Skip to content

feat(rocm): grouped + non-grouped keep-quant expert GEMM — the MoE-path terminus (issue #41) #523

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Draft
VikashLoomba wants to merge 2 commits into
base: main
Choose a base branch
from
Draft

feat(rocm): grouped + non-grouped keep-quant expert GEMM — the MoE-path terminus (issue #41) #523

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
8a4ed36
feat(rocm): MoE combine/gate ops (SharedExpertGate, MoeCombine, MoeCo…
VikashLoomba Aug 12, 2026
864b89e
feat(rocm): grouped + non-grouped keep-quant expert GEMM (kMatmulBTQu…
VikashLoomba Aug 12, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
7 changes: 7 additions & 0 deletions CMakeLists.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1363,6 +1363,8 @@ if(VLLM_CPP_HIP)
src/vt/rocm/rocm_gemma4_expert_geglu.hip
src/vt/rocm/rocm_fp8_channel_gemv.hip
src/vt/rocm/rocm_moe_router.hip
src/vt/rocm/rocm_moe_chain.hip
src/vt/rocm/rocm_grouped_gemm.hip
src/vt/rocm/rocm_sample.hip
src/vt/rocm/rocm_gdn_state.hip
src/vt/rocm/rocm_gdn_conv.hip
Expand All @@ -1383,6 +1385,11 @@ if(VLLM_CPP_HIP)
src/vt/rocm/rocm_gemma4_expert_geglu.hip
src/vt/rocm/rocm_fp8_channel_gemv.hip
src/vt/rocm/rocm_moe_router.hip
src/vt/rocm/rocm_moe_chain.hip
src/vt/rocm/rocm_grouped_gemm.hip
src/vt/rocm/rocm_grouped_gemm.hip
src/vt/rocm/rocm_moe_chain.hip
src/vt/rocm/rocm_grouped_gemm.hip
src/vt/rocm/rocm_sample.hip
src/vt/rocm/rocm_gdn_state.hip
src/vt/rocm/rocm_gdn_conv.hip
Expand Down
12 changes: 7 additions & 5 deletions docs/USAGE.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -50,11 +50,13 @@ portable scan), and the norm-gate/preamble ops (`kRmsNormGated`,
GDN-hybrid models call. Compressed conv/SSM state (bf16, the vLLM
`mamba_cache_dtype` default) is advertised via the
`SupportsCompressedConvState`/`SupportsCompressedGdnState` backend probes.
MoE-path coverage is partial: `MoeRouterTopK` (f32/bf16 logits, ungrouped
softmax, no bias) and `MoeSiluMul` are native; the remaining chain
(`kSharedExpertGate`, `kMoeCombine`/`kMoeCombineGate`, and the grouped quant
expert GEMM) is not registered yet, so MoE-bearing models still throw on
those ops. On a
MoE-path coverage: `MoeRouterTopK` (f32/bf16 logits, ungrouped softmax, no
bias), `MoeSiluMul`, `SharedExpertGate`, `MoeCombine`, and `MoeCombineGate`
are native. The grouped keep-quant expert GEMM (`kMatmulBTQuant` /
`kMatmulBTQuantGrouped`, Q8_0/Q4_K/Q5_K/Q6_K) is now native too, so GDN-MoE
GGUF models run on discrete ROCm — a 35B fits one 24GB card with
`--max-num-seqs 1` (the GDN state pool otherwise pushes past 24GB; that is a
residency note, not a kernel defect). On a
discrete card there is no CPU fallback tier, so a model whose layers call an op
that is not registered yet fails loudly with `vt: no kernel for op N on device
type 5` — that is the memory-safety design working, not a crash. Run with
Expand Down
Loading
Loading