feat: support data parallel for deepseek #1012
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
NeverRaR
force-pushed
the
dev/data_parallel
branch
from
c6b1878 to
114f072
Compare
github-actions
bot
added
module:tests
module:ops
module:core
module:quantization
labels
NeverRaR
force-pushed
the
dev/data_parallel
branch
from
114f072 to
76c143f
Compare
|
Please do a rebase |
NeverRaR
force-pushed
the
dev/data_parallel
branch
from
76c143f to
76b0b46
Compare
NeverRaR
force-pushed
the
dev/data_parallel
branch
2 times, most recently
from
650fe8f to
7c57994
Compare
NeverRaR
force-pushed
the
dev/data_parallel
branch
3 times, most recently
from
7229ab6 to
b22de07
Compare
Signed-off-by: boying <897013703@qq.com>
76 tasks
wangxiyuan
approved these changes
Jun 4, 2025
Yikun
approved these changes
Jun 4, 2025
|
I confirmed with @ganyi1996ppo offline, it's OK to merge. |
weijinqian0
pushed a commit
to weijinqian0/vllm-ascend
that referenced
this pull request
Jun 4, 2025
### What this PR does / why we need it?
feat: support data parallel for deepseek
### Does this PR introduce _any_ user-facing change?
Yes, support dp for deepseek
### How was this patch tested?
```
export VLLM_ENABLE_MC2=0
export VLLM_USE_V1=1
export TASK_QUEUE_ENABLE=1
source /usr/local/Ascend/ascend-toolkit/set_env.sh
source /usr/local/Ascend/nnal/atb/set_env.sh
nohup python -m vllm.entrypoints.openai.api_server
--model=/path/to/DeepSeek-R1-W8A8 \
--quantization ascend \
--served-model-name auto \
--trust-remote-code \
--distributed-executor-backend=mp \
--port 8006 \
-tp=8 \
-dp=2 \
--max-num-seqs 24 \
--max-model-len 4096 \
--max-num-batched-tokens 4096 \
--block-size 128 \
-O 0 \
--no-enable-prefix-caching \
--additional-config
'{"torchair_graph_batch_sizes":[24],"expert_tensor_parallel_size":16,"ascend_scheduler_config":{},"enable_graph_mode":true}'
\
--gpu-memory-utilization 0.95 &> run.log &
disown
```
Signed-off-by: boying <897013703@qq.com>
momo609
pushed a commit
to momo609/vllm-ascend
that referenced
this pull request
Jun 4, 2025
### What this PR does / why we need it?
feat: support data parallel for deepseek
### Does this PR introduce _any_ user-facing change?
Yes, support dp for deepseek
### How was this patch tested?
```
export VLLM_ENABLE_MC2=0
export VLLM_USE_V1=1
export TASK_QUEUE_ENABLE=1
source /usr/local/Ascend/ascend-toolkit/set_env.sh
source /usr/local/Ascend/nnal/atb/set_env.sh
nohup python -m vllm.entrypoints.openai.api_server
--model=/path/to/DeepSeek-R1-W8A8 \
--quantization ascend \
--served-model-name auto \
--trust-remote-code \
--distributed-executor-backend=mp \
--port 8006 \
-tp=8 \
-dp=2 \
--max-num-seqs 24 \
--max-model-len 4096 \
--max-num-batched-tokens 4096 \
--block-size 128 \
-O 0 \
--no-enable-prefix-caching \
--additional-config
'{"torchair_graph_batch_sizes":[24],"expert_tensor_parallel_size":16,"ascend_scheduler_config":{},"enable_graph_mode":true}'
\
--gpu-memory-utilization 0.95 &> run.log &
disown
```
Signed-off-by: boying <897013703@qq.com>
Signed-off-by: wangxiaoxin (A) <w00664509@china.huawei.com>
momo609
pushed a commit
to momo609/vllm-ascend
that referenced
this pull request
Jun 4, 2025
### What this PR does / why we need it?
feat: support data parallel for deepseek
### Does this PR introduce _any_ user-facing change?
Yes, support dp for deepseek
### How was this patch tested?
```
export VLLM_ENABLE_MC2=0
export VLLM_USE_V1=1
export TASK_QUEUE_ENABLE=1
source /usr/local/Ascend/ascend-toolkit/set_env.sh
source /usr/local/Ascend/nnal/atb/set_env.sh
nohup python -m vllm.entrypoints.openai.api_server
--model=/path/to/DeepSeek-R1-W8A8 \
--quantization ascend \
--served-model-name auto \
--trust-remote-code \
--distributed-executor-backend=mp \
--port 8006 \
-tp=8 \
-dp=2 \
--max-num-seqs 24 \
--max-model-len 4096 \
--max-num-batched-tokens 4096 \
--block-size 128 \
-O 0 \
--no-enable-prefix-caching \
--additional-config
'{"torchair_graph_batch_sizes":[24],"expert_tensor_parallel_size":16,"ascend_scheduler_config":{},"enable_graph_mode":true}'
