[fsdp] feat: add NPU fusion kernels for Qwen3 MoE #3221
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
tardis-key
reviewed
Aug 27, 2025
Shangwei-Li
force-pushed
the
moe_fusion
branch
from
9bbff66 to
3a394c9
Compare
tongyx361
changed the title
Shangwei-Li
force-pushed
the
moe_fusion
branch
from
1fad7ea to
d6e7822
Compare
Shangwei-Li
force-pushed
the
moe_fusion
branch
from
d6e7822 to
5c4525f
Compare
Collaborator
|
LGTM, can we review and merge it? @vermouth1992 |
vermouth1992
approved these changes
Sep 1, 2025
yellowbee686
pushed a commit
to yellowbee686/verl
that referenced
this pull request
Sep 4, 2025
### What does this PR do? This PR adds following NPU fusion kernels to Qwen3 MoE model in Transformers: GroupedMatMul, SwiGLU, RMSNorm and RoPE. ### Checklist Before Starting - [x] Search for similar PRs. Paste at least one query link here: https://github.com/volcengine/verl/pulls?q=fusion++npu+moe - [x] Format the PR title as `[{modules}] {type}: {description}` (This will be checked by the CI) - `{modules}` include `fsdp`, `megatron`, `sglang`, `vllm`, `rollout`, `trainer`, `ci`, `training_utils`, `recipe`, `hardware`, `deployment`, `ray`, `worker`, `single_controller`, `misc`, `perf`, `model`, `algo`, `env`, `tool`, `ckpt`, `doc`, `data` - If this PR involves multiple modules, separate them with `,` like `[megatron, fsdp, doc]` - `{type}` is in `feat`, `fix`, `refactor`, `chore`, `test` - If this PR breaks any API (CLI arguments, config, function signature, etc.), add `[BREAKING]` to the beginning of the title. - Example: `[BREAKING][fsdp, megatron] feat: dynamic batching` ### Test > For changes that can not be tested by CI (e.g., algorithm implementation, new model support), validate by experiment(s) and show results like training curve plots, evaluation results, etc. Tested with Qwen3-30B-A3B sp8 fsdp32 on Ascend A2: Without kernel fusion: <img width="1832" height="468" alt="image" src="https://github.com/user-attachments/assets/a8632a94-3a27-46f6-b408-2ebc09a37aa3" /> WIth kernel fusion <img width="1842" height="440" alt="image" src="https://github.com/user-attachments/assets/50b8cc21-6720-42bc-9a9d-ae684f4cb0bf" /> Test results with train_prompt_bsz=512 sp8 fsdp32 on Ascend A2. The orange line represents GPU, the pink line represents NPU, max absolute error in reward is less than 5%. <img width="718" height="444" alt="image" src="https://github.com/user-attachments/assets/3d4a47f6-fb91-40a6-a8e6-bf39545f8375" /> ### API and Usage Example > Demonstrate how the API changes if any, and provide usage example(s) if possible. ```python # Add code snippet or script demonstrating how to use this ``` ### Design & Code Changes > Demonstrate the high-level design if this PR is complex, and list the specific changes. ### Checklist Before Submitting > [!IMPORTANT] > Please check all the following items before requesting a review, otherwise the reviewer might deprioritize this PR for review. - [x] Read the [Contribute Guide](https://github.com/volcengine/verl/blob/main/CONTRIBUTING.md). - [x] Apply [pre-commit checks](https://github.com/volcengine/verl/blob/main/CONTRIBUTING.md#code-linting-and-formatting): `pre-commit install && pre-commit run --all-files --show-diff-on-failure --color=always` - [x] Add / Update [the documentation](https://github.com/volcengine/verl/tree/main/docs). - [x] Add unit or end-to-end test(s) to [the CI workflow](https://github.com/volcengine/verl/tree/main/.github/workflows) to cover all the code. If not feasible, explain why: ... - [x] Once your PR is ready for CI, send a message in [the `ci-request` channel](https://verl-project.slack.com/archives/C091TCESWB1) in [the `verl` Slack workspace](https://join.slack.com/t/verl-project/shared_invite/zt-3855yhg8g-CTkqXu~hKojPCmo7k_yXTQ). (If not accessible, please try [the Feishu group (飞书群)](https://applink.larkoffice.com/client/chat/chatter/add_by_link?link_token=772jd4f1-cd91-441e-a820-498c6614126a).) Co-authored-by: Shangwei-Li <[email protected]>
cczitong123
