Containerizing Hardware Accelerated Applications
3 likes1,841 views
The document evaluates the performance impact of using containers for hardware accelerated media applications, demonstrating that the use of Docker results in negligible performance differences during transcoding tasks. It highlights the benefits of containerization in reducing conflicts with the host system while noting the limitations in portability due to hardware dependencies. The testing was conducted by Chelsea Mafrica at Intel, utilizing specific system configurations and emphasizing the optimization for Intel microprocessors.
Containerizing Hardware Accelerated Applications
- 1. Containerizing Hardware Accelerated Applications Chelsea Mafrica Data Center Systems Engineer Intel Corporation
- 2. Motivation Evaluate performance impact of containers on a media stack that uses hardware acceleration
- 3. Agenda ● Hardware accelerators, applications, and media ● Media stack ● When & how to use containers ● Experiment & results ● Portability
- 4. Hardware accelerators, applications, and mediaA hardware accelerator is a processor or fixed function specialized to perform specific tasks (excluding a general purpose CPU) Examples: GPUs, FPGAs, ASICs Applications that typically benefit from hardware acceleration are ones that can be parallelized Examples: AI, machine learning, HPC, media Media refers to video processing Examples: Video compression and decompression (encode and decode), filters
- 5. KERNEL SERVER DRIVE R GPU LIBS APP APP APP USER SPACE Media stack Transcode applications Intel® Media Server Studio Intel® Quick Sync Video Intel® Iris® Pro Graphics
- 6. Media stack with Docker Transcode applications Intel® Media Server Studio Intel® Quick Sync Video Intel® Iris® Pro Graphics Docker KERNEL SERVER DRIVE R GPU LIBS APP USER SPACE CONTAINE R ENGINE CONTAINER
- 7. Software- only app Applications Libraries & dependencies DockerKERNEL SERVER LIBS APP USER SPACE CONTAINE R ENGINE CONTAINER
- 8. KERNEL SERVER DRIVE R GPU USER SPACE CONTAINE R ENGINE CONTAINER Media stack with Docker LIBS APP APP APP Transcode applications Intel® Media Server Studio Intel® Quick Sync Video Intel® Iris® Pro Graphics Docker
- 9. KERNEL SERVER DRIVE R GPU USER SPACE Media stack with Docker LIBS APP CONTAINER APP LIBS CONTAINER CONTAINE R ENGINE Transcode applications Intel® Media Server Studio Intel® Quick Sync Video Intel® Iris® Pro Graphics Docker
- 10. • Kernel module installation • Custom kernel build $ ls /dev/dri card0 card1 controlD64 controlD65 renderD128 Host requirements
- 11. FROM centos:7.2.1511 MAINTAINER Chelsea Mafrica <chelsea.e.mafrica@intel.com> COPY intel-linux-media_generic_16.5.1-59511_64bit.tar.gz sample_multi_transcode /root/ RUN yum -y -t install mesa-dri-drivers && yum clean all && useradd user && usermod -a -G wheel user && usermod -a -G video user && find /usr -name "libdrm*" | xargs rm -rf && find /usr -name "libva*" | xargs rm -rf && cd root && tar -xvf intel-linux-media_generic_16.5.1-59511_64bit.tar.gz && cp -r etc/* /etc && cp -r lib/* /lib && cp -r opt/* /opt && cp -r usr/* /usr && cp sample_multi_transcode /home/user && chown user:user /home/user/sample_multi_transcode && rm -rf * WORKDIR /home/user Dockerfile
- 12. docker build -t mss:centos.transcode . docker run --device=/dev/dri/renderD128 --volume=/home/user/volume/mss_content:/home/user/content -i -d mss:centos.transcode bash docker exec CONTAINER_ID su - user –c "./sample_multi_transcode -i::h264 content/video_input.264 -o::h264 content/video_output.264" Building and running the container
- 13. Experiment Test the number of transcodes that can run on a system before the average performance of a transcode drops below 30 frames per second APPNAPP1 APPNAPP1 APPNAPP1 CONTAINER HOSTHOST HOST CONTAINER1 CONTAINERN baseline single container case multiple container case
