Release Plan for v0.2.0

• Release Manager: TBD
• Code Freeze Date: TBD
• Test Verification/Bug Bash:
• Release Date:
• Release Note:

Features(P0)

• explicit warp specialize
• tile scheduler
• support transposeB=False for Rocm
  • Correctness Evaluation
  • Layout Swizzling

Kernels (P0)

• Implement Flash MLA kernel

  • init version
  • optimize to SoTA
  • MI300

• Implement NSA kernel

  • init version
  • decoding
  • varlen
  • fuse topk
  • bwd
  • MI300

• Implement Flash seerAttention

  • init version
  • different q/kv seq
  • varlen
  • bwd

• optimize TileLang Flash Attention kernel to SoTA

  • H100
  • MI300

• Complete support for commonly used attributes in Flash Attention

  • varlen
  • mask/bias
  • list all supported dims (benchmark)
  • fa3 dim 256 fwd + bwd
  • fa3 bwd (64, 128)

Backends #56

• Pass and Migrate CI to H100
  • fix fp16xfp4 dequant: testing/python/kernel/test_tilelang_kernel_dequantize_gemm.py: test_simple_impl_float16xfp4_gemm
  • fix tma load for float32: testing/python/kernel/test_tilelang_kernel_gemm.py:test_gemm_f32f32f32_nn
• Add support for WebGPU
• Add support for Metal
• Add support for Hexagon

Kernels

• compare with deepGemm
• e2e example: kernel develop flow
• Support FP8/INT8 T.gemm
• Add Examples to CI Test
• optimize TileLang Flash Attention kernel to SoTA on A100

Features

• Nightly Build
• Update API: Replace all tilelang.lower into tilelang.compile in examples and tests.
• Reduce LLVM dependencies
• Provide prebuilt and PyPI packages for ROCm platforms
• Integrate TileLang with Torch Inductor
• Configure API access level to enable advanced features

Cost Model

• Integrate Cost Model Carver into auto-tuning