pushed a commit
to cczitong123/verl
that referenced
this pull request
Sep 5, 2025
### What does this PR do? This PR adds following NPU fusion kernels to Qwen3 MoE model in Transformers: GroupedMatMul, SwiGLU, RMSNorm and RoPE. ### Checklist Before Starting - [x] Search for similar PRs. Paste at least one query link here: https://github.com/volcengine/verl/pulls?q=fusion++npu+moe - [x] Format the PR title as `[{modules}] {type}: {description}` (This will be checked by the CI) - `{modules}` include `fsdp`, `megatron`, `sglang`, `vllm`, `rollout`, `trainer`, `ci`, `training_utils`, `recipe`, `hardware`, `deployment`, `ray`, `worker`, `single_controller`, `misc`, `perf`, `model`, `algo`, `env`, `tool`, `ckpt`, `doc`, `data` - If this PR involves multiple modules, separate them with `,` like `[megatron, fsdp, doc]` - `{type}` is in `feat`, `fix`, `refactor`, `chore`, `test` - If this PR breaks any API (CLI arguments, config, function signature, etc.), add `[BREAKING]` to the beginning of the title. - Example: `[BREAKING][fsdp, megatron] feat: dynamic batching` ### Test > For changes that can not be tested by CI (e.g., algorithm implementation, new model support), validate by experiment(s) and show results like training curve plots, evaluation results, etc. Tested with Qwen3-30B-A3B sp8 fsdp32 on Ascend A2: Without kernel fusion: <img width="1832" height="468" alt="image" src="https://github.com/user-attachments/assets/a8632a94-3a27-46f6-b408-2ebc09a37aa3" /> WIth kernel fusion <img width="1842" height="440" alt="image" src="https://github.com/user-attachments/assets/50b8cc21-6720-42bc-9a9d-ae684f4cb0bf" /> Test results with train_prompt_bsz=512 sp8 fsdp32 on Ascend A2. The orange line represents GPU, the pink line represents NPU, max absolute error in reward is less than 5%. <img width="718" height="444" alt="image" src="https://github.com/user-attachments/assets/3d4a47f6-fb91-40a6-a8e6-bf39545f8375" /> ### API and Usage Example > Demonstrate how the API changes if any, and provide usage example(s) if possible. ```python # Add code snippet or script demonstrating how to use this ``` ### Design & Code Changes > Demonstrate the high-level design if this PR is complex, and list the specific changes. ### Checklist Before Submitting > [!IMPORTANT] > Please check all the following items before requesting a review, otherwise the reviewer might deprioritize this PR for review. - [x] Read the [Contribute Guide](https://github.com/volcengine/verl/blob/main/CONTRIBUTING.md). - [x] Apply [pre-commit checks](https://github.com/volcengine/verl/blob/main/CONTRIBUTING.md#code-linting-and-formatting): `pre-commit install && pre-commit run --all-files --show-diff-on-failure --color=always` - [x] Add / Update [the documentation](https://github.com/volcengine/verl/tree/main/docs). - [x] Add unit or end-to-end test(s) to [the CI workflow](https://github.com/volcengine/verl/tree/main/.github/workflows) to cover all the code. If not feasible, explain why: ... - [x] Once your PR is ready for CI, send a message in [the `ci-request` channel](https://verl-project.slack.com/archives/C091TCESWB1) in [the `verl` Slack workspace](https://join.slack.com/t/verl-project/shared_invite/zt-3855yhg8g-CTkqXu~hKojPCmo7k_yXTQ). (If not accessible, please try [the Feishu group (飞书群)](https://applink.larkoffice.com/client/chat/chatter/add_by_link?link_token=772jd4f1-cd91-441e-a820-498c6614126a).) Co-authored-by: Shangwei-Li <[email protected]>
DDVD233
pushed a commit
to DDVD233/mirl
that referenced
this pull request
Sep 5, 2025
