WTF is Open Source free hardware and how is it influencing ML Data Centers
summary
Open Compute Project designs and open-source hardware are concrete components used inside Facebook, Microsoft Azure, Goldman Sachs and Rackspace and this gear targets a projected $6 billion non-board market by 2021 at ~60% CAGR. Designs strip desktop leftovers, use DC busbars, remove AC PSUs and replace expensive management with a $25 Raspberry Pi; racks become cheaper and denser, servers halve energy use, custom accelerator modules (700W each, 5.6 kW servers) claim ~300x ML training speedups, and ML on cooling cuts energy 20–40%.
AI-written summary · about
chapters (7)
- 0:00Start
- 3:52Open-Source Hardware Manufacturing
- 6:21Open Compute Project
- 19:33Accelerator Modules
- 22:01How Can We Make the Datacenter More Efficient
- 29:28How Do You Make Money in Open Source
- 29:55How You Make Money on Open Data
related talks
32mOpen Source Is Eating the World
13mEngineer your company’s future with AI: Use open source tools on AWS
27mOpen Source in an Age of Transformation for the Energy Industry
05mAI and ML are changing traditional business
41mKeynote: How Open Data Science Opens the World of Innovation
13mDemocratizing AI with Open Source Compute
09mDemocratizing Access to GPUs for Faster MLAI Innovation with Generative AI
03mApplied machine learning at Facebook