to stock TensorFlow* in the future.
Currently, Intel® Extension for TensorFlow* has two releases: CPU & GPU.
For Intel CPU, Intel® Extension for TensorFlow* for CPU + stock TensorFlow* could replace Intel® Optimization of TensorFlow*. Install command: pip install --upgrade intel-extension-for-tensorflow[cpu].
For Intel GPU, Intel® Extension for TensorFlow* for GPU + stock TensorFlow* is only way to make TensorFlow* support Intel GPU. Install command: pip install --upgrade intel-extension-for-tensorflow[gpu].
Troubleshooting¶
This section shows common problems and solutions for compilation and runtime issues you may encounter.
Build from source¶
| Error | Solution | Comments |
|---|---|---|
| external/onednn/src/sycl/level_zero_utils.cpp:33:10: fatal error: 'level_zero/ze_api.h' file not found #include ^~~~~~~~~~~~~~~~~~~~~ |
install level-zero-dev lib |
level-zero-dev lib is needed when building from source |
Runtime¶
| Error | Solution | Comments |
|---|---|---|
| ModuleNotFoundError: No module named 'tensorflow' | install TensorFlow | Intel® Extension for TensorFlow* depends on TensorFlow |
| tensorflow.python.framework.errors_impl.NotFoundError: libmkl_sycl.so.2: cannot open shared object file: No such file or directory | source /opt/intel/oneapi/setvars.sh |
set env vars of oneAPI Base Toolkit |
| version GLIBCXX_3.4.30' not found | conda install -c conda-forge gxx_linux-64==12.1.0 |
install higher version glibcxx |