This tool allows you to compile all possible kernels in advance through the model's config.json, and avoids the situation where uncompiled kernel is encountered and JIT is executed when certain requests arrive.
* fix: handle missing logprobs at step 0 and incorrect finish reason with max_completion_tokens
* Prevent response_logprobs.logprob_token_ids[0] from going out of bounds
* Check if the controller port is available
* Global scheduler supports configuring hot updates
* add interface: /controller/scheduler
* add interface: /controller/scheduler
* Add initial support for NVIDIA Blackwell (SM100) architecture
This change introduces initial support for the NVIDIA Blackwell GPU
architecture, specifically targeting SM100 (Compute Capability 10.x)
with '100a' architecture-specific features (e.g., for CUTLASS).
Key changes:
- Updated custom_ops/setup_ops.py to generate appropriate gencode
flags (arch=compute_100a,code=sm_100a) when '100' is specified
in FD_BUILDING_ARCS. Requires CUDA 12.9+.
- Updated custom_ops/gpu_ops/cutlass_extensions/gemm_configs.h:
- Added CutlassTileConfigSM100 enum (with placeholder tile shapes).
- Added BLACKWELL to CandidateConfigTypeParam.
- Updated CutlassGemmConfig struct with is_sm100 flag,
tile_config_sm100, and new constructor for SM100.
- Modified toString() and fromString() for SM100 support.
- Updated custom_ops/gpu_ops/cutlass_kernels/cutlass_heuristic.cu:
- Added get_candidate_tiles_sm100() (with placeholder tiles).
- Added placeholder mcast support functions for SM100.
- Updated get_candidate_configs() to include SM100 paths using
the BLACKWELL flag and new SM100 config types.
- Updated build.sh with comments to guide users on specifying '100'
for Blackwell in FD_BUILDING_ARCS.
Further work:
- Optimal CUTLASS tile configurations for SM100 need to be researched
and updated in cutlass_heuristic.cu.
- Kernel auto-generation scripts in custom_ops/utils/ may need
SM100-specific versions if Blackwell's hardware features for FP8/TMA
differ significantly from SM90.
- Compatibility of third-party libraries (CUTLASS v3.8.0, DeepGEMM)
with Blackwell should be fully verified.
* Feat: Implement detailed Blackwell (SM100) CUTLASS heuristics
This change integrates specific, expert-provided CUTLASS heuristic
configurations for the NVIDIA Blackwell (SM100) GPU architecture,
replacing previous placeholders. This includes:
- Updated `custom_ops/gpu_ops/cutlass_extensions/gemm_configs.h`:
- Populated `CutlassTileConfigSM100` enum with specific tile shapes
(e.g., CtaShape64x64x128B, CtaShape128x128x128B) suitable for SM100.
- Added `FP4_ONLY` to `CandidateConfigTypeParam` for new FP4 paths.
- Updated `custom_ops/gpu_ops/cutlass_kernels/cutlass_heuristic.cu`:
- Implemented `get_candidate_tiles_sm100` with detailed logic for
selecting tile configurations based on GROUPED_GEMM and FP4_ONLY flags,
using the new SM100 tile enums.
- Implemented `supports_mcast_along_m_sm100` and
`supports_mcast_along_n_sm100` with specific tile checks for Blackwell.
- Updated the `sm == 100` (Blackwell) block in `get_candidate_configs`
to use these new helper functions and accurately populate candidate
kernel configurations for various cluster shapes.
- `custom_ops/setup_ops.py` remains configured to compile for
`arch=compute_100a,code=sm_100a` with CUDA 12.9+ for these features.
This aligns the codebase with heuristic configurations similar to those
in upstream TensorRT-LLM / CUTLASS for Blackwell, enabling more
performant kernel selection on this new architecture.
---------
Co-authored-by: google-labs-jules[bot] <161369871+google-labs-jules[bot]@users.noreply.github.com>
Co-authored-by: Jiang-Jia-Jun <163579578+Jiang-Jia-Jun@users.noreply.github.com>
