1Cat-vLLM

cpu

…

deepgemm

…

ir

…

__init__.py

…

bench_concat_mla_q.py

…

bench_cp_gather_fp8.py

…

benchmark_2d_silu_mul_fp8_quant.py

…

benchmark_activation.py

…

benchmark_block_fp8_gemm.py

…

benchmark_cutlass_moe_fp8.py

…

benchmark_cutlass_moe_nvfp4.py

…

benchmark_deepseek_v4_sm70_sparse_decode.py

…

benchmark_device_communicators.py

…

benchmark_dflash2_lookup.py

…

benchmark_dspark_markov.py

…

benchmark_dspark_metadata.py

…

benchmark_dspark_sm70_attention.py

…

benchmark_flash_v100_bfla_prefill.py

…

benchmark_flash_v100_dflash2_swa.py

…

benchmark_flash_v100_fp8_kv.py

…

benchmark_flash_v100_prefill_decay.py

…

benchmark_flashinfer_decode_sm120.py

…

benchmark_flashinfer_prefill_sm120.py

…

benchmark_fp8_gemm.py

…

benchmark_fused_collective.py

…

benchmark_fused_moe_lora_one_shot.py

…

benchmark_fused_topk.py

…

benchmark_gguf_dense_segment_banks.py

[Kernel] Cover replicated KV head projection segments

2026-10-06 00:42:10 +08:00

benchmark_gguf_dp4a_down.py

…

benchmark_gguf_dp4a_expert.py

…

benchmark_gguf_fp16_ba.py

…

benchmark_gguf_fp16_projection.py

…

benchmark_gguf_iq3_gated.py

…

benchmark_gguf_iq4_native.py

…

benchmark_gguf_lut4_experts.py

[Doc] Record matched IQ4 model quality and partition screening

2026-10-06 05:47:14 +08:00

benchmark_gguf_native_linear.py

…

benchmark_gguf_native_pair.py

[Kernel] Admit measured pure IQ3_XXS pairs with merged canonical fallback

2026-10-06 01:03:43 +08:00

benchmark_gguf_ordered_rows.py

…

benchmark_gguf_pair_reader_cursor.py

…

benchmark_gguf_pinned_ple_numa.py

[Kernel] Read packed GGUF PLE decode rows from pinned TP shards (#996 )

