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.

Built images are available via docker pull awsdeepracercommunity/deepracer-robomaker:<tag>.

Available tags are:

Version 5.1

Version 5.1 is the current development version for 2023, and it is receiving both new tracks as well as functionality updates as they are released by AWS.

New in 2023 is that the OpenGL images are no longer needed; the required drivers are included in the CPU and GPU images.

Version	Comment	AVX	AVX2	GPU
5.1.0	Initial release	5.1.0-cpu-avx	5.1.0-cpu-avx2	5.1.0-gpu

Version 5

Version 5 is the release version for 2022. It will only receive track fixes.

Version	Comment	AVX	AVX2	GPU
5.0.0	Initial release	5.0.0-cpu-avx 5.0.0-cpu-gl-avx	5.0.0-cpu-avx2 5.0.0-cpu-gl-avx2	5.0.0-gpu 5.0.0-gpu-gl
5.0.1	April Track	5.0.1-cpu-avx 5.0.1-cpu-gl-avx	5.0.1-cpu-avx2 5.0.1-cpu-gl-avx2	5.0.1-gpu 5.0.1-gpu-gl
5.0.2	May Track	5.0.2-cpu-avx 5.0.2-cpu-gl-avx	5.0.2-cpu-avx2 5.0.2-cpu-gl-avx2	5.0.2-gpu 5.0.2-gpu-gl
5.0.3	June Track	5.0.3-cpu-avx 5.0.3-cpu-gl-avx	5.0.3-cpu-avx2 5.0.3-cpu-gl-avx2	5.0.3-gpu 5.0.3-gpu-gl
5.0.4	July Track	5.0.4-cpu-avx 5.0.4-cpu-gl-avx	5.0.4-cpu-avx2 5.0.4-cpu-gl-avx2	5.0.4-gpu 5.0.4-gpu-gl
5.0.5	August Track	5.0.5-cpu-avx 5.0.5-cpu-gl-avx	5.0.5-cpu-avx2 5.0.5-cpu-gl-avx2	5.0.5-gpu 5.0.5-gpu-gl
5.0.6	September Track	5.0.6-cpu-avx 5.0.6-cpu-gl-avx	5.0.6-cpu-avx2 5.0.6-cpu-gl-avx2	5.0.6-gpu 5.0.6-gpu-gl
5.0.7	October & Reinvent 2022 Tracks	5.0.7-cpu-avx 5.0.7-cpu-gl-avx	5.0.7-cpu-avx2 5.0.7-cpu-gl-avx2	5.0.7-gpu 5.0.7-gpu-gl
5.0.8	Updates to Reinvent 2022 Tracks	5.0.8-cpu-avx 5.0.8-cpu-gl-avx	5.0.8-cpu-avx2 5.0.8-cpu-gl-avx2	5.0.8-gpu 5.0.8-gpu-gl
5.0.9	Minor updates and fixes	5.0.9-cpu-avx 5.0.9-cpu-gl-avx	5.0.9-cpu-avx2 5.0.9-cpu-gl-avx2	5.0.9-gpu 5.0.9-gpu-gl
5.0.10 (*)	Tracks expanded with CW and CCW tracks	5.0.10-cpu-avx 5.0.10-cpu-gl-avx	5.0.10-cpu-avx2 5.0.10-cpu-gl-avx2	5.0.10-gpu 5.0.10-gpu-gl

(*) 5.0.10 received a patch on 15-MAR-2023 due to a correction of the arctic_pro_ccw and arctic_pro_cw tracks.

Building the image

A build script is available as build.sh. By default all three images will be built (cpu-avx cpu-avx2 and gpu for version 5.1). Use the -a switch to limit number of images.

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