### What this PR does / why we need it?
[Feature] Adapt DispathGmmCombineDecode opertor to align with weight
scale dtype of small operators.
- **Before**: weight scale must be float32
- **After**: weight scale can be float32/float16 when x is float16,
float32/bfloat16 when x is float32/bfloat16. And w1 scale can use
different dtype with w2 scale.
More info about this operator, please refer to RFC: issue
https://github.com/vllm-project/vllm-ascend/issues/5476
### Does this PR introduce _any_ user-facing change?
### How was this patch tested?
#### Perf
> When scale is of type fp16 or bf16, it will be cast to fp32 internally
within the operator, while the subsequent computations remain unchanged.
Therefore, this PR will introduce an additional cast operation but halve
the memory copy operations for scale . Furthermore, since the scale data
is only a few KB in size and participates in relatively few
computations, its impact is almost negligible compared to major
operations like matrix multiplication. Thus, the theoretical performance
change should be minimal.
test single operator cases from qwen3-235b,
- single A3 node(ep16), 64 moe experts, 4 experts / die (like qwen3-235b
ep32)
- batch=18/32, token_hidden_size 4096, moe_intermediate_size 1536
The test was conducted for 100 rounds, and the average of the last 95
rounds was taken.
| | bs18(us)| bs32(us)|
| -----| -----| -----|
|Without this PR|96.28|108.83|
|With this PR|96.06|107.90|
Note: Single-operator benchmarks represent an ideal scenario. They are
usually only useful for referencing relative changes and may not fully
align with performance data observed within the full model.
#### Acc
test qwen3-235b eplb on a single A3 node(ep16),
with dispatch_gmm_combine_decode
| dataset | version | metric | mode | vllm-api-stream-chat |
|----- | ----- | ----- | ----- | -----|
| aime2024 | 604a78 | accuracy | gen | 83.33 |
- vLLM version: v0.13.0
- vLLM main:
2f4e6548ef
Signed-off-by: wangqiankun <wangqiankun13@huawei.com>
### What this PR does / why we need it?
This PR refactors `get_kv_cache_spec` method to delegate AttentionSpec
creation to each attention module's own `get_kv_cache_spec()` method,
aligning with the vllm source code structure.
**Changes:**
- Simplify `get_kv_cache_spec` in `model_runner_v1.py` and
`cpu_offload_connector.py`
- Remove manual `AttentionType` checks for `Attention` modules
- Delegate spec creation to each attention module's `get_kv_cache_spec`
method directly
- Let `MambaBase` layers use their own `get_kv_cache_spec` method
- Keep `use_sparse` hack for `MLAAttention` (DeepSeek DSA mode) as
Ascend-specific handling
This change follows RFC #5463 item 12: move AttentionSpec to Attention
module.
- Fixes#5463 (item 12)
### Does this PR introduce _any_ user-facing change?
No. This is an internal refactoring that simplifies code structure
without changing any external behavior.
### How was this patch tested?
- Syntax validation passed via `python -m py_compile`
- CI tests will verify the changes work correctly with existing test
cases
- vLLM version: v0.13.0
- vLLM main:
2f4e6548ef
Signed-off-by: lico67373 <918688502@qq.com>
### What this PR does / why we need it?
1. Rename dynamic_ep to default_eplb.
2. Rename dynamic_ep_v2 to swift_balancer
3. Discard func compose_expert_update_info_bipartite.
- vLLM version: v0.13.0
- vLLM main:
bde38c11df
Signed-off-by: shenchuxiaofugui <1311027364@qq.com>
This PR add `MatmulAllreduceRmsnorm` operator and introduces a graph
fusion pass for `matmul_allreduce_rmsnorm` operations. The
implementation includes a new configuration flag, a pattern matching
pass using `torch._inductor.pattern_matcher`.
Co-authored-by: Trunrain [270250579@qq.com](mailto:270250579@qq.com)
- vLLM version: v0.12.0
- vLLM main:
ad32e3e19c
---------
Signed-off-by: wxsIcey <1790571317@qq.com>
Signed-off-by: tongrunze <t00574058@china.huawei.com>
### What this PR does / why we need it?
Migrate the torch profiler configuration from deprecated environment
variables (`VLLM_TORCH_PROFILER_DIR`, `VLLM_TORCH_PROFILER_WITH_STACK`,
`VLLM_TORCH_PROFILER_WITH_PROFILE_MEMORY`) to the explicit
`ProfilerConfig` object, aligning with vLLM's configuration best
practices.
The profiler environment variable approach is deprecated in vLLM and
will be removed in v0.14.0 or v1.0.0.
