### What this PR does / why we need it?
Supports Deepseek-R1 w4a8 quantization.
Since R1 w4a8 uses mixed quantization, only the MOE layer uses
w4a8_dynamic quantization, so we added the w4a8_dynamic.py file, which
includes the AscendW4A8DynamicFusedMoEMethod class.
### Does this PR introduce _any_ user-facing change?
no
### How was this patch tested?
Adding ut case in `tests/ut/quantization/test_w4a8_dynamic.py` and
`tests/ut/quantization/test_quantizer.py`
Adding e2e case in
`tests/e2e/multicard/test_offline_inference_distributed.py::test_models_distributed_DeepSeek_W4A8DYNAMIC`
to test deepseek w4a8_dynamic quantized model
#### 1.How to get weights using Modelslim
##### Installation steps
Use the branch master, the commit id is:
298e175d69b3b855111a1e09bbe2fcd12fdb4e24
git clone https://gitee.com/ascend/msit.git
cd msit/msmodelslim
bash install.sh
##### The required transformers environment
transformers>=4.48.2
##### Generate w4a8 weights
cd /example/DeepSeek
Command reference: msmodelslim/example/DeepSeek/README.md Execute the
[pre-check](https://gitee.com/ascend/msit/blob/master/msmodelslim/example/DeepSeek/README.md#%E8%BF%90%E8%A1%8C%E5%89%8D%E5%BF%85%E6%A3%80)
and [DeepSeek-R1 w4a8 mix
quantization](https://gitee.com/ascend/msit/blob/master/msmodelslim/example/DeepSeek/README.md#deepseek-r1-w4a8-%E6%B7%B7%E5%90%88%E9%87%8F%E5%8C%96%E5%89%8D%E4%B8%89%E5%B1%82-mlpw8a8-dynamic-%E9%87%8F%E5%8C%96mla%E5%85%B1%E4%BA%AB%E4%B8%93%E5%AE%B6w8a8%E9%87%8F%E5%8C%96%E8%B7%AF%E7%94%B1%E4%B8%93%E5%AE%B6w4a8-dynamic%E9%87%8F%E5%8C%96)
chapter
Reference command：python3 quant_deepseek_w4a8.py --model_path {Original
weight path} --save_path {Generate weight path} --mindie_format
##### Adapt to vllm-ascend
Since mindie_format generates mindie format, some adaptation
modifications are needed for vllm-ascend to use it:
`quant_model_description_w8a8_dynamic.json` rename to
`quant_model_description.json`, and add `"group_size": 256`
Modification in `config.json`：`"model_type":deepseekv2` is changed to
`"model_type":deepseek_v3`; `quantization_config` is removed;
tips:The group_size and weights match. If the w4a8 weights are not
generated using msmodelslim, you can check the group_size in
quantization_config in config.json.
#### 2.How to run w4a8
##### a.How to run eager mode
export VLLM_USE_V1=1 # v1
python -m vllm.entrypoints.openai.api_server --model=$1
--trust-remote-code -tp $2 -dp $3 --enable_expert_parallel
--quantization ascend --port $4 --max-model-len $5 --max-num-seqs $6
--enforce-eager
eg: python -m vllm.entrypoints.openai.api_server
--model=/weightpath/w4a8_4_layer --trust-remote-code -tp 4 -dp 4
--enable_expert_parallel --quantization ascend --port 8002
--max-model-len 5120 --max-num-seqs 128 --enforce-eager
##### b.How to run graph mode
export VLLM_USE_V1=1 # v1
export HCCL_BUFFSIZE=1024
python -m vllm.entrypoints.openai.api_server --model=$1
--trust-remote-code -tp $2 -dp $3 --enable_expert_parallel
--quantization ascend --port $4 --max-model-len $5
--additional_config='{"ascend_scheduler_config":{"enabled":true},"torchair_graph_config":{"enabled":true}}'
eg: python -m vllm.entrypoints.openai.api_server
--model=/weight/dsr1_w4a8_vllm --trust-remote-code -tp 4 -dp 4
--enable_expert_parallel --quantization ascend --port 8002
--max-model-len 5120
--additional_config='{"ascend_scheduler_config":{"enabled":true},"torchair_graph_config":{"enabled":true}}'
- vLLM version: v0.10.0
- vLLM main:
c494f96fbc
---------
Signed-off-by: Wang Kunpeng <1289706727@qq.com>
This PR significantly optimizes performance for quantized Mixture of
Experts (MoE) layers by changing the order of quantization and
communication operations.
