EX Engine d21b2505bb fix: wire MoE topk via ixformer C++ bridge + disable broken flash_qla GDN ...

Two call chain breaks fixed:

1. MoE routing (2304 calls/token):
   BEFORE: torch.softmax + torch.topk (3 Python GPU ops, no ixformer)
   AFTER:  ix_bridge.py → ix_moe_bridge.cpp → ixformer::infer::topk_softmax()
   Source: upstream_ref/xllm/xllm/core/kernels/ilu/fused_moe.cpp line 46
   The C++ API exists in base image SDK (ixformer.h declares it),
   only the Python binding (ixformer.functions) was missing.

2. GDN prefill (4 layers, 99.98% NaN):
   BEFORE: flash_qla SM70 kernel → abs mean=inf → nan_to_num → zeros
   AFTER:  skip flash_qla, use _pytorch_forward directly
   Source: upstream_ref/xllm qwen3_gated_delta_net_base.cpp uses
   identical PyTorch chunked logic (no flash_qla).
   Sub168 (working build) never deployed flash_qla either.

Files:
- ex_engine/csrc/ix_moe_bridge.cpp: torch C++ extension calling ixformer C++ API
- ex_engine/python/ix_bridge.py: JIT-compile loader with PyTorch fallback
- qwen3_5.py: import ix_bridge for MoE, disable flash_qla for GDN
- patch_ops.sh: deploy ix_bridge .cpp + .py into vllm model dir

2026-08-10 03:00:35 +00:00

..

flash_qla_sm70

build(SM70): precompile GDN CUDA kernel to .so during docker build

2026-08-10 01:08:38 +00:00

qwen3_5

[DEPLOY] Complete submission: baseline + all optimizations

2026-07-30 16:06:20 +00:00

qwen3_5_moe

[DEPLOY] Complete submission: baseline + all optimizations

2026-07-30 16:06:20 +00:00

reasoning

fix(critical): 3 fixes from sub508 diagnosis — n>1 crash guard + thinking format + content fallback

2026-08-07 07:55:04 +00:00

_custom_ops.py

fix(critical): CCCL policy_selector degradation for MoE — PyTorch fallback for topk_softmax

2026-08-07 08:56:50 +00:00

api_server.py

refactor(api): CCCL tuning_adjacent_difference policy_selector — dynamic error handling

2026-08-08 07:01:37 +00:00

arg_utils.py

[ARCH] Eliminate AST patch scripts — full file replacements only

2026-08-05 08:24:43 +00:00

chat_utils.py

fix(critical): 3 bugs causing 0.0 score — OOM crash + thinking param + multimodal

2026-08-07 07:05:40 +00:00

cli_args.py

[DEPLOY] Complete submission: baseline + all optimizations

2026-07-30 16:06:20 +00:00

logits_processor.py

[ARCH] Eliminate AST patch scripts — full file replacements only

2026-08-05 08:24:43 +00:00

mamba_cache.py

arch(CRITICAL): deploy ALL base engine patches — paged_attn, xformers, sequence, scheduler

2026-08-08 10:48:01 +00:00

model_runner.py

fix(engine): CCCL overflow_cast + checked_allocator patterns for NaN/OOM

2026-08-07 08:56:36 +00:00

paged_attention_v2_pytorch.py

[ENGINE] CCCL SmemResource pattern: pre-allocate staging buffers in V2 paged attention

2026-08-07 04:37:44 +00:00

paged_attn.py

arch(CRITICAL): deploy ALL base engine patches — paged_attn, xformers, sequence, scheduler

2026-08-08 10:48:01 +00:00

patch_model_runner.py

arch(CRITICAL): deploy ALL base engine patches — paged_attn, xformers, sequence, scheduler

2026-08-08 10:48:01 +00:00

patch_numerical_stability.py

fix(cccl): robust fallback — inject module-level safety if regex misses

2026-08-07 09:58:35 +00:00

patch_ops.sh

fix: wire MoE topk via ixformer C++ bridge + disable broken flash_qla GDN

2026-08-10 03:00:35 +00:00

patch_transformers_qwen3_5.py

fix(critical): remove pip install transformers — was breaking corex kernel loading

2026-08-07 10:02:09 +00:00

patch_vllm_tool_parser.py

arch(CRITICAL): deploy ALL base engine patches — paged_attn, xformers, sequence, scheduler

2026-08-08 10:48:01 +00:00

patch_xformers_sdpa_batch_kernel.py

arch(CRITICAL): deploy ALL base engine patches — paged_attn, xformers, sequence, scheduler

2026-08-08 10:48:01 +00:00

patch_xformers_sdpa_batch.py

arch(CRITICAL): deploy ALL base engine patches — paged_attn, xformers, sequence, scheduler

2026-08-08 10:48:01 +00:00

patch_xformers_sdpa_seq_kernel.py

arch(CRITICAL): deploy ALL base engine patches — paged_attn, xformers, sequence, scheduler

2026-08-08 10:48:01 +00:00

patch_xformers_sdpa_seq.py

arch(CRITICAL): deploy ALL base engine patches — paged_attn, xformers, sequence, scheduler

2026-08-08 10:48:01 +00:00

precompile_gdn.py

build(SM70): precompile GDN CUDA kernel to .so during docker build

2026-08-10 01:08:38 +00:00

prefix_prefill.py

fix(critical): v2 kernel softmax normalization was commented out — outputs were wrong

2026-08-07 02:46:46 +00:00

probe_corex_api.py

fix(build): add .dockerignore + safe probe — fix docker build failure

2026-08-08 11:21:43 +00:00

protocol.py

fix(dispatch): radix_sort-inspired size-dispatch — disable thinking for small max_tokens, clamp oversized max_tokens

2026-08-08 07:36:31 +00:00

qwen3_5_base_original.py

arch(CRITICAL): replace custom qwen3_5.py with base original (1369 lines)

2026-08-08 08:10:55 +00:00

qwen3_5.py

fix: wire MoE topk via ixformer C++ bridge + disable broken flash_qla GDN

2026-08-10 03:00:35 +00:00

qwen3coder_tool_parser.py

refactor(tool_parser): CCCL agent_radix_sort_downsweep union TempStorage — phased state

2026-08-08 07:02:22 +00:00

registry.py

[ARCH] Eliminate 2 more patch scripts — registry.py + tool_parsers __init__.py

2026-08-05 08:36:52 +00:00

sampler.py

[BASE] qwen3_6_scripts/sampler.py: CCCL topk unsorted output optimization

2026-08-06 02:55:51 +00:00

scheduler.py

[DEPLOY] Complete submission: baseline + all optimizations

2026-07-30 16:06:20 +00:00

sequence.py

[DEPLOY] Complete submission: baseline + all optimizations

2026-07-30 16:06:20 +00:00

serving_chat.py

fix(d05): remove image_url stripping — model IS multimodal

2026-08-10 00:15:06 +00:00

tool_parsers_init.py

[ARCH] Eliminate 2 more patch scripts — registry.py + tool_parsers __init__.py

2026-08-05 08:36:52 +00:00

verify_functional.py

fix(verify): remove duplicate TC-22~30 test definitions — CCCL test_then.cu audit

2026-08-07 06:36:12 +00:00

xformers.py

feat(xformers): replace Q-only tiling with Q+KV tiling + online softmax

2026-08-07 01:54:52 +00:00