### Does this PR introduce _any_ user-facing change?
yes, for deverlopers who want to fetch profiler, he should use `--profiler-config` instead of `VLLM_TORCH_PROFILER_DIR`
### How was this patch tested?
- vLLM version: v0.13.0
- vLLM main:
11b6af5280
Signed-off-by: Meihan-chen <jcccx.cmh@gmail.com>
### What this PR does / why we need it?
1. If the model has dense layers, the current code will attempt to
obtain the routing experts of the dense layers, which will cause an
error. This should be fixed by modifying the code to skip the dense
layers when obtaining the routing experts.
2. The global_expert_map that the function directly outputs a affects
the performance of dsv3.2.
### Does this PR introduce _any_ user-facing change?
### How was this patch tested?
DeepSeek V3.1 conversation is normal.
#### aime precision test (dsv3.1)
baseline without eplb
| dataset | version | metric | mode | vllm-api-general-chat |
|----- | ----- | ----- | ----- | -----|
| aime2024 | 604a78 | accuracy | gen | 66.67 |
eplb
| dataset | version | metric | mode | vllm-api-general-chat |
|----- | ----- | ----- | ----- | -----|
| aime2024 | 604a78 | accuracy | gen | 70.00 |
- vLLM version: v0.13.0
- vLLM main:
11b6af5280
Signed-off-by: shenchuxiaofugui <1311027364@qq.com>
### What this PR does / why we need it?
As mentioned in https://github.com/vllm-project/vllm-ascend/issues/5339,
multi-modal inference on vllm-ascend may lead to OOM issues in some
scenarios.
After our analysis, this is due to the memory fragmentation caused by
frequent dynamic memory size adjustments during runtime. During the
inference, the figure for non-torch memory see a gradual increase from
around 1G to over 5G until the OOM issue occurs.
We find that this problem can be resolved by just directly setting
`PYTORCH_NPU_ALLOC_CONF=expandable_segments:True`. Find more details at
https://docs.vllm.ai/projects/ascend/en/latest/faqs.html#how-to-handle-the-out-of-memory-issue.
Thus, we decide to set this value by default, except RL (sleep mode)
scenarios.
It's also worthy to note that this environment variable may have more
than one key-value pairs. We should append `",expandable_segments:True"`
to the current configs.
For example:
```python
PYTORCH_NPU_ALLOC_CONF = "page_size:1g" + ",expandable_segments:True".
```
> [!NOTE]
> `max_split_size_mb` or `garbage_collection_threshold` cannot be
enabled together with `expandable_segments=True`.
### Does this PR introduce _any_ user-facing change?
Users do not need to set
`PYTORCH_NPU_ALLOC_CONF=expandable_segments:True` manually any more.
### How was this patch tested?
I have build a dataset consisting of my own photographs, which can
stably reproduce this OOM issue on Qwen3-VL serie models.
After apply this PR, this problem has been resolved and the amount of
non-torch memory will keep stable at around 1G throughout the whole
inference.
- vLLM version: v0.13.0
- vLLM main:
2f4e6548ef
---------
Signed-off-by: shen-shanshan <467638484@qq.com>
### What this PR does / why we need it?
The force-free secondary release request causes the node to crash. When
requests are pulled too quickly, they should not be added to the
delay-free queue.
### Does this PR introduce _any_ user-facing change?
No
### How was this patch tested?
By ci
- vLLM version: v0.13.0
- vLLM main:
2c24bc6996
Signed-off-by: wangxiaoteng <wangxiaoteng@huawei.com>
### What this PR does / why we need it?
As issue #5948 reported，when using cpu_offload_connector with TP=1, the
server will hang on starting, we found several bugs here to fix.
1. some crash error encountered because of code changed with vllm
version updating, some of them can be fixed as #5948, and this PR fixed
all of them.
2. hang problem described in #5948, the direct reason is that in
cpu_offload_connector, RPC client using the same client id in scheduler
and worker when tensor_parrallel_size is 1, this PR force the client id
to be different, then it is fixed.
- Why we didn't find this hang problem before?
Because we using --distributed-executor-backend mp or
tensor_parrallel_size > 1 in our test, in our old test case, the
scheduler and workers are different procceses, then client ids build by
`worker-{os.getpid()}` are not the same. But when using
tensor_parrallel_size=1, vllm will use uniproc as
distributed-executor-backend by default, the scheduler and worker will
by in the same proccess, then client ids are the same and hang.
### Does this PR introduce _any_ user-facing change?
### How was this patch tested?
- vLLM version: v0.13.0
- vLLM main:
2c24bc6996
Signed-off-by: lidenghui <lidenghui1110@gmail.com>
### What this PR does / why we need it?
Add basic 310p support. Only dense models work with eager mode now.