In the previous implementation, the `all2all` operation was performed on
unquantized `hidden_states` (in FP16/BF16) *before* quantization,
resulting in substantial communication overhead. By performing
quantization on each EP rank **first** and then sending the much smaller
quantized data, we reduce the communication volume by nearly 50%.
Additionally, this PR includes a minor optimization to cast `int` inputs
to `float` for the `argsort` operation, forcing it to run on a faster
NPU core instead of the AICPU.
These changes lead to a clear and significant performance gain in MoE
quantization scenarios.
- vLLM version: v0.10.0
- vLLM main:
7175817637
---------
Signed-off-by: SlightwindSec <slightwindsec@gmail.com>
### What this PR does / why we need it?
Adding `W4A8_DYNAMIC` quantization support for linear.
Dense models like Qwen3 can infer with `W4A8_DYNAMIC` quantization.
### Does this PR introduce _any_ user-facing change?
None
### How was this patch tested?
Adding ut case in `tests/ut/quantization/test_w4a8_dynamic.py`
Adding e2e case in
`tests/e2e/multicard/test_offline_inference_distributed.py::test_models_distributed_Qwen3_W4A8DYNAMIC`
to test qwen3 w4a8_dynamic quantized model
Note the w4a8_dynamic quantized model is quantized by `msit/msmodelslim`
of commit `d0abb0a47e1f1a473b866ad41b737fbc28fb1409`
1. Generate `W4A8_DYNAMIC` quantization weights using `msmodelslim`
```shell
git clone https://gitee.com/ascend/msit.git
cd msit/msmodelslim
git checkout d0abb0a47e1f1a473b866ad41b737fbc28fb1409
bash install.sh
```
2. Serve model using `vllm`
```shell
VLLM_USE_V1=1 python -m vllm.entrypoints.openai.api_server \
--model vllm-ascend/Qwen3-8B-W4A8 \
--port 8000 \
--quantization ascend \
--tensor_parallel_size 2 \
--enforce-eager
```
- vLLM version: v0.10.0
- vLLM main:
4cd7fe6cea
---------
Signed-off-by: ZhouXiang <zhouxiang100@huawei.com>
This PR designs the shared expert multi-stream parallelism of
w8a8-dynamic-quantized MoE stage in more detail to achieve better
performance.
- vLLM version: v0.10.0
- vLLM main:
2cc571199b
Signed-off-by: whx-sjtu <2952154980@qq.com>
### What this PR does / why we need it?
A refactoring of forward_context and model_runner_v1, add some context
which is necessary in model inference into forward_context, and refactor
dummy_run logic, make it more reasonable.
Some details for this PR:
Add `ascend_forward_context`;
Update mc2_v2 op, and support `active_mask` param;
Update scripts in examples dir;
refactor `dummy_run` logic;
Add soc_version for A2 and A3;
### Does this PR introduce _any_ user-facing change?
No change at user-facing.
### How was this patch tested?
- vLLM version: v0.10.0
- vLLM main:
57c22e57f9
Signed-off-by: zzzzwwjj <1183291235@qq.com>
There is a lot torchair specified logic in common code. It results hard
code maintenance. We will create a new torchair module to launch
torchair related logic there. I plan to add 4 PR.
1. Refactor worker
2. Refactor utils (this PR)
- simple change that move all torchair related util function to torchair
module
3. Refactor model_runner
4. Refactor attention
- vLLM version: v0.9.2
- vLLM main:
8188196a1c
Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>
### What this PR does / why we need it?
Remove ETP/EP maintained in branch main. We drop this as there is no
relevant scenarios to use ETP now, and we may subsequently advocate
implementing expert tensor parallelism in vLLM to support scenarios
where the expert is needed to be sliced
This is a part of #1422 backport.