### What does this PR do? This PR adds following NPU fusion kernels to Qwen3 MoE model in Transformers: GroupedMatMul, SwiGLU, RMSNorm and RoPE. ### Checklist Before Starting - [x] Search for similar PRs. Paste at least one query link here: https://github.com/volcengine/verl/pulls?q=fusion++npu+moe - [x] Format the PR title as `[{modules}] {type}: {description}` (This will be checked by the CI) - `{modules}` include `fsdp`, `megatron`, `sglang`, `vllm`, `rollout`, `trainer`, `ci`, `training_utils`, `recipe`, `hardware`, `deployment`, `ray`, `worker`, `single_controller`, `misc`, `perf`, `model`, `algo`, `env`, `tool`, `ckpt`, `doc`, `data` - If this PR involves multiple modules, separate them with `,` like `[megatron, fsdp, doc]` - `{type}` is in `feat`, `fix`, `refactor`, `chore`, `test` - If this PR breaks any API (CLI arguments, config, function signature, etc.), add `[BREAKING]` to the beginning of the title. - Example: `[BREAKING][fsdp, megatron] feat: dynamic batching` ### Test > For changes that can not be tested by CI (e.g., algorithm implementation, new model support), validate by experiment(s) and show results like training curve plots, evaluation results, etc. Tested with Qwen3-30B-A3B sp8 fsdp32 on Ascend A2: Without kernel fusion: <img width="1832" height="468" alt="image" src="https://github.com/user-attachments/assets/a8632a94-3a27-46f6-b408-2ebc09a37aa3" /> WIth kernel fusion <img width="1842" height="440" alt="image" src="https://github.com/user-attachments/assets/50b8cc21-6720-42bc-9a9d-ae684f4cb0bf" /> Test results with train_prompt_bsz=512 sp8 fsdp32 on Ascend A2. The orange line represents GPU, the pink line represents NPU, max absolute error in reward is less than 5%. <img width="718" height="444" alt="image" src="https://github.com/user-attachments/assets/3d4a47f6-fb91-40a6-a8e6-bf39545f8375" /> ### API and Usage Example > Demonstrate how the API changes if any, and provide usage example(s) if possible. ```python # Add code snippet or script demonstrating how to use this ``` ### Design & Code Changes > Demonstrate the high-level design if this PR is complex, and list the specific changes. ### Checklist Before Submitting > [!IMPORTANT] > Please check all the following items before requesting a review, otherwise the reviewer might deprioritize this PR for review. - [x] Read the [Contribute Guide](https://github.com/volcengine/verl/blob/main/CONTRIBUTING.md). - [x] Apply [pre-commit checks](https://github.com/volcengine/verl/blob/main/CONTRIBUTING.md#code-linting-and-formatting): `pre-commit install && pre-commit run --all-files --show-diff-on-failure --color=always` - [x] Add / Update [the documentation](https://github.com/volcengine/verl/tree/main/docs). - [x] Add unit or end-to-end test(s) to [the CI workflow](https://github.com/volcengine/verl/tree/main/.github/workflows) to cover all the code. If not feasible, explain why: ... - [x] Once your PR is ready for CI, send a message in [the `ci-request` channel](https://verl-project.slack.com/archives/C091TCESWB1) in [the `verl` Slack workspace](https://join.slack.com/t/verl-project/shared_invite/zt-3855yhg8g-CTkqXu~hKojPCmo7k_yXTQ). (If not accessible, please try [the Feishu group (飞书群)](https://applink.larkoffice.com/client/chat/chatter/add_by_link?link_token=772jd4f1-cd91-441e-a820-498c6614126a).) Co-authored-by: Shangwei-Li <[email protected]>
WncFht
pushed a commit
to WncFht/verl
that referenced
this pull request
Oct 10, 2025