- vLLM version: v0.13.0
- vLLM main:
2f4e6548ef
---------
Signed-off-by: Tflowers-0129 <2906339855@qq.com>
Signed-off-by: Shaoxu Cheng <2906339855@qq.com>
### What this PR does / why we need it?
reverts #5761 to fix accuracy issues when using piecewise graph mode.
### Does this PR introduce _any_ user-facing change?
### How was this patch tested?
- vLLM version: v0.13.0
- vLLM main:
2c24bc6996
Signed-off-by: Angazenn <supperccell@163.com>
### What this PR does / why we need it?
Convert `vllm-ascend/compilation` to ruff format.
### Does this PR introduce _any_ user-facing change?
During this migration, we encountered some **errors** in our CI and
testing environments, such as:
```
vllm_ascend/utils.py:653: in <module>
def register_ascend_customop(vllm_config: VllmConfig | None = None):
^^^^^^^^^^^^^^^^^
E TypeError: unsupported operand type(s) for |: 'NoneType' and 'NoneType'
```
**1. Root Cause Analysis:**
The project uses a common pattern to break circular dependencies:
```python
if TYPE_CHECKING:
from vllm.config import VllmConfig
else:
VllmConfig = None # Placeholder assigned at runtime
```
When Python parses the function definition `def
register_ascend_customop(vllm_config: VllmConfig | None)`, it attempts
to evaluate the expression `VllmConfig | None`.
Since `VllmConfig` is assigned `None` at runtime, the expression
effectively becomes `None | None`. In Python, `None` is an instance of
`NoneType`. While the `|` operator is implemented for Type objects
(classes), it is not supported for `NoneType` instances, leading to the
`TypeError` shown above.
**2. Solution:**
To maintain the modern `|` syntax required by our new linting standards
while preserving our dependency management strategy, I have introduced:
```python
from __future__ import annotations
```
at the top of the affected files. This enables **Postponed Evaluation of
Annotations (PEP 563)**.
**3. Impact and Benefits:**
- By enabling `annotations`, Python no longer executes the `VllmConfig |
None` operation during module load. Instead, it stores the annotation as
a string literal, completely avoiding the `None | None` calculation.
- We can keep the `VllmConfig = None` placeholders. This ensures that
other modules can still import these symbols without triggering an
`ImportError`, maintaining a stable dependency graph.
- IDEs and static type checkers (MyPy/Pyright) continue to resolve the
types correctly. This allows us to use modern syntax without sacrificing
type safety or runtime stability.
- The only side effect is that `__annotations__` will now return strings
instead of type objects. Since this module does not use runtime type
enforcement or reflection, this change has zero negative impact on
existing functionality.
### How was this patch tested?
- vLLM version: v0.13.0
- vLLM main:
11b6af5280
---------
Signed-off-by: MrZ20 <2609716663@qq.com>
### What this PR does / why we need it?
This PR fix the input constraints checks for the mlapo and bmm_transpose
operators.
### Does this PR introduce _any_ user-facing change?
No.
### How was this patch tested?
CI passed with new added/existing test.
- vLLM version: v0.13.0
- vLLM main:
2f4e6548ef
### Perf
64K/3K，1P1D，bs=32
before this pr:
TPOT 29ms, TTFT 47s，TPS 606 token/s
after this pr:
TPOT 29ms, TTFT 48s，TPS 636 token/s
Signed-off-by: rjg-lyh <1318825571@qq.com>
### What this PR does / why we need it?
1. Fix DeepSeek-V3.2-W8A8-Pruning mtp
2. Add DeepSeek-V3.2-W8A8-Pruning e2e test
### How was this patch tested?
- vLLM version: v0.13.0
- vLLM main:
11b6af5280
Signed-off-by: hfadzxy <starmoon_zhang@163.com>
### What this PR does / why we need it?
Based on the RFC:https://github.com/vllm-project/vllm-ascend/issues/5604
This PR is a refactoring of vllm_ascend/distributed.
### Does this PR introduce _any_ user-facing change?
NA
### How was this patch tested?
- vLLM version: v0.13.0
- vLLM main:
11b6af5280
Signed-off-by: lty <linhebiwen@gmail.com>
### What this PR does / why we need it?
Fix XliteModelRunner init failed when aclgraph is enabled. Ensure
function graph_capture of vllm.v1.worker.gpu_model_runner is replaced.