Fixes https://github.com/vllm-project/vllm-ascend/issues/1396https://github.com/vllm-project/vllm-ascend/issues/1154
### Does this PR introduce _any_ user-facing change?
We'll not maintain etp/ep in vllm-ascend anymore, and use the tp/ep in
vllm instead.
### How was this patch tested?
CI passed with new added and existing test.
- vLLM version: v0.9.2
- vLLM main:
fe8a2c544a
Signed-off-by: MengqingCao <cmq0113@163.com>
### What this PR does / why we need it?
perf: use multicast to avoid padding decode request to prefill size
### How was this patch tested?
- vLLM version: v0.9.1
- vLLM main:
1fd471e957
Signed-off-by: boying <897013703@qq.com>
### What this PR does / why we need it?
support fused_moe_allgather_ep
### How was this patch tested?
It was tested by UT.
Signed-off-by: lyj-jjj <liuyingjun5@huawei.com>
<!-- Thanks for sending a pull request!
BEFORE SUBMITTING, PLEASE READ
https://docs.vllm.ai/en/latest/contributing/overview.html
-->
### What this PR does / why we need it?
<!--
- Please clarify what changes you are proposing. The purpose of this
section is to outline the changes and how this PR fixes the issue.
If possible, please consider writing useful notes for better and faster
reviews in your PR.
- Please clarify why the changes are needed. For instance, the use case
and bug description.
- Fixes #
-->
1.add static EPLB unit test
2.fix bug: Tensor cannot be directly judged by if statements
### Does this PR introduce _any_ user-facing change?
<!--
Note that it means *any* user-facing change including all aspects such
as API, interface or other behavior changes.
Documentation-only updates are not considered user-facing changes.
-->
### How was this patch tested?
<!--
CI passed with new added/existing test.
If it was tested in a way different from regular unit tests, please
clarify how you tested step by step, ideally copy and paste-able, so
that other reviewers can test and check, and descendants can verify in
the future.
If tests were not added, please describe why they were not added and/or
why it was difficult to add.
-->
Run the unit test.
---------
Signed-off-by: songshanhu07 <1763685535@qq.com>
<!-- Thanks for sending a pull request!
BEFORE SUBMITTING, PLEASE READ
https://docs.vllm.ai/en/latest/contributing/overview.html
-->
### What this PR does / why we need it?
This PR is used for resolved [issue
1147](https://github.com/vllm-project/vllm-ascend/issues/1147)
1. Move fused_moe code into one file `fused_moe.py`.
2. Integrate branch conditions into function `get_fused_moe_state`.
<!--
- Please clarify what changes you are proposing. The purpose of this
section is to outline the changes and how this PR fixes the issue.
If possible, please consider writing useful notes for better and faster
reviews in your PR.
- Please clarify why the changes are needed. For instance, the use case
and bug description.
- Fixes #
-->
### Does this PR introduce _any_ user-facing change?
1. This PR has removed the env `VLLM_ENABLE_MC2`, because I think this
env is useless, we can make judgments based on the current scenario
without this env, it will only increase complexity.
2. This PR has removed the env `USING_LCCL_COM`, because this env has
already expired.
3. `additional_config.expert_tensor_parallel_size` has already expired,
and now we also use parameter `enable_expert_parallel`, consistent with
the vLLM.
<!--
Note that it means *any* user-facing change including all aspects such
as API, interface or other behavior changes.
Documentation-only updates are not considered user-facing changes.
-->
### How was this patch tested?
<!--
CI passed with new added/existing test.
If it was tested in a way different from regular unit tests, please
clarify how you tested step by step, ideally copy and paste-able, so
that other reviewers can test and check, and descendants can verify in
the future.
If tests were not added, please describe why they were not added and/or
why it was difficult to add.
-->
Signed-off-by: zzzzwwjj <1183291235@qq.com>
Contains on #1111 for completeness.
<!-- Thanks for sending a pull request!
BEFORE SUBMITTING, PLEASE READ
https://docs.vllm.ai/en/latest/contributing/overview.html
-->
### What this PR does / why we need it?
Implement multi-stream parallelism for MoE layers with shared experts,
where computation of shared experts will be overlapped with expert token
dispatch and combine. Also, when multi-stream is enabled, weights of
shared experts will be force to replicate across all cards, regardless
of any tensor parallelism configurations, to avoid AllReduce operations.
