Skip to content

[Kernel] Re-enable mrope triton kernel for CUDA/ROCM platform by default #27216

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

Open
wants to merge 2 commits into
base: main
Choose a base branch
from
+9 −0
Open

[Kernel] Re-enable mrope triton kernel for CUDA/ROCM platform by default #27216

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
2c363df
enable mrope triton kernel
Isotr0py Oct 20, 2025
b4fc9a8
use vllm config
Isotr0py Oct 21, 2025
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 vllm/config/vllm.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -538,6 +538,13 @@ def __post_init__(self):
self.compilation_config.cudagraph_mode.has_full_cudagraphs()
)

# enable mrope custom op with triton kernel when model uses mrope
if (
self.model_config.uses_mrope
and "-mrope" not in self.compilation_config.custom_ops
):
self.compilation_config.custom_ops.append("+mrope")

if self.parallel_config.enable_dbo:
a2a_backend = self.parallel_config.all2all_backend
assert a2a_backend in ["deepep_low_latency", "deepep_high_throughput"], (
Expand Down
2 changes: 2 additions & 0 deletions vllm/model_executor/layers/rotary_embedding/mrope.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -5,6 +5,7 @@
import numpy as np
import torch

from vllm.model_executor.custom_op import CustomOp
from vllm.triton_utils import tl, triton

from .base import RotaryEmbedding
Expand Down Expand Up @@ -199,6 +200,7 @@ def apply_interleaved_rope(x: torch.Tensor, mrope_section: list[int]) -> torch.T
return x_t


@CustomOp.register(name="mrope")
class MRotaryEmbedding(RotaryEmbedding):
"""Rotary Embedding with Multimodal Sections."""

Expand Down