Mook 7a59e05dd1 [Kernel] Fuse temperature + softmax in sampling for decode speedup (#20501)

2026-04-02 12:46:36 +08:00

..

all_reduce

refactor: consolidate is_in_ci (jit_kernel, sgl-kernel benchmarks, tests) (#21009)

2026-03-20 05:55:36 -07:00

decoding_attention_triton

Fix benchmark import for should_use_tensor_core (#17232)

2026-01-16 17:48:36 -05:00

deepep

Add CLI args to conveniently support tuning more models (#12922)

2026-03-12 23:10:55 -07:00

deepseek

[Benchmark] use flashinfer bench_gpu_time instead of triton do_bench (#20305)

2026-03-12 04:04:30 +00:00

elementwise

[Benchmark] use flashinfer bench_gpu_time instead of triton do_bench (#20305)

2026-03-12 04:04:30 +00:00

flashinfer_allreduce_fusion

[kernel slimming] Clean many useless sgl-kernel deprecated kernels (#20277)

2026-03-14 16:45:54 +08:00

fused_moe_triton

[Intel GPU] Enable DeepSeek R1 inference on XPU (#18461)

2026-03-29 22:35:59 -07:00

quantization

[Intel GPU] Enable DeepSeek R1 inference on XPU (#18461)

2026-03-29 22:35:59 -07:00

scheduler_batch

[Benchmark] use flashinfer bench_gpu_time instead of triton do_bench (#20305)

2026-03-12 04:04:30 +00:00

sliding_window_attention_triton

[Benchmark] use flashinfer bench_gpu_time instead of triton do_bench (#20305)

2026-03-12 04:04:30 +00:00

bench_fused_temperature_softmax.py

[Kernel] Fuse temperature + softmax in sampling for decode speedup (#20501)

2026-04-02 12:46:36 +08:00