With the expected overlaping being:
```
| shared gate_up | shared act | | shared down |
| dispatch | routed gate_up, act, down | combine |
```
<!--
- Please clarify what changes you are proposing. The purpose of this
section is to outline the changes and how this PR fixes the issue.
If possible, please consider writing useful notes for better and faster
reviews in your PR.
- Please clarify why the changes are needed. For instance, the use case
and bug description.
- Fixes #
-->
### Does this PR introduce _any_ user-facing change?
No.
<!--
Note that it means *any* user-facing change including all aspects such
as API, interface or other behavior changes.
Documentation-only updates are not considered user-facing changes.
-->
### How was this patch tested?
Tested on 1x16 910 node, with tailored 2 layer DSKv2.
<!--
CI passed with new added/existing test.
If it was tested in a way different from regular unit tests, please
clarify how you tested step by step, ideally copy and paste-able, so
that other reviewers can test and check, and descendants can verify in
the future.
If tests were not added, please describe why they were not added and/or
why it was difficult to add.
-->
---------
Signed-off-by: sdmyzlp <lrwei2@petalmail.com>
### What this PR does / why we need it?
Fix incompatibility problem for non-EPLB scenarios in #1116
### Does this PR introduce _any_ user-facing change?
No.
### How was this patch tested?
Tested with online serving and e2e CI.
Signed-off-by: linfeng-yuan <1102311262@qq.com>
### What this PR does / why we need it?
Add EPLB expert map import capabilities
### Does this PR introduce _any_ user-facing change?
When importing the EPLB expert map you need import expert map file by
vllm args additional_config
### How was this patch tested?
1.You need to collect expert hotness and generate an expert placement
file based on the hotness and the EPLB algorithm, or you can directly
use an existing expert placement table.
2.When launching vLLM, enable EC2 and pass the configuration via the
command-line argument:
--additional-config '{"expert_map_path": "/xxx/xxx/xx.json"}
Co-authored-by: songshanhu07 <1763685535@qq.com>
---------
Signed-off-by: songshanhu07 <1763685535@qq.com>
Signed-off-by: Yuxiao-Xu <664988918@qq.com>
Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>
Co-authored-by: songshanhu07 <1763685535@qq.com>
Co-authored-by: Xu Yuxiao <xuyuxiao2@huawei.com>
Co-authored-by: wangxiyuan <wangxiyuan1007@gmail.com>
### What this PR does / why we need it?
Support MOE inner Multi-stream for Deepseek.
This feature requires graph mode with mc2 enabled.
---------
Signed-off-by: David9857 <985700846@qq.com>
More and more config options are added to additional_config. This PR
provide a new AscendConfig to manage these config options by an easier
way to make code cleaner and readable.
This PR also added the `additional_config` doc for users.
Added the test_ascend_config.py to make sure the new AscendConfig works
as expect.
TODO: Add e2e test with torchair and deepseek once the CI resource is
available.
Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>
### What this PR does / why we need it?
1. In previous PRs https://github.com/vllm-project/vllm-ascend/pull/580https://github.com/vllm-project/vllm-ascend/pull/784, I saved GPU memory
by promptly deleting unnecessary tensors. For tensors passed from
upper-layer functions, I used a list container to transfer the parameter
and then popped the tensor from the list within the inner function to
achieve deletion. Recently, I discovered a better implementation in
sglang—the `dispose_tensor` function and I recommend adopting this
approach.
2. Dispose `hidden_states` and `residual` from the previous layer once
they're no longer used.
3. Avoid to generate `self.inputs_embeds` in `ModelRunnerV1` in
non-multimodal scenarios.
With the aforementioned optimizations, using the DeepSeek-R1-W8A8 model
under the conditions of `TP=16` and `max-model-len=32768`, we can save
1.3GB of npu memory.
