Mohammad Miadh Angkad 2bdd89a6cd [Kernel Slimming] Migrate NVFP4 kernels to JIT (#19437)

2026-03-05 15:22:28 +08:00

..

gptq

…

marlin

[Kernel Slimming] Remove sgl-kernel AOT marlin kernels (#19241)

2026-02-25 10:08:22 +08:00

awq_kernel.cu

…

bmm_fp8.cu

move all get_stream in sgl_kernel to c++ to reduce the launch overhead (#12521)

2025-11-02 13:15:05 -08:00

dsv3_fused_a_gemm.cu

…

dsv3_router_gemm_bf16_out.cu

…

dsv3_router_gemm_entry.cu

…

dsv3_router_gemm_float_out.cu

…

fp8_blockwise_gemm_kernel.cu

fix(sgl-kernel): use >= 120 for SM12x CUDA kernel dispatch (#18750)

2026-02-16 00:44:47 +08:00

fp8_gemm_kernel.cu

[sgl-kernel] feat: Support sm120 cutlass fp8 gemm kernel (#9403)

2025-10-26 23:45:45 -07:00

int8_gemm_kernel.cu

Optimize cutlass int8 gemm kernel for large M on SM89 Ada GPU (#10714)

2025-09-21 17:04:27 -07:00

math.hpp

…

per_tensor_quant_fp8.cu

…

per_token_group_quant_8bit_v2.cu

update pre-commit config (#18860)

2026-02-16 00:18:31 +08:00

per_token_group_quant_8bit.cu

Reland [1/2] Optimizations and refactors about quant kernel (#10312)

2025-10-11 15:59:03 +08:00

per_token_quant_fp8.cu

support smem in per_token_quant_fp8 kernel (#16725)

2026-02-02 17:18:50 +08:00

qserve_w4a8_per_chn_gemm.cu

…

qserve_w4a8_per_group_gemm.cu

…