\
--gpu-memory-utilization 0.95 &> run.log &
disown
```
Signed-off-by: boying <897013703@qq.com>
Signed-off-by: wangxiaoxin (A) <w00664509@china.huawei.com>
momo609
pushed a commit
to momo609/vllm-ascend
that referenced
this pull request
Jun 5, 2025
### What this PR does / why we need it?
feat: support data parallel for deepseek
### Does this PR introduce _any_ user-facing change?
Yes, support dp for deepseek
### How was this patch tested?
```
export VLLM_ENABLE_MC2=0
export VLLM_USE_V1=1
export TASK_QUEUE_ENABLE=1
source /usr/local/Ascend/ascend-toolkit/set_env.sh
source /usr/local/Ascend/nnal/atb/set_env.sh
nohup python -m vllm.entrypoints.openai.api_server
--model=/path/to/DeepSeek-R1-W8A8 \
--quantization ascend \
--served-model-name auto \
--trust-remote-code \
--distributed-executor-backend=mp \
--port 8006 \
-tp=8 \
-dp=2 \
--max-num-seqs 24 \
--max-model-len 4096 \
--max-num-batched-tokens 4096 \
--block-size 128 \
-O 0 \
--no-enable-prefix-caching \
--additional-config
'{"torchair_graph_batch_sizes":[24],"expert_tensor_parallel_size":16,"ascend_scheduler_config":{},"enable_graph_mode":true}'
\
--gpu-memory-utilization 0.95 &> run.log &
disown
```
Signed-off-by: boying <897013703@qq.com>
Signed-off-by: wangxiaoxin (A) <w00664509@china.huawei.com>
chopper0126
pushed a commit
to chopper0126/vllm-ascend
that referenced
this pull request
Oct 16, 2025
### What this PR does / why we need it?
feat: support data parallel for deepseek
### Does this PR introduce _any_ user-facing change?
Yes, support dp for deepseek
### How was this patch tested?
```
export VLLM_ENABLE_MC2=0
export VLLM_USE_V1=1
export TASK_QUEUE_ENABLE=1
source /usr/local/Ascend/ascend-toolkit/set_env.sh
source /usr/local/Ascend/nnal/atb/set_env.sh
nohup python -m vllm.entrypoints.openai.api_server
--model=/path/to/DeepSeek-R1-W8A8 \
--quantization ascend \
--served-model-name auto \
--trust-remote-code \
--distributed-executor-backend=mp \
--port 8006 \
-tp=8 \
-dp=2 \
--max-num-seqs 24 \
--max-model-len 4096 \
--max-num-batched-tokens 4096 \
--block-size 128 \
-O 0 \
--no-enable-prefix-caching \
--additional-config
'{"torchair_graph_batch_sizes":[24],"expert_tensor_parallel_size":16,"ascend_scheduler_config":{},"enable_graph_mode":true}'
\
--gpu-memory-utilization 0.95 &> run.log &
disown
```
Signed-off-by: boying <897013703@qq.com>
chopper0126
pushed a commit
to chopper0126/vllm-ascend
that referenced
this pull request
Oct 16, 2025
…llm-project#1094) ### What this PR does / why we need it? Add `with_prefill_across_dp` to AscendMetadata to fix dp This pr fixes the bug introduced by vllm-project#1012, which add an arg `with_prefill_across_dp` when dp_size > 1. Signed-off-by: MengqingCao <cmq0113@163.com>
Angazenn
pushed a commit
to Angazenn/vllm-ascend
that referenced
this pull request
Oct 21, 2025
### What this PR does / why we need it?
feat: support data parallel for deepseek
### Does this PR introduce _any_ user-facing change?
Yes, support dp for deepseek
### How was this patch tested?
```
export VLLM_ENABLE_MC2=0
export VLLM_USE_V1=1
export TASK_QUEUE_ENABLE=1
source /usr/local/Ascend/ascend-toolkit/set_env.sh
source /usr/local/Ascend/nnal/atb/set_env.sh
nohup python -m vllm.entrypoints.openai.api_server
--model=/path/to/DeepSeek-R1-W8A8 \
--quantization ascend \
--served-model-name auto \
--trust-remote-code \
--distributed-executor-backend=mp \
--port 8006 \
-tp=8 \
-dp=2 \
--max-num-seqs 24 \
--max-model-len 4096 \
--max-num-batched-tokens 4096 \
--block-size 128 \
-O 0 \
--no-enable-prefix-caching \
--additional-config
'{"torchair_graph_batch_sizes":[24],"expert_tensor_parallel_size":16,"ascend_scheduler_config":{},"enable_graph_mode":true}'
\
--gpu-memory-utilization 0.95 &> run.log &
disown
```
Signed-off-by: boying <897013703@qq.com>
Angazenn
pushed a commit
to Angazenn/vllm-ascend
that referenced
this pull request
Oct 21, 2025
…llm-project#1094) ### What this PR does / why we need it? Add `with_prefill_across_dp` to AscendMetadata to fix dp This pr fixes the bug introduced by vllm-project#1012, which add an arg `with_prefill_across_dp` when dp_size > 1. Signed-off-by: MengqingCao <cmq0113@163.com>
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 this PR does / why we need it?
feat: support data parallel for deepseek
Does this PR introduce any user-facing change?
Yes, support dp for deepseek
How was this patch tested?