Interview
summary
MapD runs a SQL database and visual analytics entirely on GPUs, compiling SQL into GPU machine code with an LLVM compiler and cutting workloads that once needed hundreds or thousands of CPU nodes down to five or six by scaling up across multiple GPUs per box. Backend is C++ with that compiler, frontend is React, deployment runs on-premise and in the cloud, and the stack already supports interactive analytics and early machine learning experiments.
AI-written summary · about
related talks
19mScalable Data and AI Solutions with Cloud-Native Architecture
29mLive Keynote
24mMatei Zaharia, Databricks | Databricks Data + AI Summit 2023
12mDigital & Data/AI Strategy and Transformation
14mJohn Mao, VAST Data | Nvidia GTC 2024
13mRAG Without the Hassle: Building AI-Powered Applications
12mDemo: Building cloud-native, AI-powered applications with GKE
34mScaling AI Like a Pro-PepsiCo’s LLM Deployment Strategy on K...