- vLLM version: v0.13.0
- vLLM main:
bde38c11df
Signed-off-by: changdawei1 <changdawei3@huawei.com>
### What this PR does / why we need it?
1. Rename num_iterations_eplb_update to expert_heat_collection_interval.
2. Rename num_wait_worker_iterations to algorithm_execution_interval.
3. Rename init_redundancy_expert to num_redundant_experts because the
variable with the same meaning in vLLM is named this way.
4. Delete gate_eplb because we don't need this feature.
5. Move eplb config into a dict in additional config.
6. Depend on pr5817
### Does this PR introduce _any_ user-facing change?
before this pr：
`--additional-config '{"dynamic_eplb":true,
"num_iterations_eplb_update": 4000, "num_wait_worker_iterations": 150,
"init_redundancy_expert": 16, "expert_map_path": "xxx.json"}'`
after this pr:
`--additional-config
'{"eplb_config":{"dynamic_eplb":true,"expert_heat_collection_interval":4000,
"algorithm_execution_interval":150,"num_redundant_experts": 16,
"expert_map_path": "xxx.json"}}'`
### How was this patch tested?
#### test qwen3-235b eplb num_redundant_experts=16
without pr5817
| dataset | version | metric | mode | vllm-api-general-chat |
|----- | ----- | ----- | ----- | -----|
| aime2024 | 604a78 | accuracy | gen | 83.33 |
with pr5817
| dataset | version | metric | mode | vllm-api-general-chat |
|----- | ----- | ----- | ----- | -----|
| aime2024 | 604a78 | accuracy | gen | 86.67 |
- vLLM version: v0.13.0
- vLLM main:
45c1ca1ca1
Signed-off-by: shenchuxiaofugui <1311027364@qq.com>
### What this PR does / why we need it?
This PR aims to extract common methods from eagle_proposer and
mtp_proposer. This is a small step towards merging eagle and mtp.
### Does this PR introduce _any_ user-facing change?
N/A
### How was this patch tested?
by ci
- vLLM version: v0.13.0
- vLLM main:
bde38c11df
---------
Signed-off-by: Zetong Li <slippersss@126.com>
### What this PR does / why we need it?
Fix acc bug when enbale dispatch_gmm_combine_decode and eplb.
After eplb, expert table may change, so mapping is needed, while
fused_mc2 miss the mapping.
More info about this operator, please refer to RFC: issue
https://github.com/vllm-project/vllm-ascend/issues/5476
### Does this PR introduce _any_ user-facing change?
No
### How was this patch tested?
without this pr, qwen3-235b eplb with dispatch_gmm_combine_decode get
acc 3.33% on aime2024.
with this pr,
test qwen3-235b eplb on a single A3 node(ep16)
without dispatch_gmm_combine_decode
| dataset | version | metric | mode | vllm-api-stream-chat |
|----- | ----- | ----- | ----- | -----|
| aime2024 | 604a78 | accuracy | gen | 86.67 |
with dispatch_gmm_combine_decode
| dataset | version | metric | mode | vllm-api-stream-chat |
|----- | ----- | ----- | ----- | -----|
| aime2024 | 604a78 | accuracy | gen | 86.67 |
- vLLM version: v0.13.0
- vLLM main:
2f4e6548ef
Signed-off-by: wangqiankun <wangqiankun13@huawei.com>
### What this PR does / why we need it?
we implement set_additional_forward_context in platform, it's necessary
to reuse code of gpu in model runner v2 by inheriting method in gpu
model runer v2. please see model runner v2's plan #5208
### Does this PR introduce _any_ user-facing change?
no
### How was this patch tested?
- vLLM version: v0.13.0
- vLLM main:
2f4e6548ef
---------
Signed-off-by: Ronald1995 <ronaldautomobile@163.com>
### What this PR does / why we need it?
Based on the RFC:https://github.com/vllm-project/vllm-ascend/issues/5604
This PR is a refactoring of vllm_ascend/distributed, moving all
kv_transfer realtaed codes into a dedicated folder, which has already
been done in vLLM
### Does this PR introduce _any_ user-facing change?
NA
### How was this patch tested?
- vLLM version: v0.13.0
- vLLM main:
2f4e6548ef
---------
Signed-off-by: lty <linhebiwen@gmail.com>
### What this PR does / why we need it?
support dsv3.2 enable both mtp and full_decode_only
PR5626 To align with the community, the branch logic was modified.
Previously, dsv32 could not reach inside the branch, and now an
additional unpadded step is required, which causes transformations in
positions and num_input_tokens, leading to changes in the cos and sin
dimensions in sfa_v1.py. This, in turn, causes an illegal shape error
when passed to the operator.
1. The unpadded function is introduced to align with the community， and
in the community the function does not have the parameters
num_input_tokens and positions.