### What does this PR do? This PR adds following NPU fusion kernels to Qwen3 MoE model in Transformers: GroupedMatMul, SwiGLU, RMSNorm and RoPE. ### Checklist Before Starting - [x] Search for similar PRs. Paste at least one query link here: https://github.com/volcengine/verl/pulls?q=fusion++npu+moe - [x] Format the PR title as `[{modules}] {type}: {description}` (This will be checked by the CI) - `{modules}` include `fsdp`, `megatron`, `sglang`, `vllm`, `rollout`, `trainer`, `ci`, `training_utils`, `recipe`, `hardware`, `deployment`, `ray`, `worker`, `single_controller`, `misc`, `perf`, `model`, `algo`, `env`, `tool`, `ckpt`, `doc`, `data` - If this PR involves multiple modules, separate them with `,` like `[megatron, fsdp, doc]` - `{type}` is in `feat`, `fix`, `refactor`, `chore`, `test` - If this PR breaks any API (CLI arguments, config, function signature, etc.), add `[BREAKING]` to the beginning of the title. - Example: `[BREAKING][fsdp, megatron] feat: dynamic batching` ### Test > For changes that can not be tested by CI (e.g., algorithm implementation, new model support), validate by experiment(s) and show results like training curve plots, evaluation results, etc. Tested with Qwen3-30B-A3B sp8 fsdp32 on Ascend A2: Without kernel fusion: <img width="1832" height="468" alt="image" src="https://github.com/user-attachments/assets/a8632a94-3a27-46f6-b408-2ebc09a37aa3" /> WIth kernel fusion <img width="1842" height="440" alt="image" src="https://github.com/user-attachments/assets/50b8cc21-6720-42bc-9a9d-ae684f4cb0bf" /> Test results with train_prompt_bsz=512 sp8 fsdp32 on Ascend A2. The orange line represents GPU, the pink line represents NPU, max absolute error in reward is less than 5%. <img width="718" height="444" alt="image" src="https://github.com/user-attachments/assets/3d4a47f6-fb91-40a6-a8e6-bf39545f8375" /> ### API and Usage Example > Demonstrate how the API changes if any, and provide usage example(s) if possible. ```python # Add code snippet or script demonstrating how to use this ``` ### Design & Code Changes > Demonstrate the high-level design if this PR is complex, and list the specific changes. ### Checklist Before Submitting > [!IMPORTANT] > Please check all the following items before requesting a review, otherwise the reviewer might deprioritize this PR for review. - [x] Read the [Contribute Guide](https://github.com/volcengine/verl/blob/main/CONTRIBUTING.md). - [x] Apply [pre-commit checks](https://github.com/volcengine/verl/blob/main/CONTRIBUTING.md#code-linting-and-formatting): `pre-commit install && pre-commit run --all-files --show-diff-on-failure --color=always` - [x] Add / Update [the documentation](https://github.com/volcengine/verl/tree/main/docs). - [x] Add unit or end-to-end test(s) to [the CI workflow](https://github.com/volcengine/verl/tree/main/.github/workflows) to cover all the code. If not feasible, explain why: ... - [x] Once your PR is ready for CI, send a message in [the `ci-request` channel](https://verl-project.slack.com/archives/C091TCESWB1) in [the `verl` Slack workspace](https://join.slack.com/t/verl-project/shared_invite/zt-3855yhg8g-CTkqXu~hKojPCmo7k_yXTQ). (If not accessible, please try [the Feishu group (飞书群)](https://applink.larkoffice.com/client/chat/chatter/add_by_link?link_token=772jd4f1-cd91-441e-a820-498c6614126a).) Co-authored-by: Shangwei-Li <[email protected]>
masoudhashemi
pushed a commit
to masoudhashemi/verl
that referenced
this pull request
Oct 19, 2025
### What does this PR do? This PR adds following NPU fusion kernels to Qwen3 MoE model in Transformers: GroupedMatMul, SwiGLU, RMSNorm and RoPE. ### Checklist Before Starting - [x] Search for similar PRs. Paste at least one query link here: https://github.com/volcengine/verl/pulls?q=fusion++npu+moe - [x] Format the PR title as `[{modules}] {type}: {description}` (This will be checked by the CI) - `{modules}` include `fsdp`, `megatron`, `sglang`, `vllm`, `rollout`, `trainer`, `ci`, `training_utils`, `recipe`, `hardware`, `deployment`, `ray`, `worker`, `single_controller`, `misc`, `perf`, `model`, `algo`, `env`, `tool`, `ckpt`, `doc`, `data` - If