yangzhuxinyzx 9c5660574b [Core] Share DFlash attention and norm boundaries across TP sizes (#1025) ...

* [Core] Share DFlash attention and norm boundaries across TP sizes

Signed-off-by: yangzhuxinyzx <153831768+yangzhuxinyzx@users.noreply.github.com>

* [Test] Validate installed TP2 and TP4 DFlash boundaries

Signed-off-by: yangzhuxinyzx <153831768+yangzhuxinyzx@users.noreply.github.com>

---------

Signed-off-by: yangzhuxinyzx <153831768+yangzhuxinyzx@users.noreply.github.com>

2026-10-07 15:10:21 +08:00

..

test_activation.py

…

test_apply_rotary_emb.py

…

test_batched_weight_rms_norm.py

…

test_cpu_activation.py

…

test_dflash2_sampler_graph.py

[Perf] Capture compact DFlash2 rejection dispatch on SM70

2026-10-04 11:13:03 +08:00

test_dflash_compact_aux.py

[Memory] Reduce SM70 graph scratch and DFlash active peak

2026-09-23 09:58:34 +08:00

test_fused_q_kv_rmsnorm.py

…

test_fused_qk_norm_rope.py

…

test_fused_quant_layernorm.py

…

test_fused_rms_norm_gated.py

…

test_fused_silu_mul_block_quant.py

…

test_layernorm.py

…

test_minimax_reduce_rms.py

…

test_mrope.py

…

test_opcheck.py

…

test_permute_cols.py

…

test_pos_encoding.py

…

test_rotary_embedding_mla_cache_fused.py

…

test_rotary_embedding.py

…

test_sm70_compact_half_topk.py

[Kernel] Reduce SM70 target compact top-k launches (#942)

2026-10-05 06:55:03 +08:00

test_sm70_dflash2_bmhd.py

[Kernel] Parallelize DFlash2 small-query window attention on SM70 (#944)

2026-10-05 08:13:09 +08:00

test_sm70_dflash2_gemma_rms.py

…

test_sm70_dflash2_qk_rope.py

[Kernel] Fuse SM70 DFlash2 query preparation and attention layouts (#935)

2026-10-05 06:25:52 +08:00

test_sm70_dflash2_split_dispatch.py

[Core] Share DFlash attention and norm boundaries across TP sizes (#1025)

2026-10-07 15:10:21 +08:00

test_sm70_dflash2_split.py

[Core] Share DFlash attention and norm boundaries across TP sizes (#1025)

2026-10-07 15:10:21 +08:00

test_sm70_dflash2_state_audit.py

[Audit] Explain physical state slots in natural DFlash2 comparisons

2026-09-10 05:18:03 +08:00

test_sm70_flashqla_runtime_compiler.py

[Bugfix] Compile SM70 TileLang kernels with pip CUDA runtime

2026-10-04 23:40:56 +08:00

test_sm70_fp8_projection_output.py

[Kernel] Write GDN projections into the declared output buffer

2026-10-04 22:35:23 +08:00

test_sm70_fp32_lm_head.py

…

test_sm70_gdn_preprocess.py

[Test] Check fused GDN graph replay with changed state metadata

2026-10-05 02:32:52 +08:00

test_sm70_gdn_projection_dump.py

…

test_sm70_gdn_verify_tile.py

[Kernel] Tile single-request SM70 GDN verification columns

2026-10-04 12:51:02 +08:00

test_sm70_qwen38_qk_rope.py

[Kernel] Preserve verifier FP16 key rounding before cache encoding

2026-10-04 23:58:17 +08:00

test_sm70_qwen_layer_dump.py

…

test_sm70_sampler_graph_branch.py

[Perf] Capture compact DFlash2 rejection dispatch on SM70

2026-10-04 11:13:03 +08:00

test_uva.py

…

test_vit_bilinear_pos_embed.py

…

test_vit_fp8_attn.py

…

test_vit_fp8_quant.py

…

test_vit_fp8_scaling.py

…