2. The positions are split and num_input_tokens=num_actual_tokens are
used to correspond to the function name unpad, so that the padded
positions and num_input_tokens are not output.
However, in fact, attention_v1 does not use the above two parameters.
This is done because we are concerned that some people might use these
parameters later and encounter shape mismatch issues if they are not
aware of this. Therefore, we have performed the cropping.
From the perspective of the source of acquisition, positions are not
cropped, so there is actually no need to add unpad in this case.
- vLLM version: v0.13.0
- vLLM main:
2f4e6548ef
Signed-off-by: cookieyyds <126683903+cookieyyds@users.noreply.github.com>
### What this PR does / why we need it?
When there is no kv cache in some devices, the `_compute_prefill_context
func` will return `None`, which is unexecpted. This PR replaces None
with full zeros/-inf tensors to avoid TypeError.
### Does this PR introduce _any_ user-facing change?
No
### How was this patch tested?
```bash
pytest tests/e2e/multicard/4-cards/long_sequence/test_chunked_prefill.py -k test_models_chunked_prefill_with_empty_kvcache
```
- vLLM version: v0.13.0
- vLLM main:
2f4e6548ef
---------
Signed-off-by: QiuChunshuo <qiuchunshuo@huawei.com>
### What this PR does / why we need it?
This PR removes the unused `make_attention_mask` function from
`vllm_ascend/worker/v2/attn_utils.py`.
**Why it's dead code:**
- After PR #4870 (attention mask unification refactor), attention mask
generation has been centralized in the `AttentionMaskBuilder` singleton
class
- The mask is now generated directly by metadata builders when needed
(e.g., `AscendAttentionMetadataBuilder`, `AscendMLAMetadataBuilder`)
- The `make_attention_mask` function is no longer called anywhere in the
codebase
- The function's parameters (including `attn_mask` and `spec_attn_mask`)
were also removed from `build_attn_metadata` in the same refactor
**Changes:**
- Remove `make_attention_mask` function (24 lines) from
`vllm_ascend/worker/v2/attn_utils.py`
### Does this PR introduce _any_ user-facing change?
No. This is a code cleanup that removes dead code. No user-facing
behavior changes.
### How was this patch tested?
- Verified that `make_attention_mask` is not called anywhere in the
codebase (via `grep`)
- CI tests pass to ensure no regressions
- The function has been unused since PR #4870 was merged
- vLLM version: v0.13.0
- vLLM main:
2f4e6548ef
Signed-off-by: lico67373 <918688502@qq.com>
Co-authored-by: weijinqian0 <1184188277@qq.com>
### What this PR does / why we need it?
this pr implement eagle spec decoding for model runner v2, please see
RFC https://github.com/vllm-project/vllm-ascend/issues/5208
### Does this PR introduce _any_ user-facing change?
No
### How was this patch tested?
vLLM version: v0.13.0
---------
Signed-off-by: Ronald1995 <ronaldautomobile@163.com>
### What this PR does / why we need it?
While using the LLM Compressor quantization tool from the VLLM community
to generate quantized weights, the VLLM Ascend engine needs to be
adapted to support the compressed tensors quantization format.
1. Support Moe model W8A8 Int8 dynamic weight.
2. Specify W4A16 quantization configuration.
Co-authored-by: menogrey 1299267905@qq.com
Co-authored-by: kunpengW-code 1289706727@qq.com
### Does this PR introduce _any_ user-facing change?
No
- vLLM version: v0.13.0
- vLLM main:
2f4e6548ef
---------
Signed-off-by: LHXuuu <scut_xlh@163.com>
Signed-off-by: menogrey <1299267905@qq.com>
Signed-off-by: Wang Kunpeng <1289706727@qq.com>
Co-authored-by: menogrey <1299267905@qq.com>
Co-authored-by: Wang Kunpeng <1289706727@qq.com>
### What this PR does / why we need it?
Fixed an accuracy problem when using eagle3 with sp.
The problem is described in
https://github.com/vllm-project/vllm-ascend/issues/5825.
It also adds a much more precise way to determine whether drafter should
use `sp` or not.
Also, it changes the `eager` of drafter to be a real `eager` in frontend
to avoid a `fx-graph` problem.
### Does this PR introduce _any_ user-facing change?
N/A
### How was this patch tested?
For simpilicity, we test it as in
https://github.com/vllm-project/vllm-ascend/issues/5825.
And we get the same result of `eagle3` with `sp` disabled.