**Reference**: https://github.com/sgl-project/sglang/pull/6147
### Does this PR introduce _any_ user-facing change?
No
### How was this patch tested?
---------
Signed-off-by: ApsarasX <apsarax@outlook.com>
### What this PR does / why we need it?
This PR fixes two accuracy bugs incurred by PR #819 when running
deepseekv3 series models:
1. #819 adds `all_to_all` communication in quantized cases, but
`all_gather` && `reduce_scatter` are removed in both of quantized and
unquantized cases. When running unquantized deepseekv3 models with
`ep_size == world_size`, the moe modules fail to communicate. Therefore,
this PR adds `all_to_all` communication on unquantized situation to
solve this accuracy issue.
2. Use `ep_size` rather than `dp_size` to decide whether to use
`all_to_all` in moe.
### Does this PR introduce _any_ user-facing change?
No.
### How was this patch tested?
CI passed with new added/existing test.
---------
Signed-off-by: angazenn <zengyanjia@huawei.com>
Co-authored-by: angazenn <zengyanjia@huawei.com>
<!-- Thanks for sending a pull request!
BEFORE SUBMITTING, PLEASE READ
https://docs.vllm.ai/en/latest/contributing/overview.html
-->
### What this PR does / why we need it?
1. This PR introduces native `all_to_all` communication operator to fix
`allgather` bugs when dp_size > 1. Besides, it adds a naive
implementation of force-load-balance when doing profile runs.
2. The operator `npu_dequant_swiglu_quant` only supports input
hidden_states with dtype `torch.int32`. This tensor occupies space of
`global_bs * seq_len * topk * hidden_size`, which might be very large as
`ep_size` grows. Therefore we need to disable this operator and use
original `swiglu` && `quantize`.
### Does this PR introduce _any_ user-facing change?
No.
### How was this patch tested?
By performing offline inference:
---------
Signed-off-by: angazenn <zengyanjia@huawei.com>
Co-authored-by: angazenn <zengyanjia@huawei.com>
### What this PR does / why we need it?
In the w8a8 quantization code of `fused_experts`, the output of almost
every operator is assigned a new variable name. If we want to save NPU
memory, we manually `del` these variables to end their lifecycle, which
fills the code with `del` statements and looks inelegant.
Therefore, I plan to names the output of most operators as
`hidden_states`, thereby ending the lifecycle of the previous
`hidden_states`.
### Does this PR introduce _any_ user-facing change?
No
### How was this patch tested?
Signed-off-by: ApsarasX <apsarax@outlook.com>
### What this PR does / why we need it?
The root cause of the bug is that numerical computations involving NaNs
cannot eliminate them. We addressed it by using `masked_fill_` to
eliminate NaNs while avoiding memory-wasting `torch.where` approach.
### Does this PR introduce _any_ user-facing change?
No
### How was this patch tested?
This patch was tested with vllm v0.8.5 and vllm-ascend master. I run
deepseek_v3 model with offline inference scripts
(examples/dp_offline/run_dp.sh & data_parallel.py).
Signed-off-by: linfeng-yuan <1102311262@qq.com>
### What this PR does / why we need it?
Optimize NPU memory usage.
https://github.com/vllm-project/vllm-ascend/issues/723
vllm v0.8.4.rc2 and DeepSeek R1 can only support a model length of 16K.
When attempting to run with a model length of 32K, an "Out of Memory"
(OOM) error will occur.
### Does this PR introduce _any_ user-facing change?
No
### How was this patch tested?
CI passed
Signed-off-by: sunbaosong <13793883820@163.com>
-->
### What this PR does / why we need it?
<!--
- Please clarify what changes you are proposing. The purpose of this
section is to outline the changes and how this PR fixes the issue.
If possible, please consider writing useful notes for better and faster
reviews in your PR.
- Please clarify why the changes are needed. For instance, the use case
and bug description.
- Fixes #
-->
1. Improve inference speed and usability for deepsek models with NPU
graph mode.
2. Modify some codes to adapt to CANN 8.1.RC1.beta1.
3. Add a switch for NPU graph mode and its cache.
### Does this PR introduce _any_ user-facing change?
<!--
Note that it means *any* user-facing change including all aspects such
as API, interface or other behavior changes.
Documentation-only updates are not considered user-facing changes.
