Published on September 18, 2025, this bookmark collects material on CUDA, H100 and cuBLAS matmul optimization, nondeterminism in LLM inference, transformers, scaling, and hardware-model co-design.
Published on September 18, 2025, this bookmark collects readings on GPU and AI performance. Its stated scope includes matmul optimization in CUDA, H100 and cuBLAS, nondeterminism in LLM inference, transformer inference, scaling, and hardware-model co-design. The text points to material on kernel optimization and performance in AI workloads and hardware, but gives no quantitative results or specific conclusions.
To verify the details, consult the original materials referenced by the bookmark and check their methods, test conditions and results; the record provides no individual titles or links. If you use AI to study or apply these techniques, avoid submitting personal or internal data without authorization, and apply your organization’s policy to anything you send.