In the news
Faster Scientific Simulations with New FP64 Algorithms in NVIDIA cuBLAS
NVIDIA Developer · Published · 3 min read
In 30 seconds
- What happened
- CUDA Toolkit 13.4.2 introduces Ozaki-II and 2M algorithms that accelerate double-precision matrix multiplication on integer Tensor Cores while maintaining full FP64 accuracy.
- Why it matters
- Matters for engineers running scientific simulations in quantum chemistry, materials science, and HPC who need both speed and precision in dense matrix operations.
- Watch out
- Algorithms require sufficient mantissa bits to preserve FP64 accuracy; cuBLAS falls back to native FP64 if precision requirements exceed configured limits at runtime.
The patterns behind this
Each one covers how the technique works, when it earns its cost, and where it breaks.
The Agent Architect
One pattern, one tradeoff, one production failure story. A short weekly briefing for people building agentic systems.
Weekly email, one-click unsubscribe. We only use your address to send the briefing.