profiling.md 11.8 KB
Newer Older
1
2
# Profiling vLLM

3
4
!!! warning
    Profiling is only intended for vLLM developers and maintainers to understand the proportion of time spent in different parts of the codebase. **vLLM end-users should never turn on profiling** as it will significantly slow down the inference.
5

6
7
## Profile with PyTorch Profiler

8
9
We support tracing vLLM workers using the `torch.profiler` module. You can enable the torch profiler by setting `--profiler-config`
when launching the server, and setting the entries `profiler` to `'torch'` and `torch_profiler_dir` to the directory where you want to save the traces. Additionally, you can control the profiling content by specifying the following additional arguments in the config:
10

11
12
13
14
15
16
- `torch_profiler_record_shapes` to enable recording Tensor Shapes, off by default
- `torch_profiler_with_memory` to record memory, off by default
- `torch_profiler_with_stack` to enable recording stack information, on by default
- `torch_profiler_with_flops` to enable recording FLOPs, off by default
- `torch_profiler_use_gzip` to control gzip-compressing profiling files, on by default
- `torch_profiler_dump_cuda_time_total` to control dumping and printing the aggregated CUDA self time table, on by default
17

18
When using `vllm bench serve`, you can enable profiling by passing the `--profile` flag.
19
20
21

Traces can be visualized using <https://ui.perfetto.dev/>.

22
!!! tip
23
    You can directly call bench module without installing vLLM using `python -m vllm.entrypoints.cli.main bench`.
24

25
26
!!! tip
    Only send a few requests through vLLM when profiling, as the traces can get quite large. Also, no need to untar the traces, they can be viewed directly.
27

28
29
30
31
!!! tip
    To stop the profiler - it flushes out all the profile trace files to the directory. This takes time, for example for about 100 requests worth of data for a llama 70b, it takes about 10 minutes to flush out on a H100.
    Set the env variable VLLM_RPC_TIMEOUT to a big number before you start the server. Say something like 30 minutes.
    `export VLLM_RPC_TIMEOUT=1800000`
32

33
### Example commands and usage
34

35
#### Offline Inference
36

37
Refer to [examples/offline_inference/simple_profiling.py](../../examples/offline_inference/simple_profiling.py) for an example.
38

39
#### OpenAI Server
40
41

```bash
42
vllm serve meta-llama/Llama-3.1-8B-Instruct --profiler-config '{"profiler": "torch", "torch_profiler_dir": "./vllm_profile"}'
43
44
```

45
vllm bench command:
46
47

```bash
48
vllm bench serve \
49
    --backend vllm \
50
    --model meta-llama/Llama-3.1-8B-Instruct \
51
52
53
54
    --dataset-name sharegpt \
    --dataset-path sharegpt.json \
    --profile \
    --num-prompts 2
55
```
56

57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
Or use http request:

```shell
# We need first call /start_profile api to start profile.
$ curl -X POST http://localhost:8000/start_profile

# Call model generate.
curl -X POST http://localhost:8000/v1/chat/completions \
    -H "Content-Type: application/json" \
    -d '{
                "model": "meta-llama/Llama-3.1-8B-Instruct",
                "messages": [
                        {
                                "role": "user",
                                "content": "San Francisco is a"
                        }
                ]
    }'

# After need call /stop_profile api to stop profile.
$ curl -X POST http://localhost:8000/stop_profile
```

80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
## Profile with NVIDIA Nsight Systems

Nsight systems is an advanced tool that exposes more profiling details, such as register and shared memory usage, annotated code regions and low-level CUDA APIs and events.