-->
This PR provides an experimental configuration to enable NPU graph mode
for Deepseek models. User can set
additional_config={'enable_graph_mode': True} to try this feature. Note
that this feature currently only supports for V0 engine.
### How was this patch tested?
<!--
CI passed with new added/existing test.
If it was tested in a way different from regular unit tests, please
clarify how you tested step by step, ideally copy and paste-able, so
that other reviewers can test and check, and descendants can verify in
the future.
If tests were not added, please describe why they were not added and/or
why it was difficult to add.
-->
This patch was tested with the newest torch_npu 2.5.1
(https://pypi.org/project/torch-npu/#files) and CANN 8.1.RC1.beta1
toolkit&nnal&kernels
(https://www.hiascend.com/developer/download/community/result?module=cann)
released in 25/30 April.
Signed-off-by: linfeng-yuan <1102311262@qq.com>
### What this PR does / why we need it?
1. support deepseek with w8a8 quant;
2. support deepseek with mix-parallel(multi-DP, EP+TP);
3. support deepseek with graphmode.
---------
Signed-off-by: wen-jie666 <wenjie39@huawei.com>
Signed-off-by: Yizhou Liu <liuyizhou5@h-partners.com>
Signed-off-by: libaokui <libaokui@huawei.com>
Signed-off-by: linfeng-yuan <1102311262@qq.com>
Co-authored-by: wen-jie666 <wenjie39@huawei.com>
### What this PR does / why we need it?
Add a `VLLMAscendQuantizer` to support w8a8 static (W8A8) and dynamic on
linear and moe (W8A8_DYNAMIC), the quantizer will be enable if a model
has [quantize
filed](https://huggingface.co/vllm-ascend/Qwen2.5-0.5B-Instruct-w8a8/blob/main/config.json#L27).
If MindIE Turbo is installed, the MindIE Turbo Quantizer will apply,
otherwise will use VLLMAscendQuantizer directly.
- This patch fix installation docs to make installation work
- This patch enable norm quantization by patch `RMSNorm.__init__`,
`RMSNorm.forward_oot`, `NPUModelRunnerBase.load_model`
- Add `AscendW8A8LinearMethod` for W8A8
- Add `AscendW8A8DynamicLinearMethod` and
`AscendW8A8DynamicFusedMoEMethod` for W8A8_DYNAMIC
- Add a e2e test for `vllm-ascend/Qwen2.5-0.5B-Instruct-w8a8`
### Does this PR introduce _any_ user-facing change?
Yes, support w8a8 quantization. After this patch supported, users can
use below commands to run w8a8 models:
```
vllm serve /root/.cache/modelscope/hub/Qwen/Qwen2.5-7B-Instruct-w8a8 --served-model-name "qwen2.5-7B"
```
### How was this patch tested?
0. CI passed: add e2e test for `vllm-ascend/Qwen2.5-0.5B-Instruct-w8a8`
1. From @Yikun:
I test Qwen2.5-0.5B-Instruct-w8a8 for functional test all is well, pls
refer to
https://github.com/vllm-project/vllm-ascend/pull/580#issuecomment-2816747613
2. From @dingdingchaomian :
Use qwen2.5-72b-instruct model and deepseek-v2-lite-chat tested, both
models were quantized using Ascend's msmodelslim tool:
- Qwen2.5-72b-instruct were tested twice, one for w8a8 static and one
for w8a8 dynamic.
- Deepseek-v2-lite-chat were tested once because its quantization used
both static and dynamic w8a8.
Models were tested using both off line inference and online serving, and
both work well. The inference codes are exactly the same with the
examples in
https://vllm-ascend.readthedocs.io/en/latest/quick_start.html, with
model path and tensor parallel number changed.
---------
Signed-off-by: dingdingchaomian <wangce21@huawei.com>
Signed-off-by: Yikun Jiang <yikunkero@gmail.com>
Co-authored-by: dingdingchaomian <wangce21@huawei.com>
Co-authored-by: Angazenn <zengyanjia@huawei.com>
Co-authored-by: liujiaxu <liujiaxu4@huawei.com>
Co-authored-by: ApsarasX <apsarax@outlook.com>
Co-authored-by: ganyi1996ppo <pleaplusone.gy@gmail.com>
