| WARNING: CPU IP/backtrace sampling not supported, disabling. |
| Try the 'nsys status --environment' command to learn more. |
|
|
| WARNING: CPU context switch tracing not supported, disabling. |
| Try the 'nsys status --environment' command to learn more. |
|
|
| INFO 08-13 16:39:02 [__init__.py:235] Automatically detected platform cuda. |
| CUDA_VISIBLE_DEVICES = 3 |
| --- vLLM V1 基准测试（含 NVTX 标记）--- |
| 模型: Qwen/Qwen2-1.5B |
| 批量大小: [1, 2, 4, 8, 16, 32, 64, 128, 256] |
| 场景: ['prefill640_decode1', 'prefill1_decode512'] |
| ------------------------------------------------------------ |
| 加载分词器/模型中... |
| INFO 08-13 16:39:11 [config.py:1604] Using max model len 8192 |
| INFO 08-13 16:39:11 [config.py:2434] Chunked prefill is enabled with max_num_batched_tokens=8192. |
| INFO 08-13 16:39:17 [__init__.py:235] Automatically detected platform cuda. |
| INFO 08-13 16:39:19 [core.py:572] Waiting for init message from front-end. |
| INFO 08-13 16:39:19 [core.py:71] Initializing a V1 LLM engine (v0.10.0) with config: model='Qwen/Qwen2-1.5B', speculative_config=None, tokenizer='Qwen/Qwen2-1.5B', skip_tokenizer_init=False, tokenizer_mode=auto, revision=None, override_neuron_config={}, tokenizer_revision=None, trust_remote_code=True, dtype=torch.bfloat16, max_seq_len=8192, download_dir=None, load_format=LoadFormat.AUTO, tensor_parallel_size=1, pipeline_parallel_size=1, disable_custom_all_reduce=False, quantization=None, enforce_eager=False, kv_cache_dtype=auto, device_config=cuda, decoding_config=DecodingConfig(backend='auto', disable_fallback=False, disable_any_whitespace=False, disable_additional_properties=False, reasoning_backend=''), observability_config=ObservabilityConfig(show_hidden_metrics_for_version=None, otlp_traces_endpoint=None, collect_detailed_traces=None), seed=0, served_model_name=Qwen/Qwen2-1.5B, num_scheduler_steps=1, multi_step_stream_outputs=True, enable_prefix_caching=True, chunked_prefill_enabled=True, use_async_output_proc=True, pooler_config=None, compilation_config={"level":3,"debug_dump_path":"","cache_dir":"","backend":"","custom_ops":[],"splitting_ops":["vllm.unified_attention","vllm.unified_attention_with_output","vllm.mamba_mixer2"],"use_inductor":true,"compile_sizes":[],"inductor_compile_config":{"enable_auto_functionalized_v2":false},"inductor_passes":{},"use_cudagraph":true,"cudagraph_num_of_warmups":1,"cudagraph_capture_sizes":[512,504,496,488,480,472,464,456,448,440,432,424,416,408,400,392,384,376,368,360,352,344,336,328,320,312,304,296,288,280,272,264,256,248,240,232,224,216,208,200,192,184,176,168,160,152,144,136,128,120,112,104,96,88,80,72,64,56,48,40,32,24,16,8,4,2,1],"cudagraph_copy_inputs":false,"full_cuda_graph":false,"max_capture_size":512,"local_cache_dir":null} |
| INFO 08-13 16:39:21 [parallel_state.py:1102] rank 0 in world size 1 is assigned as DP rank 0, PP rank 0, TP rank 0, EP rank 0 |
| WARNING 08-13 16:39:21 [topk_topp_sampler.py:59] FlashInfer is not available. Falling back to the PyTorch-native implementation of top-p & top-k sampling. For the best performance, please install FlashInfer. |
