High-performance computing on gamer PCs
summary
A GPU cluster built from consumer cards runs reservoir computing for tomographic deconvolution, using Python with PyCUDA and small C kernels, NetworkX, Cython and Sage. The system combines reservoir networks with particle‑swarm + genetic optimization and a Jeopardy‑like selector, maps everything to matrix–vector kernels, and fits up to seven GPUs per node (GTX‑series cards). Measured wins include a Python 11s routine reduced to 0.12s; practical issues: PCI lanes, heat and failing cards.
AI-written summary · about
chapters (8)
- 0:00High-performance computing on gamer PCs
- 6:51Reservoir computing & Fly High
- 13:35Swarm optimization & genetic algorithms
- 20:52Literate programming for science
- 26:58HPC software stack for gamer hardware
- 32:16Building multi-GPU cluster
- 38:04Performance vs supercomputers
- 40:14Q&A
related talks
41mLuxuries, necessities, and the challenges that remain
30mCUDA in Your Python: Effective Parallel Programming on the GPU
27mThe art of wrangling your GPU Python environments
36mThe Beauty and the Beast: Python on GPUS
02h 28mGPU Accelerated Python
35mCUDA in Python- A New Era for GPU Acceleration
33mWhen GPUs Make Python Slower: Patterns and Pitfalls
29mA Beginners Guide to GPUs for Pythonistas