This is a list of optimized WHL Tensorflow builds for cloud computing use with Intel® Xeon® Scalable CPUs.
Date of build | Tensorflow | Python | Platform | Architecture | Links |
---|---|---|---|---|---|
2019-07-29 | 1.14.0 | 3.6 | Linux | skylake | download |
In your Python environment, run
pip install --upgrade /path/to/file.whl
A non exhaustive list of VMs that work best for CPU Tensorflow inference.
Intel® Xeon® Platinum 8124M
- c5.4xlarge (16 vCPUs / 32GB)
- c5.9xlarge (36 vCPUs / 72GB)
Intel® Xeon® Platinum 8275L
- c5.12xlarge (48 vCPUs / 96GB)
Intel® Xeon® Platinum 8168
- F16s_v2 (16 vCPUs / 32GB)
- F32s_v2 (32 vCPUs / 64GB)
Undisclosed Intel® Xeon® Scalable CPU
- c2-standard-30 (30 vCPUs, 120GB)