this PR involves multiple modules, separate them with `,` like `[megatron, fsdp, doc]` - `{type}` is in `feat`, `fix`, `refactor`, `chore`, `test` - If this PR breaks any API (CLI arguments, config, function signature, etc.), add `[BREAKING]` to the beginning of the title. - Example: `[BREAKING][fsdp, megatron] feat: dynamic batching` ### Test > For changes that can not be tested by CI (e.g., algorithm implementation, new model support), validate by experiment(s) and show results like training curve plots, evaluation results, etc. Tested with Qwen3-30B-A3B sp8 fsdp32 on Ascend A2: Without kernel fusion: <img width="1832" height="468" alt="image" src="https://github.com/user-attachments/assets/a8632a94-3a27-46f6-b408-2ebc09a37aa3" /> WIth kernel fusion <img width="1842" height="440" alt="image" src="https://github.com/user-attachments/assets/50b8cc21-6720-42bc-9a9d-ae684f4cb0bf" /> Test results with train_prompt_bsz=512 sp8 fsdp32 on Ascend A2. The orange line represents GPU, the pink line represents NPU, max absolute error in reward is less than 5%. <img width="718" height="444" alt="image" src="https://github.com/user-attachments/assets/3d4a47f6-fb91-40a6-a8e6-bf39545f8375" /> ### API and Usage Example > Demonstrate how the API changes if any, and provide usage example(s) if possible. ```python # Add code snippet or script demonstrating how to use this ``` ### Design & Code Changes > Demonstrate the high-level design if this PR is complex, and list the specific changes. ### Checklist Before Submitting > [!IMPORTANT] > Please check all the following items before requesting a review, otherwise the reviewer might deprioritize this PR for review. - [x] Read the [Contribute Guide](https://github.com/volcengine/verl/blob/main/CONTRIBUTING.md). - [x] Apply [pre-commit checks](https://github.com/volcengine/verl/blob/main/CONTRIBUTING.md#code-linting-and-formatting): `pre-commit install && pre-commit run --all-files --show-diff-on-failure --color=always` - [x] Add / Update [the documentation](https://github.com/volcengine/verl/tree/main/docs). - [x] Add unit or end-to-end test(s) to [the CI workflow](https://github.com/volcengine/verl/tree/main/.github/workflows) to cover all the code. If not feasible, explain why: ... - [x] Once your PR is ready for CI, send a message in [the `ci-request` channel](https://verl-project.slack.com/archives/C091TCESWB1) in [the `verl` Slack workspace](https://join.slack.com/t/verl-project/shared_invite/zt-3855yhg8g-CTkqXu~hKojPCmo7k_yXTQ). (If not accessible, please try [the Feishu group (飞书群)](https://applink.larkoffice.com/client/chat/chatter/add_by_link?link_token=772jd4f1-cd91-441e-a820-498c6614126a).) Co-authored-by: Shangwei-Li <[email protected]>
7 tasks
techkang
pushed a commit
to techkang/verl
that referenced
this pull request
Oct 31, 2025
### What does this PR do? This PR adds following NPU fusion kernels to Qwen3 MoE model in Transformers: GroupedMatMul, SwiGLU, RMSNorm and RoPE. ### Checklist Before Starting - [x] Search for similar PRs. Paste at least one query link here: https://github.com/volcengine/verl/pulls?q=fusion++npu+moe - [x] Format the PR title as `[{modules}] {type}: {description}` (This will be checked by the CI) - `{modules}` include `fsdp`, `megatron`, `sglang`, `vllm`, `rollout`, `trainer`, `ci`, `training_utils`, `recipe`, `hardware`, `deployment`, `ray`, `worker`, `single_controller`, `misc`, `perf`, `model`, `algo`, `env`, `tool`, `ckpt`, `doc`, `data` - If this PR involves multiple modules, separate them with `,` like `[megatron, fsdp, doc]` - `{type}` is in `feat`, `fix`, `refactor`, `chore`, `test` - If this PR breaks any API (CLI arguments, config, function signature, etc.), add `[BREAKING]` to the beginning of the