2026-10-06 04:00:12 +08:00

benchmark_gguf_q8_intermediate.py

…

benchmark_gguf_qkv.py

[Kernel] Wire GGUF small-M attention inputs by measured capability

2026-10-06 04:09:09 +08:00

benchmark_gguf_qkvz.py

[Kernel] Wire measured joint GDN inputs with canonical stream aliases

2026-10-06 01:44:44 +08:00

benchmark_gguf_raw_grouped.py

…

benchmark_gguf_scalar_lut.py

[Kernel] Screen bank-aware IQ lookup without expanding weights

2026-10-07 03:47:57 +08:00

benchmark_gguf_sm70.py

…

benchmark_gguf_small_down.py

…

benchmark_gguf_small_outputs.py

[Kernel] Admit measured shared-activation output projections

2026-10-06 03:50:09 +08:00

benchmark_gguf_turbomind_grouped.py

…

benchmark_gguf_turbomind.py

…

benchmark_glm53_fp16_gemv.py

…

benchmark_grouped_gemm_cutlass.py

…

benchmark_h3_vsa_fp32.py

…

benchmark_int8_gemm.py

…

benchmark_layernorm.py

…

benchmark_lora.py

…

benchmark_machete.py

…

benchmark_marlin.py

…

benchmark_mla_k_concat.py

…

benchmark_moe_align_block_size.py

…

benchmark_moe_defaults.py

…

benchmark_moe_permute_unpermute.py

…

benchmark_moe.py

…

benchmark_mrope.py

…

benchmark_mxfp4_qutlass.py

…

benchmark_nvfp4_gemm.py

…

benchmark_nvfp4_quant.py

…

benchmark_nvfp4_qutlass.py

…

benchmark_paged_attention.py

…

benchmark_per_token_group_quant.py

…

benchmark_per_token_quant_fp8.py

…

benchmark_quant.py

…

benchmark_qwen38_fp16_hc_mix_mtp5.py

…

benchmark_qwen38_fp16_verifier_m5.py

…

benchmark_qwen38_nvfp4_qpn_mtp5.py

…

benchmark_qwen38_qsa_mtp5_flash_v100.py

…

benchmark_qwen38_router_mtp5.py

…

benchmark_reshape_and_cache_flash.py

…

benchmark_reshape_and_cache.py

…

benchmark_rmsnorm.py

…

benchmark_rope.py

…

benchmark_selective_state_update.py

…

benchmark_shapes.py

…

benchmark_silu_mul_fp8_quant.py

…

benchmark_sm70_compact_half_topk.py

…

benchmark_sm70_compact_top1_ipc.py

…

benchmark_sm70_device_qsa.py

[Benchmark] Separate direct and protected QSA output buffers

2026-10-09 03:44:34 +08:00

benchmark_sm70_dflash2_grouped_verify.py

…

benchmark_sm70_dflash2_qk_rope.py

…

benchmark_sm70_dflash_batched_prefill.py

…

benchmark_sm70_draft_f16_layout.py

…

benchmark_sm70_exact_row_reduce.py

…

benchmark_sm70_flash_v100_xqa_layout.py

…

benchmark_sm70_fp8_e5m2_cache_write.py

…

benchmark_sm70_fp8_graph_capture.py

…

benchmark_sm70_fp16_gemv_silu.py

…

benchmark_sm70_gdn_input_batch.py

…

benchmark_sm70_gdn_preprocess.py

…

benchmark_sm70_gdn_projection_split.py

…

benchmark_sm70_gdn_projection_tails.py

…

benchmark_sm70_gguf_cache_copy.py

[Perf] Parallelize Flash-Next expert copies and skip empty KV installs

2026-10-14 04:42:46 +08:00

benchmark_sm70_gguf_head.py

…

benchmark_sm70_grouped_attention_long.py

…

benchmark_sm70_grouped_attention_precision.py

…

benchmark_sm70_grouped_attention_qk_stage.py

…

benchmark_sm70_grouped_attention_splits.py

…

benchmark_sm70_hc_batch_native_tp4.py

…

benchmark_sm70_hc_batch_reuse.py

…

benchmark_sm70_hc_batch_tp4.py

…

benchmark_sm70_hc_batch.py

…

benchmark_sm70_hc_cta_chain.py

…

benchmark_sm70_hc_full_chain.py

…

benchmark_sm70_hc_gguf_tp4.py

…

benchmark_sm70_hc_ll.py

[Bench] Compare packaged HC shards and replicated kernels in ABBA

2026-10-06 01:37:05 +08:00

benchmark_sm70_hc_loads.py

[Test] Compare HC replay outputs by FP16 bit pattern

2026-10-07 00:16:17 +08:00

benchmark_sm70_hc_tp4.py

…

benchmark_sm70_hc_waits.py

[Kernel] Record HC boundary screens and sample replay phases

2026-10-06 22:06:51 +08:00

benchmark_sm70_hcx_schedule.py

[Misc] Pre-commit compliance for the Flash-Next prefill branch

2026-10-13 05:41:13 -07:00

benchmark_sm70_hcx_stages.py

[Perf] Attribute packaged HCX phases with coupled TP4 graph replay

2026-10-07 03:39:19 +08:00

benchmark_sm70_hcxo_schedule.py

[Misc] Pre-commit compliance for the Flash-Next prefill branch

2026-10-13 05:41:13 -07:00

benchmark_sm70_lm_head_local_vocab.py

…

benchmark_sm70_long_attention_batch.py

…

benchmark_sm70_long_context_attention.py

…

benchmark_sm70_moe_packed_w13.py

…

benchmark_sm70_mtp5_grouped.py

…

benchmark_sm70_mtp_current_chains.py

…

benchmark_sm70_mtp_fc_gather_pair.py

…

benchmark_sm70_mtp_fp32_dense.py

…

benchmark_sm70_mtp_gdn_output.py

…

benchmark_sm70_mtp_local_top1.py

…

benchmark_sm70_mtp_moe_fp16.py

…

benchmark_sm70_mtp_qpn8_dense.py

…

benchmark_sm70_mtp_strided_qkv.py

…

benchmark_sm70_multihead_attention.py

…

benchmark_sm70_ngram_input.py

…

benchmark_sm70_nvfp4_bundle.py

[Kernel] Bundle native SM70 NVFP4 codes and scales

2026-10-06 22:37:40 +08:00

benchmark_sm70_nvfp4_qpn2.py

…

benchmark_sm70_nvfp4_shared_weight.py

…

benchmark_sm70_packed_topk_gather.py

[Kernel] Gather compact SM70 vocabulary candidates in one message

2026-10-08 05:43:32 +08:00

benchmark_sm70_ple_ngram_batch.py

…

benchmark_sm70_qpn2_chunked.py

…

benchmark_sm70_qpn2_cooperative_mlp.py

…

benchmark_sm70_qpn2_graph_layout.py

…

benchmark_sm70_qpn2_packed_mlp.py

…

benchmark_sm70_qpn2_publish.py

…

benchmark_sm70_qpn2_working_set.py

…

benchmark_sm70_qsa_batch_decode.py

…

benchmark_sm70_qsa_shared_key.py

[Test] Record paired QSA results and check short-context selection

2026-10-09 15:34:11 +08:00

benchmark_sm70_quasar_nvfp4_oracle.py

…

benchmark_sm70_quasar_tp_quality.py

…

benchmark_sm70_qwen38_batch_gate.py

…

benchmark_sm70_qwen38_dense_batch.py

…

benchmark_sm70_ring_atomic_allreduce.py

…

benchmark_sm70_rmsnorm_gated_exact.py

…

benchmark_sm70_router_moe_chain.py

[Kernel] Fuse SM70 router top-k with Q8 activation quantization (#1147 )

