tf-builds

This is a list of optimized WHL Tensorflow builds for cloud computing use with Intel® Xeon® Scalable CPUs.

Download links

Date of build	Tensorflow	Python	Platform	Architecture	Links
2019-07-29	1.14.0	3.6	Linux	skylake	download

Installation

In your Python environment, run pip install --upgrade /path/to/file.whl

Best VMs

A non exhaustive list of VMs that work best for CPU Tensorflow inference.

AWS

Intel® Xeon® Platinum 8124M

• c5.4xlarge (16 vCPUs / 32GB)
• c5.9xlarge (36 vCPUs / 72GB)

Intel® Xeon® Platinum 8275L

• c5.12xlarge (48 vCPUs / 96GB)

Azure

Intel® Xeon® Platinum 8168

• F16s_v2 (16 vCPUs / 32GB)
• F32s_v2 (32 vCPUs / 64GB)

GCP

Undisclosed Intel® Xeon® Scalable CPU

• c2-standard-30 (30 vCPUs, 120GB)