[Install nsight-systems](https://docs.nvidia.com/nsight-systems/InstallationGuide/index.html) using your package manager.
The following block is an example for Ubuntu.

```bash
apt update
apt install -y --no-install-recommends gnupg
echo "deb http://developer.download.nvidia.com/devtools/repos/ubuntu$(source /etc/lsb-release; echo "$DISTRIB_RELEASE" | tr -d .)/$(dpkg --print-architecture) /" | tee /etc/apt/sources.list.d/nvidia-devtools.list
apt-key adv --fetch-keys http://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64/7fa2af80.pub
apt update
apt install nsight-systems-cli
```

96
97
!!! tip
    When profiling with `nsys`, it is advisable to set the environment variable `VLLM_WORKER_MULTIPROC_METHOD=spawn`. The default is to use the `fork` method instead of `spawn`. More information on the topic can be found in the [Nsight Systems release notes](https://docs.nvidia.com/nsight-systems/ReleaseNotes/index.html#general-issues).
98

99
100
101
The Nsight Systems profiler can be launched with `nsys profile ...`, with a few recommended flags for vLLM: `--trace-fork-before-exec=true --cuda-graph-trace=node`.

### Example commands and usage
102

103
104
#### Offline Inference

105
For basic usage, you can just append the profiling command before any existing script you would run for offline inference.
106

107
The following is an example using the `vllm bench latency` script:
108
109

```bash
110
nsys profile  \
111
112
    --trace-fork-before-exec=true \
    --cuda-graph-trace=node \
113
vllm bench latency \
114
115
116
117
118
119
    --model meta-llama/Llama-3.1-8B-Instruct \
    --num-iters-warmup 5 \
    --num-iters 1 \
    --batch-size 16 \
    --input-len 512 \
    --output-len 8
120
121
122
123
```

#### OpenAI Server

124
To profile the server, you will want to prepend your `vllm serve` command with `nsys profile` just like for offline inference, but you will need to specify a few other arguments to enable dynamic capture similarly to the Torch Profiler:
125
126
127

```bash
# server
128
nsys profile \
129
130
    --trace-fork-before-exec=true \
    --cuda-graph-trace=node \
131
132
    --capture-range=cudaProfilerApi \
    --capture-range-end repeat \
133
    vllm serve meta-llama/Llama-3.1-8B-Instruct --profiler-config.profiler cuda
134
135

# client
136
vllm bench serve \
137
138
    --backend vllm \
    --model meta-llama/Llama-3.1-8B-Instruct \
139
140
141
142
    --dataset-name sharegpt \
    --dataset-path sharegpt.json \
    --profile \
    --num-prompts 2
143
144
```

145
With `--profile`, vLLM will capture a profile for each run of `vllm bench serve`. Once the server is killed, the profiles will all be saved.
146
147
148
149
150

#### Analysis

You can view these profiles either as summaries in the CLI, using `nsys stats [profile-file]`, or in the GUI by installing Nsight [locally following the directions here](https://developer.nvidia.com/nsight-systems/get-started).

151
??? console "CLI example"
152
153
154
155
156
157

    ```bash
    nsys stats report1.nsys-rep
    ...
    ** CUDA GPU Kernel Summary (cuda_gpu_kern_sum):

158
    Time (%)  Total Time (ns)  Instances   Avg (ns)     Med (ns)    Min (ns)  Max (ns)   StdDev (ns)                                                  Name
159
160
161
162
163
164
165
166
167
168
    --------  ---------------  ---------  -----------  -----------  --------  ---------  -----------  ----------------------------------------------------------------------------------------------------
        46.3   10,327,352,338     17,505    589,965.9    144,383.0    27,040  3,126,460    944,263.8  sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_of…
        14.8    3,305,114,764      5,152    641,520.7    293,408.0   287,296  2,822,716    867,124.9  sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_of…
        12.1    2,692,284,876     14,280    188,535.4     83,904.0    19,328  2,862,237    497,999.9  sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off…
        9.5    2,116,600,578     33,920     62,399.8     21,504.0    15,326  2,532,285    290,954.1  sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_…
        5.0    1,119,749,165     18,912     59,208.4      9,056.0     6,784  2,578,366    271,581.7  void vllm::act_and_mul_kernel<c10::BFloat16, &vllm::silu_kernel<c10::BFloat16>, (bool)1>(T1 *, cons…
        4.1      916,662,515     21,312     43,011.6     19,776.0     8,928  2,586,205    199,790.1  void cutlass::device_kernel<flash::enable_sm90_or_later<flash::FlashAttnFwdSm90<flash::CollectiveMa…
        2.6      587,283,113     37,824     15,526.7      3,008.0     2,719  2,517,756    139,091.1  std::enable_if<T2>(int)0&&vllm::_typeConvert<T1>::exists, void>::type vllm::fused_add_rms_norm_kern…
        1.9      418,362,605     18,912     22,121.5      3,871.0     3,328  2,523,870    175,248.2  void vllm::rotary_embedding_kernel<c10::BFloat16, (bool)1>(const long *, T1 *, T1 *, const T1 *, in…
        0.7      167,083,069     18,880      8,849.7      2,240.0     1,471  2,499,996    101,436.1  void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0…
169
    ...
170
    ```
171
172
173
174