```text
--------------------------------------------------
total_num_output_tokens: 1000
num_drafts: 437
num_draft_tokens: 1311
num_accepted_tokens: 564
mean acceptance length: 2.29
--------------------------------------------------
acceptance at token 0: 0.62
acceptance at token 1: 0.40
acceptance at token 2: 0.27
acceptance at token 3: 0.00
acceptance at token 4: 0.00
acceptance at token 5: 0.00
```
* vLLM version: v0.13.0
* vLLM main:
2f4e6548ef
Signed-off-by: drslark <slarksblood@qq.com>
### What this PR does / why we need it?
Fix the bug where the P-node's schedule dead after it force-frees a
request due to timeout and then receives the completed kv cache pulled
by the D-node again. By add list to recode all requests.
- vLLM version: release/v0.13.0
- vLLM main:
81786c8774
---------
Signed-off-by: liziyu <liziyu16@huawei.com>
Signed-off-by: wangxiaoteng <wangxiaoteng@huawei.com>
Co-authored-by: wangxiaoteng <wangxiaoteng@huawei.com>
### What this PR does / why we need it?
Fix layerwise connector for decoder tp size > num kv heads. In this case
prefiller should push kv cache to all decoder npu.
- vLLM version: v0.13.0
- vLLM main:
2f4e6548ef
Signed-off-by: liziyu <liziyu16@huawei.com>
### What this PR does / why we need it?
- Use upstream util function (`_pre_process()` and `_post_process()`) to
reduce redundant codes. (Find more details at
https://github.com/vllm-project/vllm/blob/main/vllm/model_executor/layers/rotary_embedding/common.py#L184-L213)
- Merge Q/K split to simplify the logic of calling
`torch_npu.npu_rotary_mul()` for better performance (TPOT has been
reduced by **6.22%**).
### Does this PR introduce _any_ user-facing change?
no.
### How was this patch tested?
#### ✅ Functional test
Launch the server:
```bash
export VLLM_USE_MODELSCOPE=True
vllm serve /root/.cache/modelscope/hub/models/Qwen/Qwen2.5-VL-7B-Instruct \
--dtype bfloat16 \
--limit-mm-per-prompt '{"image": 1}' \
--max-model-len 16384 \
--max-num-batched-tokens 16384
```
Query the server:
```bash
curl -X POST http://localhost:8000/v1/chat/completions \
-H "Content-Type: application/json" \
-d '{
"model": "/root/.cache/modelscope/hub/models/Qwen/Qwen2.5-VL-7B-Instruct",
"messages": [
{"role": "system", "content": "You are a helpful assistant."},
{"role": "user", "content": [
{"type": "image_url", "image_url": {"url": "https://modelscope.oss-cn-beijing.aliyuncs.com/resource/qwen.png"}},
{"type": "text", "text": "What is the text in the illustrate? How does it look?"}
]}
],
"max_tokens": 100
}'
```
Output:
```
{"id":"chatcmpl-b2911ab6989ef098","object":"chat.completion","created":1768202780,"model":"/root/.cache/modelscope/hub/models/Qwen/Qwen2.5-VL-7B-Instruct","choices":[{"index":0,"message":{"role":"assistant","content":"The text in the illustration is \"TONGYI Qwen.\" The word \"TONGYI\" is written in blue, and \"Qwen\" is written in gray. The text appears to be part of a logo or branding design, with \"TONGYI\" being more prominent and \"Qwen\" being slightly smaller and positioned below it. The font style is modern and clean, with \"TONGYI\" having a slightly bolder appearance compared to \"Qwen.\"","refusal":null,"annotations":null,"audio":null,"function_call":null,"tool_calls":[],"reasoning":null,"reasoning_content":null},"logprobs":null,"finish_reason":"length","stop_reason":null,"token_ids":null}],"service_tier":null,"system_fingerprint":null,"usage":{"prompt_tokens":78,"total_tokens":178,"completion_tokens":100,"prompt_tokens_details":null},"prompt_logprobs":null,"prompt_token_ids":null,"kv_transfer_params":null}
```
#### ✅ Benchmark
Run:
```bash
export VLLM_USE_MODELSCOPE=False
export HF_ENDPOINT="https://hf-mirror.com"
vllm bench serve \
--model /root/.cache/modelscope/hub/models/Qwen/Qwen2.5-VL-7B-Instruct \
--backend openai-chat \
--endpoint /v1/chat/completions \
--dataset-name hf \
--hf-split train \
--dataset-path lmarena-ai/vision-arena-bench-v0.1 \
--num-prompts 10 \
--no-stream
```
Before this PR:
```
============ Serving Benchmark Result ============
Successful requests: 10
Failed requests: 0
Benchmark duration (s): 5.96
Total input tokens: 7191
Total generated tokens: 996
Request throughput (req/s): 1.68