- 14. Observations ● Variability in container startup time as the system reaches capacity ● Running in detached mode, negligible change in performance
- 15. framespersecond Legal Disclaimer: Software and workloads used in performance tests may have been optimized for performance only on Intel microprocessors. Performance tests, such as SYSmark and MobileMark, are measured using specific computer systems, components, software, operations and functions. Any change to any of those factors may cause the results to vary. You should consult other information and performance tests to assist you in fully evaluating your contemplated purchases, including the performance of that product when combined with other products. For more complete information visit http://www.intel.com/performance. *Other names and brands may be claimed as the property of others. See backup for configuration details. Transcode Performance apps real-time (30 fps)
- 16. Observations ● Variability in container startup time as the system reaches capacity ● Running in detached mode, negligible change in performance ● Portability is limited due to driver and hardware requirements
- 17. Media stack with Docker KERNEL SERVER DRIVE R GPU LIBS APP USER SPACE CONTAINE R ENGINE CONTAINER Transcode applications Intel® Media Server Studio Intel® Quick Sync Video Intel® Iris® Pro Graphics Docker
- 18. Media stack with Docker KERNEL SERVER DRIVE R GPU LIBS APP USER SPACE CONTAINE R ENGINE CONTAINER Transcode applications Intel® Media Server Studio Intel® Quick Sync Video Intel® Iris® Pro Graphics Docker
- 19. Media stack with Docker KERNEL SERVER DRIVE R GPU LIBS APP USER SPACE CONTAINE R ENGINE CONTAINER Transcode applications Intel® Media Server Studio Intel® Quick Sync Video Intel® Iris® Pro Graphics Docker
- 20. Media stack with Docker KERNEL SERVER DRIVE R GPU LIBS APP USER SPACE CONTAINE R ENGINE CONTAINER Transcode applications Intel® Media Server Studio Intel® Quick Sync Video Intel® Iris® Pro Graphics Docker
- 21. Summary ● Running accelerated apps in containers uses existing Docker capabilities ● The use of containers resulted in negligible performance difference for transcode apps in capacity test ● Containers are helpful for reducing conflicts with the host, but this isn’t specific to hardware accelerators ● Dependency on hardware and custom kernels limits portability of container, but the app will have better performance because of the hardware
- 22. Links & current work Intel® Media Server Studio: http://intel.ly/MediaServerStudio Intel ® MediaSDK http://github.com/Intel-Media-SDK Intel® OTC: http://github.com/vmmqa/dockerGpuStack twitter: mafrica_ chelsea.e.mafrica at intel dot com
- 23. Legal Information Testing by Chelsea Mafrica, January 2017 – June 2017 System Configuration: BASELINE: Intel® Xeon® CPU E3-1585L v5, 3.5GHz, 4 cores, turbo and HT on, BIOS AMI 1.0, 32GB total memory, 2 slots / 16GB / 2133MHz / DDR4 DIMM, 480GB total storage / 2 240GB SSDs (2.5”), Intel® I350 Gigabit Network Connection, CentOS Linux* 7.2.1511 kernel 3.10.0-327.13.1.x86_64, Media Server Studio 2017 R1 NEW: Baseline configuration, Docker* 1.12.3 Disclaimer Software and workloads used in performance tests may have been optimized for performance only on Intel microprocessors. Performance tests, such as SYSmark and MobileMark, are measured using specific computer systems, components, software, operations and functions. Any change to any of those factors may cause the results to vary. You should consult other information and performance tests to assist you in fully evaluating your contemplated purchases, including the performance of that product when combined with other products. For more complete information visit http://www.intel.com/performance. *Other names and brands may be claimed as the property of others