Ce diaporama a bien été signalé.
Le téléchargement de votre SlideShare est en cours. ×

GPU power consumption and performance trends

Publicité
Publicité
Publicité
Publicité
Publicité
Publicité
Publicité
Publicité
Publicité
Publicité
Publicité
Publicité
Prochain SlideShare
Graphic Processing Unit
Graphic Processing Unit
Chargement dans…3
×

Consultez-les par la suite

1 sur 25 Publicité

GPU power consumption and performance trends

A brief technical overview about GPU power consumption and performance, with references to the latest architecture developed by Nvidia: Maxwell and Tegra X1.

Co-Author: Pietro Piscione (https://www.linkedin.com/pub/pietro-piscione/84/b37/926)

A brief technical overview about GPU power consumption and performance, with references to the latest architecture developed by Nvidia: Maxwell and Tegra X1.

Co-Author: Pietro Piscione (https://www.linkedin.com/pub/pietro-piscione/84/b37/926)

Publicité
Publicité

Plus De Contenu Connexe

Diaporamas pour vous (20)

Similaire à GPU power consumption and performance trends (20)

Publicité

Plus récents (20)

GPU power consumption and performance trends

  1. 1. Power Consumption and Performance trends on GPUs Computer Architecture Authors: Piscione Pietro Villardita Alessio A.Y. 2014/2015 Degree: Computer Engineering
  2. 2. Nowadays focus ...and CPU CUDA Performance comparison Power consumption and performance Architecture overview Purposes and application Mobile devices GPU Benchmark
  3. 3. Why GPU ? ● Multimedia and general purpose applications and recently, also ● High Performance Computing applications, "autonomous machines" and automotive CAD and 3D apps Office and PDF readers Multimedia Audio/Video Browsers
  4. 4. Power Consumption Overview Domain GPU power consumption Desktop and Workstation (no mobility) 300W (50%) Notebook (mobility) 50W (71%) And the smartphones ?
  5. 5. Power Consumption Overview: smartphone Gaming 3D Rendering Power(mW)
  6. 6. GPU power consumption on mobile How much power does a mobile application require in order to execute? Why is the GPU so “energy-hungry”? The answer is in the GPU Architecture.
  7. 7. Power Consumption vs Performance: SPECInt benchmark
  8. 8. Nowadays focus ...and CPU CUDA Performance comparison Power consumption and performance Architecture overview Purposes and application Mobile devices GPU Benchmark
  9. 9. GPU Architecture - Overview ● CUDA cores (up to thousands) ● Streaming Multiprocessor (SMM) ● Global memory ● Giga Thread (Scheduler) ● Cache (L1 and L2)
  10. 10. GPU Architecture - (SMM) ● Cuda Cores ● SFU ● LDST units ● Dispatcher unit ● Warp scheduler
  11. 11. What is CUDA ? (Computer Unified Device Architecture) Multithreading on multiple simple cores It implies an higher throughtput than CPU
  12. 12. GPU and CPU communication
  13. 13. CUDA study case: quick sort
  14. 14. GPU vs CPU GPU ● hundreds of simpler cores CPU ● few very complex cores ● thousand of concurrent hardware threads ● single-thread performance optimization ● maximize floating-point throughput ● transistor space dedicated to complex ILP ● most die surface for integer and fp units ● few die surface for integer and fp units
  15. 15. Performance Comparison: computation power
  16. 16. Performance Comparison: throughput
  17. 17. Nowadays focus ...and CPU CUDA Performance comparison Power consumption and performance Architecture overview Purposes and application Mobile devices GPU Benchmark
  18. 18. Benchmark study case: Tegra X1 Main features: ● GPU Maxwell 256 core ● 20nm tecnology ● CPU Octacore One Teraflop of computation power.
  19. 19. GFXBenchmark description GFXBenchmark metrics: ● Graphics performance ● Long-term performance stability ● Render quality ● Power consumption Composed by: ● Manhattan test ● T-Rex test ● Battery and stability test
  20. 20. GFXBenchmark description Manhattan test: ● GPU-intensive OpenGL ES 3 test ● High detailed scenes (gaming field) ● Low effects level
  21. 21. GFXBenchmark description T-Rex test: ● GPU-intensive OpenGL ES 2 test ● Low detailed scenes ● High effects level
  22. 22. GFXBenchmark description Battery and Stability test: Battery life Performance stability How it’s made ? Logging (FPS)
  23. 23. Tegra X1 on GFXBenchmark
  24. 24. Tegra X1 vs K1: power consumption Architectural solutions to improve energy efficiency: ● big.LITTLE processing design ● Using of cluster migration ● Cache coherence solution DRAM Energy efficiency
  25. 25. Conclusions Power consumption reduction: current and future trends involve both Hardware AND Software design ● Reduce data movement ● Lots of local processing in parallel ● Efficient caching and memory usage ● Where data lives ● Where computation happens, how it is scheduled Energy efficiency must now be key metric.

×