Open Source RAPIDS GPU Platform to Accelerate Predictive Data Analytics
1 like1,013 views
The document discusses the evolution and market growth of high-performance computing (HPC) and its applications in various sectors, highlighting a $9 billion market in 2005 projected to reach $36 billion. It emphasizes the importance of tools such as RAPIDS for accelerated data science and machine learning, showcasing contributions from significant figures and companies in the field. Additionally, it notes advancements in GPU technology, particularly from NVIDIA, to enhance the efficiency of data processing and analytics.
Open Source RAPIDS GPU Platform to Accelerate Predictive Data Analytics
- 1. THE NEW HPC MARKET $9B 2005 SCIENTIFIC COMPUTING
- 2. INTERNET DEEP LEARNING SCIENTIFIC COMPUTING THE NEW HPC MARKET SCIENTIFIC COMPUTING $9B 2005 2018 CERN — Switzerland CSCS — Switzerland DFKI — Germany Técnico Lisboa — Portugal CUDA
- 3. INTERNET THE NEW HPC MARKET SCIENTIFIC COMPUTING $9B 2005 2018 MACHINE LEARNING SCIENTIFIC COMPUTING RETAIL HEALTHCARE FINANCIAL SERVICES LOGISTICS TELECOM AD TECH $36B HADOOP NUMPY SKL PANDAS DEEP LEARNING
- 4. 1991 | Guido van Rossum Interpreted language emphasizing readability 2006 | Travis Oliphant Multi-dimensional arrays, math functions 2010 | Inria Machine learning library 2008 | Wes McKinney Data manipulation and analysis NUMPY PYTHON PANDAS SKLEARN THE DEFACTO DATA SCIENCE PLATFORM
- 5. Matthew Rocklin Parallel processing in Python data analytics Dynamic task scheduling Collection of parallel arrays, data frames, lists NUMPY PYTHON PANDAS SKLEARN DASK NUMPY PYTHON PANDAS SKLEARN THE DEFACTO DATA SCIENCE PLATFORM
- 6. CUDA PYTHON CUDF “PANDAS-LIKE” CUML “SKLEARN-LIKE” ARROW DASK NUMPY PYTHON PANDAS SKLEARN DASK NUMPY PYTHON PANDAS SKLEARN 2016 | Wes McKinney Cross-language platform for in-memory data Columnar memory format Vectorized execution engine Zero-copy IPC Designed with GPU in mind ANNOUNCING RAPIDS — ACCELERATED DATA SCIENCE
- 7. CUDA PYTHON ARROW DASK NUMPY PYTHON PANDAS SKLEARN DASK NUMPY PYTHON PANDAS SKLEARN DEEP LEARNING FRAMEWORKS CUDNN RAPIDS CUMLCUDF CUGRAPH ANNOUNCING RAPIDS — ACCELERATED DATA SCIENCE
- 8. CUDA PYTHON ARROW DASK DEEP LEARNING FRAMEWORKS CUDNN RAPIDS CUMLCUDF CUGRAPH ANNOUNCING RAPIDS — ACCELERATED DATA SCIENCE 2010 | Matei Zaharia, CTO, Databricks Unified analytics cluster-computing framework for big data and machine learning Spark Streaming GPU Support cuML and cuGraph with Spark 2.4 Batch cuDF with Spark 2.4
- 9. ENTERPRISE-SCALE DATA SCIENCE DGX-2 DGX-1 DGX STATION RTX 8000 TESLA V100 512GB 256GB 128GB 96GB 32GB
- 10. NEW NVIDIA DGX-2 THE LARGEST GPU EVER CREATED 2 PFLOPS 512GB HBM2 16 TB/sec Memory Bandwidth 10 kW | 160 kg
- 12. 0 3600 7200 10800 DGX-2 100 CPU Nodes 50 CPU Nodes 20 CPU Nodes ETL ML FASTER SPEEDS, REAL WORLD BENEFITS ETL ML End-to-End SECONDS 0 600 1200 1800 2400 3000 3600 DGX-2 100 CPU Nodes 50 CPU Nodes 20 CPU Nodes SECONDSSECONDS 2 Hours 1 Hour 3 Hours 0 1800 3600 5400 7200 DGX-2 100 CPU Nodes 50 CPU Nodes 20 CPU Nodes
- 14. TRADITIONAL DATA SCIENCE CLUSTER 300 Servers $3M 180 kW
- 15. GPU-ACCELERATED MACHINE LEARNING CLUSTER DGX-2 AND RAPIDS FOR PREDICTIVE ANALYTICS 1 DGX-2 10 kW 1/8 the Cost 1/15 the Space 1/18 the Power
- 16. ACCELERATING MACHINE LEARNING WITH RAPIDS RAPIDS software has immensely improved how we use data — enabling the most complex models to run at scale and deliver even more accurate forecasting.” - Jeremy King, EVP & CTO “
- 17. BRINGING RAPIDS-OPTIMIZED SYSTEMS TO ENTERPRISES WORLDWIDE We are excited to partner with NVIDIA on RAPIDS to accelerate the application of data science and machine learning to help our customers drive faster and more insightful outcomes.” - Antonio Neri, CEO “
- 18. INTEGRATING RAPIDS TO ACCELERATE DATA SCIENCE AND MACHINE LEARNING PLATFORMS We look forward to extending our successful partnership with NVIDIA, leveraging RAPIDS to provide new machine learning tools for our clients.” - Arvind Krishna, SVP, Hybrid Cloud and Director of IBM Research “
- 19. ACCELERATING ORACLE CLOUD INFRASTRUCTURE WITH NVIDIA HGX-2 AND RAPIDS RAPIDS software runs seamlessly on the Oracle Cloud, allowing customers to support their HPC, AI and data science needs, all while taking advantage of the portfolio of GPU instances available on Oracle Cloud Infrastructure.” - Clay Magouyrk, SVP, Software Development, Oracle Cloud Infrastructure “
- 20. NVIDIA ACCELERATED HPC PLATFORM DENSE HPC NVIDIA HPC ACCELERATION STACKS HYPERSCALE HPC SCIENCE CUDA DL TRAINING cuDNN DL INFERENCE New TRT Hyperscale MACHINE LEARNING New RAPIDS