zhenghongpeng / deepracer-simapp

Extract from AWS DeepRacer Robomaker Bundle

AWS DeepRacer Robomaker Image

This repository contains the extracts from the AWS DeepRacer Robomaker simapp; including

• the environment and tracks
• the robotics agent code ('markov')
• Dockerfile

Available pre-built versions

In most cases it will be sufficient to use one of our pre-built images which use optimized Tensorflow binaries. Depending on how modern processor you have available you can pick which instruction sets to use.

• AVX is available on most Intel Core processors from the second generation on - running DeepRacer on an older processor really does not make sense; hence we do not have pre-built images without this instruction set.
• AVX2 is available on more modern Intel Core processors; starting from the Haswell architecture (4th generation), and is probably the image most will use.
• AVX-512 was available on newer Xeon and Core i7-X processors. Due to convergence issues it has been removed.
• GPU uses Tensorflow-GPU instead of CPU. It is not compiled with special instruction sets for the CPU.

Available tags are:

Version 3

Version 3 is the current version, and it is receiving both new tracks as well as functionality updates as they are released by AWS.

Version	Comment	AVX	AVX2	GPU
3.0.1	Upgrade to Simapp 3	3.0.1-cpu-avx 3.0.1-cpu-gl-avx	3.0.1-cpu-avx2 3.0.1-cpu-gl-avx	3.0.1-gpu 3.0.1-gpu-gl
3.0.2	September Track	3.0.2-cpu-avx 3.0.2-cpu-gl-avx	3.0.2-cpu-avx2 3.0.2-cpu-gl-avx	3.0.2-gpu 3.0.2-gpu-gl

Version 2

Version 2 is in maintenance state, and will only receive bug-fixes for breaking issues as well as new tracks.

Version	Comment	AVX	AVX2	GPU
Release	Version 2.0.11	cpu, cpu-avx, cpu-gl-avx	cpu-avx2, cpu-gl-avx2	gpu, gpu-gl
2.0.5	Update 07-MAY	2.0.5-cpu-avx	2.0.5-cpu-avx2	2.0.5-gpu
2.0.6	Update 20-MAY and 22-May	2.0.6-cpu-avx	2.0.6-cpu-avx2	2.0.6-gpu
2.0.7	June track + fixes	2.0.7-cpu-avx	2.0.7-cpu-avx2	2.0.7-gpu
2.0.8	Update 8-JUN + fixes	2.0.8-cpu-avx 2.0.8-cpu-gl-avx	2.0.8-cpu-avx2 2.0.8-cpu-gl-avx	2.0.8-gpu 2.0.8-gpu-gl
2.0.9	July Track	2.0.9-cpu-avx 2.0.9-cpu-gl-avx	2.0.9-cpu-avx2 2.0.9-cpu-gl-avx	2.0.9-gpu 2.0.9-gpu-gl
2.0.10	Training improvements	2.0.10-cpu-avx 2.0.10-cpu-gl-avx	2.0.10-cpu-avx2 2.0.10-cpu-gl-avx	2.0.10-gpu 2.0.10-gpu-gl
2.0.11	August Track	2.0.11-cpu-avx 2.0.11-cpu-gl-avx	2.0.11-cpu-avx2 2.0.11-cpu-gl-avx	2.0.11-gpu 2.0.11-gpu-gl
2.0.12	September Track	2.0.12-cpu-avx 2.0.12-cpu-gl-avx	2.0.12-cpu-avx2 2.0.12-cpu-gl-avx	2.0.12-gpu 2.0.12-gpu-gl

The built image is available via docker pull awsdeepracercommunity/deepracer-robomaker:cpu. For other images replace cpu with the tag of your choice from the above table.

Building the image

To build the CPU image run docker build -t awsdeepracercommunity/deepracer-robomaker:cpu -f docker/Dockerfile.cpu . --build-arg TENSORFLOW_VER='tensorflow==1.11.0

To build the GPU image run docker build -t awsdeepracercommunity/deepracer-robomaker:gpu -f docker/Dockerfile.gpu .

To add a custom tensorflow library run docker build -t awsdeepracercommunity/deepracer-robomaker:cpu -f docker/Dockerfile.cpu . --build-arg TENSORFLOW_VER=<http-url-to-file>

Tensorflow build

Tensorflow has been built with the following parameters:

• cpu-avx with -march=x86-64 -maes -mavx -mmmx -mpclmul -mpopcnt -msse -msse2 -msse3 -msse4.1 -msse4.2 -mssse3
• cpu-avx2 with -march=x86-64 -maes -mavx -mavx2 -mmmx -mpclmul -mpopcnt -msse -msse2 -msse3 -msse4.1 -msse4.2 -mssse3 -mf16c -mfma