Deschide articolul original într-un tab nou Schimbarea acestei opțiuni se aplică imediat și reîncarcă pagina.
LECTURA
Înapoi la articole →

Rubber Duck as a Service

kyrieblunders.bearblog.dev/feed/
Vizitează site-ul

Cel mai recent articol

Ultima verificare

Urmăritori 0 urmăritori
Articole: 4

  1. The kernels that were there: where a hand kernel beats the library, and where it can't

    (Rubber Duck as a Service)

    Sequel to "The kernel that wasn't there".

  2. The kernel that wasn't there: profiling my way to real-time streaming Zonos

    (Rubber Duck as a Service)

    Or: what 7% taught me before I wrote a single kernel.

  3. Making Dr GRPO go brrr

    (Rubber Duck as a Service)

    I wrote a fused decode-attention kernel for an RL training loop, got it 2.2× faster than the SDPA path it replaces at the microbenchmark level, dropped it into HuggingFace's generate, and watched the decode step get nearly 3× slower.

  4. Flash Attention 2 in CuteDSL: A Naive Kernel, Three Optimizations, and Where I Got Stuck

    (Rubber Duck as a Service)

    After writing Flash Attention 1 in plain CUDA, I wanted to take the next step and port Flash Attention 2 to CuteDSL, the Python front-end NVIDIA ships with CUTLASS 4.