title. - Example: `[BREAKING][fsdp, megatron] feat: dynamic batching` ### Test > For changes that can not be tested by CI (e.g., algorithm implementation, new model support), validate by experiment(s) and show results like training curve plots, evaluation results, etc. Tested with Qwen3-30B-A3B sp8 fsdp32 on Ascend A2: Without kernel fusion: <img width="1832" height="468" alt="image" src="https://github.com/user-attachments/assets/a8632a94-3a27-46f6-b408-2ebc09a37aa3" /> WIth kernel fusion <img width="1842" height="440" alt="image" src="https://github.com/user-attachments/assets/50b8cc21-6720-42bc-9a9d-ae684f4cb0bf" /> Test results with train_prompt_bsz=512 sp8 fsdp32 on Ascend A2. The orange line represents GPU, the pink line represents NPU, max absolute error in reward is less than 5%. <img width="718" height="444" alt="image" src="https://github.com/user-attachments/assets/3d4a47f6-fb91-40a6-a8e6-bf39545f8375" /> ### API and Usage Example > Demonstrate how the API changes if any, and provide usage example(s) if possible. ```python # Add code snippet or script demonstrating how to use this ``` ### Design & Code Changes > Demonstrate the high-level design if this PR is complex, and list the specific changes. ### Checklist Before Submitting > [!IMPORTANT] > Please check all the following items before requesting a review, otherwise the reviewer might deprioritize this PR for review. - [x] Read the [Contribute Guide](https://github.com/volcengine/verl/blob/main/CONTRIBUTING.md). - [x] Apply [pre-commit checks](https://github.com/volcengine/verl/blob/main/CONTRIBUTING.md#code-linting-and-formatting): `pre-commit install && pre-commit run --all-files --show-diff-on-failure --color=always` - [x] Add / Update [the documentation](https://github.com/volcengine/verl/tree/main/docs). - [x] Add unit or end-to-end test(s) to [the CI workflow](https://github.com/volcengine/verl/tree/main/.github/workflows) to cover all the code. If not feasible, explain why: ... - [x] Once your PR is ready for CI, send a message in [the `ci-request` channel](https://verl-project.slack.com/archives/C091TCESWB1) in [the `verl` Slack workspace](https://join.slack.com/t/verl-project/shared_invite/zt-3855yhg8g-CTkqXu~hKojPCmo7k_yXTQ). (If not accessible, please try [the Feishu group (飞书群)](https://applink.larkoffice.com/client/chat/chatter/add_by_link?link_token=772jd4f1-cd91-441e-a820-498c6614126a).) Co-authored-by: Shangwei-Li <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
What does this PR do?
This PR adds following NPU fusion kernels to Qwen3 MoE model in Transformers: GroupedMatMul, SwiGLU, RMSNorm and RoPE.
Checklist Before Starting
[{modules}] {type}: {description}(This will be checked by the CI){modules}includefsdp,megatron,sglang,vllm,rollout,trainer,ci,training_utils,recipe,hardware,deployment,ray,worker,single_controller,misc,perf,model,algo,env,tool,ckpt,doc,data,like[megatron, fsdp, doc]{type}is infeat,fix,refactor,chore,test[BREAKING]to the beginning of the title.[BREAKING][fsdp, megatron] feat: dynamic batchingTest
Tested with Qwen3-30B-A3B sp8 fsdp32 on Ascend A2:
Without kernel fusion:
WIth kernel fusion
Test results with train_prompt_bsz=512 sp8 fsdp32 on Ascend A2. The orange line represents GPU, the pink line represents NPU, max absolute error in reward is less than 5%.
API and Usage Example
# Add code snippet or script demonstrating how to use thisDesign & Code Changes
Checklist Before Submitting
Important
Please check all the following items before requesting a review, otherwise the reviewer might deprioritize this PR for review.
pre-commit install && pre-commit run --all-files --show-diff-on-failure --color=alwaysci-requestchannel in theverlSlack workspace. (If not accessible, please try the Feishu group (飞书群).)