2026-10-10 03:57:28 +08:00

benchmark_sm70_router_native.py

[Kernel] Fuse SM70 router top-k with Q8 activation quantization (#1147 )

2026-10-10 03:57:28 +08:00

benchmark_sm70_scalar_attention.py

…

benchmark_sm70_small_moe_routing.py

…

benchmark_sm70_sparse_dense_topk.py

…

benchmark_sm70_tp2_e4m3_scalar.py

…

benchmark_sm70_tp4_norm_packets.py

[Kernel] Reuse M8 gate operands and TP4 norm packets (#1041 )

2026-10-08 08:50:20 +08:00

benchmark_sm70_tp4_small_message_push.py

…

benchmark_trtllm_decode_attention.py

…

benchmark_trtllm_prefill_attention.py

…

benchmark_vit_bilinear_pos_embed.py

…

benchmark_vit_fp8_attn.py

…

benchmark_w8a8_block_fp8.py

…

build_sm70_device_qsa.py

[Attention] Read device QSA history without protected-cache resolution

2026-10-08 13:53:28 +08:00

build_sm70_exact_decode_screen.py

[Kernel] Reuse M8 gate operands and TP4 norm packets (#1041 )

2026-10-08 08:50:20 +08:00

build_sm70_grouped_attention_candidate.py

…

build_sm70_grouped_attention_compact_cta.py

…

build_sm70_grouped_attention_direct_qk.py

…

build_sm70_grouped_attention_n64.py

…

build_sm70_grouped_attention_phase_probe.py

…

build_sm70_grouped_attention_pipeline.py

…

build_sm70_grouped_attention_qk_parallel.py

…

build_sm70_grouped_attention_staged.py

…

build_sm70_grouped_attention_swizzle.py

…

build_sm70_long_context_attention.py

…

build_sm70_native_sort_candidate.py

…

build_sm70_qpn2_chunked_candidate.py

…

build_sm70_qpn2_cooperative_mlp.py

…

build_sm70_qpn2_dual_norm.py

…

build_sm70_qpn2_packed_input_candidate.py

…

build_sm70_qpn2_publish_candidate.py

…

build_sm70_qpn2_q8_candidate.py

…

build_sm70_qsa_topk_sidecar.py

…

build_sm70_qwen38_runtime.py

…

build_sm70_scalar_attention_candidate.py

…

build_sm70_tp2_matched_qpn2.py

…

build_sm70_w13_exact_sidecar.py

…

deepseek_v4_sm70_aot_check.py

…

dflash2_split_attention_reference.py

…

graph_machete_bench.py

…

h3_vsa_fp32.cu

…

requirements.txt

…

sm70_context_probe_candidate_route.py

…

sm70_dflash2_common_candidate_route.py

…

sm70_dflash2_qpn2_candidate_route.py

…

sm70_draft_column_candidate_route.py

…

sm70_fp4_stage_parity.py

[Core] Cache native policy transport outside token execution

2026-10-09 14:32:24 +08:00

sm70_gdn_value_tile_candidate_route.py

…

sm70_grouped_attention_candidate_route.py

…

sm70_hc_batch_screen.cu

…

sm70_moe_stage_parity.py

[Core] Share FP4 MoE stages and capture format policy

2026-10-09 11:50:23 +08:00

sm70_native_artifact_parity.py

[Core] Resolve MoE native stages and close Phase B operator validation

2026-10-09 16:01:01 +08:00

sm70_native_policy_isolation.py

[Core] Cache native policy transport outside token execution

2026-10-09 14:32:24 +08:00

sm70_nvfp4_shared_sidecar.cpp

…

sm70_paired_stats.py

…

sm70_qpn2_graph_layout.py

…

sm70_qpn2_graph_nodes.py

…

sm70_qsa_topk_sidecar.cu

…

sm70_qwen38_custom_ar_sidecar.cu

…

sm70_qwen38_nvfp4_sidecar.cpp

…

sm70_qwen38_qk_rope_micro.py

…

sm70_router_benchmark_utils.py

[Kernel] Fuse SM70 router top-k with Q8 activation quantization (#1147 )

2026-10-10 03:57:28 +08:00

sm70_w13_exact_sidecar.cu

…

utils.py

…

verify_sm70_hc_down_scatter.py

…

verify_sm70_nvfp4_moe_raw_storage.py

…

verify_sm70_output_projection_policy.py

…

verify_sm70_qsa_nvfp4_relocation.py

…

verify_sm70_qsa_resolved.py

…

verify_sm70_qsa_router_exact.py

…

verify_sm70_w13_exact.py

…

weight_shapes.py

…