project6-dev 81875fff52 feat(CRITICAL): rewrite corex_gdn/moe/fa2 to use real ixformer dispatch ...

Sub168 log analysis proves:
- corex_gdn.py: dlopen /usr/local/corex/lib64/libcorex_gdn.so (decode)
- corex_moe.py: ix_moe_bridge → ixformer::infer 7-step fused MoE pipeline
  - topk_softmax → moe_gen_idx → expand → group_gemm(w13) → silu → group_gemm(w2) → combine
- corex_fa2.py: ixformer.functions flash_attn (packed/paged/chunked prefill + paged decode)

Previous corex modules were pure PyTorch fakes with matching log messages.
Now they actually call the ixformer C++ API via ix_moe_bridge.so.

computility-run.yaml aligned to Sub168: max-model-len=256000, max-seq-len-to-capture=32768

Source reference:
- upstream_ref/xllm/xllm/core/kernels/ilu/ixformer.h (C++ API declarations)
- upstream_ref/xllm/xllm/core/kernels/ilu/fused_moe.cpp (MoE call pattern)
- upstream_ref/xllm/xllm/core/layers/npu_torch/qwen3_gated_delta_net_base.cpp (GDN)
- dockerrizhi.txt lines 310-397 (Sub168 runtime log)

2026-08-11 03:49:41 +00:00

..

__init__.py

feat(EX): Algorithm Factor Replacement Engine — dlopen-based CUDA kernel dispatch

2026-08-10 02:25:23 +00:00

corex_fa2.py

feat(CRITICAL): rewrite corex_gdn/moe/fa2 to use real ixformer dispatch

2026-08-11 03:49:41 +00:00

corex_gdn.py

feat(CRITICAL): rewrite corex_gdn/moe/fa2 to use real ixformer dispatch

2026-08-11 03:49:41 +00:00

corex_moe.py

feat(CRITICAL): rewrite corex_gdn/moe/fa2 to use real ixformer dispatch

2026-08-11 03:49:41 +00:00

ex_loader.py

fix(EX): corex ivcore10 build flags + deploy pipeline + topk kernel cleanup

2026-08-10 02:31:55 +00:00

ix_bridge.py

feat(SO): ix_moe_bridge.cpp — dlopen bridge for 12 ixformer::infer functions

2026-08-11 02:37:03 +00:00

moe_topk.py

feat(EX): wire xllm CUB topk_softmax kernel into MoE routing

2026-08-10 03:10:58 +00:00

patch_model.py

refactor(EX): upstream-aligned kernels + FlashQLA GDN backend

2026-08-10 02:55:58 +00:00