Posts

All the articles I've posted.

• [TIL] Tilelang Element-wise Op & Vectorization

  2 Jun, 2026

  Notes on three TileLang vector-add kernels, how vectorization appears in IR/CUDA/PTX, and why one-element-per-thread leaves bandwidth on the table.

• [Book Club] FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness

  29 May, 2026

  Paper reading notes on FlashAttention, online softmax, IO complexity, and block-sparse attention.

• [TIL] Multihead Attention

  28 May, 2026

• [Book Club] TileLang: Bridge Programmability And Performance In Modern Neural Kernels

  25 May, 2026