Clojure is Not Afraid of the GPU
summary
Clojure can drive GPUs with ClojureCL, run high‑performance linear algebra with Neanderthal, and execute massive MCMC with Bayadera. Example results: an 8,000×8,000 matrix multiply drops from 18 s on the CPU to ~0.3 s on the GPU (~60×, ~1,200× versus an optimized Java library). Bayadera produces ~60 million complex samples in 267 ms (~4 ns/sample) and 100 million normal draws in 4 ms versus 27 s in the comparator (≈7,000×).
AI-written summary · about
chapters (6)
- 0:00Start
- 12:06Undertone Support for Gpus
- 15:17Matrix Multiplication
- 20:53Opencl
- 33:04Markov Chain Monte Carlo Simulation
- 37:44Mcmc Algorithm in Closure on Jvm
related talks
25mMore Clojure Less Complication
26mThe Feynman Machine: AI architecture in Clojure/Neanderthal
50mUnder the HAT: Empowering GPU Acceleration for Java
38mScientific Clojure, a bird's eye view
40mThe Tensors Must Flow
32mCyfra: Scala on GPU
40mBuilding Machine Learning Models with Clojure and Cortex
29mDeep Learning Needs Clojure