AI APPLICATIONS ON POWER9 SYSTEM
summary
Claims that POWER9 with NVIDIA Tesla V100 GPUs runs enterprise ML/DL workloads that would otherwise need hundreds of CPUs, citing a single GPU being 100 CPUs equivalent for some workloads. Describes using Python stack (NumPy, Pandas, scikit‑learn, PyTorch) and Jupyter, training n‑layer neural nets in hours, and using workload optimized frequency (WOF) to boost throughput without crashes. Shows applied examples: vehicle price prediction and diabetic retinopathy image models, and positions POWER9 with NVIDIA Tesla V100 as the production platform.
AI-written summary · about
chapters (5)
- 0:00Introduction to AI on Power9
- 14:04Power9 system features
- 27:55AI applications and accelerators
- 39:53Use cases and future outlook
- 42:42Q&A session
related talks
44mArchitecture Effects When Interacting with GPU Technologies
25mHAL: Computer System for Scalable Deep Learning
12mAI-powered infrastructure:
40mFueling the AI Revolution with Gaming
46mWhen AI Meets High Performance Computing
25mBenchmarking Graphical Models on POWER with Healthcare...
50mThe democratization of tomorrow’s technology
31mHigh-Performance MPI and Deep Learning on OpenPOWER Platform