Output token throughput (tok/s): 167.05
Peak output token throughput (tok/s): 261.00
Peak concurrent requests: 10.00
Total token throughput (tok/s): 1373.16
---------------Time to First Token----------------
Mean TTFT (ms): 964.43
Median TTFT (ms): 858.48
P99 TTFT (ms): 1691.45
-----Time per Output Token (excl. 1st token)------
Mean TPOT (ms): 63.08
Median TPOT (ms): 40.86
P99 TPOT (ms): 241.30
---------------Inter-token Latency----------------
Mean ITL (ms): 40.16
Median ITL (ms): 33.61
P99 ITL (ms): 250.30
==================================================
```
After this PR:
```
============ Serving Benchmark Result ============
Successful requests: 10
Failed requests: 0
Benchmark duration (s): 5.71
Total input tokens: 7191
Total generated tokens: 996
Request throughput (req/s): 1.75
Output token throughput (tok/s): 174.45
Peak output token throughput (tok/s): 279.00
Peak concurrent requests: 10.00
Total token throughput (tok/s): 1433.95
---------------Time to First Token----------------
Mean TTFT (ms): 992.14
Median TTFT (ms): 938.30
P99 TTFT (ms): 1728.71
-----Time per Output Token (excl. 1st token)------
Mean TPOT (ms): 59.16
Median TPOT (ms): 37.65
P99 TPOT (ms): 234.89
---------------Inter-token Latency----------------
Mean ITL (ms): 36.55
Median ITL (ms): 30.73
P99 ITL (ms): 170.72
==================================================
```
- vLLM version: v0.13.0
- vLLM main:
2f4e6548ef
---------
Signed-off-by: shen-shanshan <467638484@qq.com>
### What this PR does / why we need it?
To support dispatch_ffn_combine ep32 enabled
### Does this PR introduce _any_ user-facing change?
N/A
### How was this patch tested?
Single operator tested
---------
Signed-off-by: lhchg <lhao_cheng@163.com>
come from: https://github.com/vllm-project/vllm-ascend/issues/5463
Reason:
During the iteration process of the hardware version, there may be a
large number of iterations for the operators, which can lead to
short-term compatibility differences. Therefore, an intermediate
adaptation layer is provided to accommodate the short-term differences
in operators.
- vLLM version: v0.13.0
- vLLM main:
2f4e6548ef
---------
Signed-off-by: weijinqian_v1 <weijinqian@huawei.com>
Signed-off-by: weijinqian0 <1184188277@qq.com>
Co-authored-by: weijinqian_v1 <weijinqian@huawei.com>
[Refactor] Modify the binding logic to allocate CPU cores for each NPU
card
### What this PR does / why we need it?
Modify the binding logic to allocate CPU cores for each NPU card based
on NUMA affinity, while isolating acl_thread/release_thread and other
processes to prevent mutual interference.
### Does this PR introduce _any_ user-facing change?
No
### How was this patch tested?
c85cc045f8
Signed-off-by: rowzwel_dx <1392851715@qq.com>
- vLLM version: v0.13.0
- vLLM main:
7157596103
Signed-off-by: Rozwel-dx <1392851715@qq.com>
### What this PR does / why we need it?
According to the official documentation, the parameter
"draft_tensor_parallel_size": 1 is supposed to be applied to the Eagle3
model. However, based on actual debugging, it was found that the number
of tensor parallelisms (tp) of the Eagle model is consistent with that
of the target model. The setting of tp for the draft model did not take
effect as expected.
**Note:** This feature has not been superimposed and tested with `sp`
and `dp`. It will be adapted later
### Does this PR introduce _any_ user-facing change?
No
### How was this patch tested?
```python
from vllm import LLM, SamplingParams
def main():
prompts = [
"The future of AI is",
]
# Create a sampling params object.
sampling_params = SamplingParams(temperature=0.8, top_p=0.95)
# Create an LLM.
llm = LLM(
model="meta-llama/Llama-3.1-8B-Instruct",
tensor_parallel_size=4,
gpu_memory_utilization=0.9,
enforce_eager=True,
speculative_config={
"method": "eagle3",
"model": "yuhuili/EAGLE3-LLaMA3.1-Instruct-8B"
"draft_tensor_parallel_size": 1,
"num_speculative_tokens": 3,
},
)
# Generate texts from the prompts.
outputs = llm.generate(prompts, sampling_params)
print(f"Outputs: {outputs}")
for output in outputs:
prompt = output.prompt
generated_text = output.outputs[0].text
print(f"Prompt: {prompt!r}, Generated text: {generated_text!r}")
```
- vLLM version: v0.13.0
- vLLM main:
45c1ca1ca1Fixesvllm-project/vllm#31345
Signed-off-by: zhaomingyu <zhaomingyu13@h-partners.com>
Co-authored-by: drslark <slarksblood@qq.com>
### What this PR does / why we need it?
this pr support use triton mrope like cuda_forward, which performance is
equal to ascendc ops
this triton ops should use cann 8.5.0
### Does this PR introduce _any_ user-facing change?
### How was this patch tested?
test in qwen3-vl-235b acc textvqa
native 81.82
npu triton 81.58
cuda triton 81.52
- vLLM version: v0.13.0
- vLLM main:
2f4e6548ef
Signed-off-by: shiyuan680 <917935075@qq.com>
### What this PR does / why we need it?
When running memcache in the A2 environment, the logic for registering
memory needs to be added. Additionally, there is a link establishment
conflict between memcache and HCCS during initialization in A2, so the
link should be established in advance.
### Does this PR introduce _any_ user-facing change?
### How was this patch tested?
- vLLM version: v0.13.0
- vLLM main:
7157596103
---------
Signed-off-by: fangjianwei <f30058701@china.huawei.com>
Co-authored-by: fangjianwei <f30058701@china.huawei.com>
### What this PR does / why we need it?
Add docstrings for Metadata and MetadataBuilder classes in the attention
module to improve code readability.
Related to #5463 (Item 11: Add some comments for CommonMetadata and
others)
**Modified files:**
- `vllm_ascend/attention/context_parallel/common_cp.py`: Added comments
for `AscendPCPMetadata`, `CPChunkedContextMetadata`,
`AscendMetadataForPrefill`, `AscendMetadataForDecode`
- `vllm_ascend/attention/utils.py`: Added comments for
`AscendPrefillContextParallelMetadata`
- `vllm_ascend/attention/mla_v1.py`: Added comments for
`ChunkedContextMetadata`, `AscendMLADecodeMetadata`
- `vllm_ascend/attention/attention_v1.py`: Added comments for
`AscendMetadata`, `AscendAttentionMetadataBuilder`
- `vllm_ascend/attention/context_parallel/attention_cp.py`: Added
comments for `AscendAttentionCPMetadataBuilder`
### Does this PR introduce _any_ user-facing change?
No.
### How was this patch tested?
Documentation only, no functional changes.
Signed-off-by: lico67373 <918688502@qq.com>
### What this PR does / why we need it?
This PR addresses an issue in piecewise graph mode when Multi-Threading
Parallelism (MTP) is enabled. Specifically, the original dummy run
sequence performs the following steps in order:
1. Sync DP (input length = 1 + k)
2. Dispatch (input length = 1 + k, with padding==graph size)
However, in the model execution phase, the sequence differs, resulting
in:
1. Padding (input length = 1, with padding)
2. Sync DP (input length = 1 + k)
3. Dispatch (input length 1 + k != graph size 1 + k, with padding)
This discrepancy leads to a mismatch between the input sizes used in the
model execution and those expected by the dispatch graph, causing an
inconsistency in graph size.
This PR ensures that the dispatch graph size aligns correctly by
modifying the sequence of operations during model execution to match the
dummy run sequence, resolving the mismatch issue.
### Does this PR introduce _any_ user-facing change?
No
### How was this patch tested?
- vLLM version: v0.13.0
- vLLM main:
2f4e6548ef
Signed-off-by: LiuYi-UP <1150854440@qq.com>
### What this PR does / why we need it?
In the chunked prefill scenario, CP needs to align the
`max_context_chunk` to the `cp_virtual_block_size`, but the current
implementation only aligns it to the `block_size`. For
PD-disaggregation, `cp_kv_cache_interleave_size` is typically set equal
to `block_size`, in which case `cp_virtual_block_size=block_size *
dcp_size * pcp_size`. Under specific conditions, this can lead to
misalignment of certain chunks, subsequently triggering assertion check
errors.
### Does this PR introduce _any_ user-facing change?
No
- vLLM version: v0.13.0
- vLLM main:
2f4e6548ef
Signed-off-by: QiuChunshuo <qiuchunshuo@huawei.com>
### What this PR does / why we need it?
when graph mode is picewise，replay by synchronize will be effect
performance, sync almost cost 250us
### Does this PR introduce _any_ user-facing change?
only sync when graph mode contain full mode
### How was this patch tested?
- vLLM version: v0.13.0
- vLLM main:
2f4e6548ef
---------
Signed-off-by: wangyongjun <wangyongjun7@huawei.com>