GUI example:

<img width="1799" alt="Screenshot 2025-03-05 at 11 48 42 AM" src="https://github.com/user-attachments/assets/c7cff1ae-6d6f-477d-a342-bd13c4fc424c" />
175

176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
## Continuous Profiling

There is a [GitHub CI workflow](https://github.com/pytorch/pytorch-integration-testing/actions/workflows/vllm-profiling.yml) in the PyTorch infrastructure repository that provides continuous profiling for different models on vLLM. This automated profiling helps track performance characteristics over time and across different model configurations.

### How It Works

The workflow currently runs weekly profiling sessions for selected models, generating detailed performance traces that can be analyzed using different tools to identify performance regressions or optimization opportunities. But, it can be triggered manually as well, using the Github Action tool.

### Adding New Models

To extend the continuous profiling to additional models, you can modify the [profiling-tests.json](https://github.com/pytorch/pytorch-integration-testing/blob/main/vllm-profiling/cuda/profiling-tests.json) configuration file in the PyTorch integration testing repository. Simply add your model specifications to this file to include them in the automated profiling runs.

### Viewing Profiling Results

The profiling traces generated by the continuous profiling workflow are publicly available on the [vLLM Performance Dashboard](https://hud.pytorch.org/benchmark/llms?repoName=vllm-project%2Fvllm). Look for the **Profiling traces** table to access and download the traces for different models and runs.

192
193
194
195
196
## Profiling vLLM Python Code

The Python standard library includes
[cProfile](https://docs.python.org/3/library/profile.html) for profiling Python
code. vLLM includes a couple of helpers that make it easy to apply it to a section of vLLM.
197
Both the `vllm.utils.profiling.cprofile` and `vllm.utils.profiling.cprofile_context` functions can be
198
199
used to profile a section of code.

200
201
202
203
!!! note
    The legacy import paths `vllm.utils.cprofile` and `vllm.utils.cprofile_context` are deprecated.
    Please use `vllm.utils.profiling.cprofile` and `vllm.utils.profiling.cprofile_context` instead.

204
205
206
207
208
209
210
### Example usage - decorator

The first helper is a Python decorator that can be used to profile a function.
If a filename is specified, the profile will be saved to that file. If no filename is
specified, profile data will be printed to stdout.

```python
211
from vllm.utils.profiling import cprofile
212

213
@cprofile("expensive_function.prof")
214
215
216
217
218
219
220
221
222
223
224
def expensive_function():
    # some expensive code
    pass
```

### Example Usage - context manager

The second helper is a context manager that can be used to profile a block of
code. Similar to the decorator, the filename is optional.

```python
225
from vllm.utils.profiling import cprofile_context
226
227
228
229
230

def another_function():
    # more expensive code
    pass

231
with cprofile_context("another_function.prof"):
232
233
234
235
236
237
238
239
240
241
242
243
    another_function()
```

### Analyzing Profile Results

There are multiple tools available that can help analyze the profile results.
One example is [snakeviz](https://jiffyclub.github.io/snakeviz/).

```bash
pip install snakeviz
snakeviz expensive_function.prof
```
244
245
246
247
248

### Analyzing Garbage Collection Costs

Leverage VLLM_GC_DEBUG environment variable to debug GC costs.

249
- VLLM_GC_DEBUG=1: enable GC debugger with gc.collect elapsed times
250
251
- VLLM_GC_DEBUG='{"top_objects":5}': enable GC debugger to log top 5
  collected objects for each gc.collect