Wang Yixuan c68ddc11ce [OPS] add bmm_transpose ops (#3990) ...

### What this PR does / why we need it?
Add a new fusion ops to custom_op, which can cobime the torch.bmm() and
transpsose to achieve better peformance. This ops is used in mla_v1 to
replace the bmm and transpose

### Does this PR introduce _any_ user-facing change?
No

### How was this patch tested?


- vLLM version: v0.11.2

---------

Signed-off-by: hust17yixuan <303660421@qq.com>

2025-12-01 09:09:51 +08:00

..

aclnn_torch_adapter

[feature] Add Custom Op grouped_matmul_swiglu_quant (#4431)

2025-11-27 21:56:18 +08:00

batch_matmul_transpose

[OPS] add bmm_transpose ops (#3990)

2025-12-01 09:09:51 +08:00

cmake

[Kernel] add custom op GmmSwigluQuantWeightNzTensorList (#3804)

2025-11-28 18:06:39 +08:00

grouped_matmul_swiglu_quant_weight_nz_tensor_list

[Kernel] add custom op GmmSwigluQuantWeightNzTensorList (#3804)

2025-11-28 18:06:39 +08:00

kernels

[OPS] add bmm_transpose ops (#3990)

2025-12-01 09:09:51 +08:00

mla_preprocess

remove redundant params in mla_preprocess kernel (#3530)

2025-10-21 19:20:13 +08:00

utils

[Kernel] add custom op GmmSwigluQuantWeightNzTensorList (#3804)

2025-11-28 18:06:39 +08:00

build_aclnn.sh

[Kernel] add custom op GmmSwigluQuantWeightNzTensorList (#3804)

2025-11-28 18:06:39 +08:00

build.sh

[Kernel] add custom op GmmSwigluQuantWeightNzTensorList (#3804)

2025-11-28 18:06:39 +08:00

camem_allocator.cpp

Quality enhancement: Immediately interrupt execution when memory OOM (#3932)

2025-11-04 08:55:09 +08:00

CMakeLists.txt

[Kernel] add custom op GmmSwigluQuantWeightNzTensorList (#3804)

2025-11-28 18:06:39 +08:00

ops.h

[OPS] add bmm_transpose ops (#3990)

2025-12-01 09:09:51 +08:00

torch_binding_meta.cpp

[OPS] add bmm_transpose ops (#3990)

2025-12-01 09:09:51 +08:00

torch_binding.cpp

[OPS] add bmm_transpose ops (#3990)

2025-12-01 09:09:51 +08:00

utils.h

[core] Support capture custom ops into aclgraph (#2113)

2025-08-11 15:59:42 +08:00