biondizzle 84a2f6d441 perf: replace expert counting O(n*E) comparison with torch.bincount O(n) ...

Bug #5 fix: (sorted_ids.unsqueeze(1) == expert_id_range.unsqueeze(0)).sum(dim=0)
materializes a (num_slots × num_experts) bool tensor every forward — 48K × 384 = 18M
elements. torch.bincount(sorted_ids, minlength=num_experts) gives the same result
in O(n) with no intermediate allocation. ~200× less work.

Also removes the now-unused _expert_id_range buffer.

2026-05-20 02:17:23 +00:00

..

kernel

refactor: copy CuTeDSL kernel into repo with local imports

2026-05-16 02:57:54 +00:00

__init__.py

refactor: copy CuTeDSL kernel into repo with local imports

2026-05-16 02:57:54 +00:00

blackwell_attention.py

Add blackwell_attention module and comprehensive test

2026-05-19 15:30:29 +00:00

bridge.py

fix: zero out x_norm for underflow blocks before division in NVFP4 quantization

2026-05-20 02:16:49 +00:00

csa_attention.py

Fix device reference in full_attention_reference

2026-05-19 08:01:31 +00:00

custom_ops.py

Replace autograd.Function with torch.library.custom_op for Dynamo compat

2026-05-19 01:54:48 +00:00

inverse_rope.py

Replace DeepGEMM fp8_einsum with CuTeDSL NVFP4 for wo_a (o_proj)

2026-05-19 02:36:30 +00:00

moe_pipeline.py

Add pipeline test with real model weights, add swiglu_limit to reference moe_pipeline

2026-05-17 18:07:44 +00:00

native_swa_decode.py

fix: handle 3D swa_indices and correct kv_bf16 expand dims

2026-05-20 01:36:27 +00:00

nvfp4_linear.py

fix: root-cause JIT memory corruption myth, add eager warmup, remove _needs_token_refill

2026-05-20 02:08:01 +00:00

runner.py

perf: replace expert counting O(n*E) comparison with torch.bincount O(n)

2026-05-20 02:17:23 +00:00

shared_expert_pipeline.py

Fix garbled shared_expert_pipeline.py: imports/class were merged

2026-05-19 07:18:10 +00:00

wo_a_grouped_linear.py

Fix wo_a: scatter each group's data at correct offset in padded buffer

2026-05-19 02:45:57 +00:00