Upgrade to Pro — share decks privately, control downloads, hide ads and more …

Accelerating the Machine Learning Pipeline on V...

Sponsored · Ship Features Fearlessly Turn features on and off without deploys. Used by thousands of Ruby developers. →
OmniSci
April 11, 2018

Accelerating the Machine Learning Pipeline on Very Large Datasets with the GPU Data Frame | QCon.AI

Veda Shankar (Senior Developer Advocate) and Wamsi Viswanath (Data Scientist) demonstrates on a very large dataset how to manage a full Machine Learning Pipeline with minimal data exchange overhead between MapD’s SQL engine and H2O’s generalized linear model library (GLM)..

OmniSci

April 11, 2018

Transcript

15 slides
  1. © MapD 2018 Accelerating the Machine Learning Pipeline on Very

    Large Datasets with the GPU Data Frame Veda Shankar & Wamsi Viswanath | April 11, 2018
  2. © MapD 2018 Veda Shankar Sr Developer Advocate , MapD

    Community [email protected] slides: https://speakerdeck.com/mapd Wamsi Viswanath Data Scientist, MapD [email protected]
  3. © MapD 2018 Advanced memory management Three-tier caching to GPU

    RAM for speed and to SSDs for persistent storage 8 SSD or NVRAM STORAGE (L3) 250GB to 20TB 1-2 GB/sec CPU RAM (L2) 32GB to 3TB 70-120 GB/sec GPU RAM (L1) 24GB to 256GB 1000-6000 GB/sec Hot Data Speedup = 1500x to 5000x Over Cold Data Warm Data Speedup = 35x to 120x Over Cold Data Cold Data COMPUTE LAYER STORAGE LAYER Data Lake/Data Warehouse/System Of Record
  4. © MapD 2018 The GPU Open Analytics Initiative (GOAI) Seamless

    data interchange framework in GPU memory 9
  5. © MapD 2018 The GPU Open Analytics Initiative (GOAI) Creating

    common data frameworks to accelerate data science on GPUs 1 0 /mapd/pymapd /gpuopenanalytics/pygdf
  6. © MapD 2018 Machine Learning Pipeline 11 Personas in Analytics

    Lifecycle (Illustrative) Business Analyst Data Scientist Data Engineer IT Systems Admin Data Scientist / Business Analyst Data Preparation Data Discovery & Feature Engineering Model & Validate Predict Operationalize Monitoring & Refinement Evaluate & Decide GPUs
  7. © MapD 2018 • We’ve published a few notebooks showing

    how to connect to a MapD database and use an ML algorithm to make predictions 12 Github ML Examples /gpuopenanalytics/demo-docker /mapd/mapd-ml-demo
  8. © MapD 2018 © MapD 2018 • community.mapd.com Ask questions

    and share your experiences • mapd.com/cloud Try 14-day free trial, no credit card needed • mapd.com/demos Play with our demos • mapd.com/platform/download-community/ Get our free Community Edition and start playing 14 Next Steps

More Decks by OmniSci

Other Decks in Technology