| INFO 08-13 16:39:21 [gpu_model_runner.py:1843] Starting to load model Qwen/Qwen2-1.5B... |
| INFO 08-13 16:39:21 [gpu_model_runner.py:1875] Loading model from scratch... |
| INFO 08-13 16:39:22 [cuda.py:290] Using Flash Attention backend on V1 engine. |
| INFO 08-13 16:39:22 [weight_utils.py:296] Using model weights format ['*.safetensors'] |
| INFO 08-13 16:39:23 [weight_utils.py:349] No model.safetensors.index.json found in remote. |
|
Loading safetensors checkpoint shards: 0% Completed | 0/1 [00:00<?, ?it/s] |
|
Loading safetensors checkpoint shards: 100% Completed | 1/1 [00:00<00:00, 1.75it/s] |
|
Loading safetensors checkpoint shards: 100% Completed | 1/1 [00:00<00:00, 1.75it/s] |
|
|
| INFO 08-13 16:39:23 [default_loader.py:262] Loading weights took 0.65 seconds |
| INFO 08-13 16:39:24 [gpu_model_runner.py:1892] Model loading took 2.9105 GiB and 1.834124 seconds |
| INFO 08-13 16:39:30 [backends.py:530] Using cache directory: /home/cy/.cache/vllm/torch_compile_cache/8e19b7790a/rank_0_0/backbone for vLLM's torch.compile |
| INFO 08-13 16:39:30 [backends.py:541] Dynamo bytecode transform time: 6.19 s |
| INFO 08-13 16:39:35 [backends.py:161] Directly load the compiled graph(s) for dynamic shape from the cache, took 4.361 s |
| INFO 08-13 16:39:36 [monitor.py:34] torch.compile takes 6.19 s in total |
| INFO 08-13 16:39:37 [gpu_worker.py:255] Available KV cache memory: 16.94 GiB |
| INFO 08-13 16:39:37 [kv_cache_utils.py:833] GPU KV cache size: 634,464 tokens |
| INFO 08-13 16:39:37 [kv_cache_utils.py:837] Maximum concurrency for 8,192 tokens per request: 77.45x |
|
Capturing CUDA graph shapes: 0%| | 0/67 [00:00<?, ?it/s]
Capturing CUDA graph shapes: 6%|▌ | 4/67 [00:00<00:01, 33.61it/s]
Capturing CUDA graph shapes: 12%|█▏ | 8/67 [00:00<00:01, 34.83it/s]
Capturing CUDA graph shapes: 18%|█▊ | 12/67 [00:00<00:01, 35.01it/s]
Capturing CUDA graph shapes: 24%|██▍ | 16/67 [00:00<00:01, 35.52it/s]
Capturing CUDA graph shapes: 30%|██▉ | 20/67 [00:00<00:01, 35.95it/s]
Capturing CUDA graph shapes: 36%|███▌ | 24/67 [00:00<00:01, 35.73it/s]
Capturing CUDA graph shapes: 42%|████▏ | 28/67 [00:00<00:01, 36.24it/s]
Capturing CUDA graph shapes: 48%|████▊ | 32/67 [00:00<00:00, 35.94it/s]
Capturing CUDA graph shapes: 54%|█████▎ | 36/67 [00:01<00:00, 34.85it/s]
Capturing CUDA graph shapes: 60%|█████▉ | 40/67 [00:01<00:00, 35.36it/s]
Capturing CUDA graph shapes: 66%|██████▌ | 44/67 [00:01<00:00, 35.23it/s]
Capturing CUDA graph shapes: 72%|███████▏ | 48/67 [00:01<00:00, 35.49it/s]
Capturing CUDA graph shapes: 78%|███████▊ | 52/67 [00:01<00:00, 35.42it/s]
Capturing CUDA graph shapes: 84%|████████▎ | 56/67 [00:01<00:00, 34.72it/s]
Capturing CUDA graph shapes: 90%|████████▉ | 60/67 [00:01<00:00, 34.21it/s]
Capturing CUDA graph shapes: 96%|█████████▌| 64/67 [00:01<00:00, 33.91it/s]
Capturing CUDA graph shapes: 100%|██████████| 67/67 [00:01<00:00, 34.80it/s] |
| INFO 08-13 16:39:39 [gpu_model_runner.py:2485] Graph capturing finished in 2 secs, took 0.49 GiB |
| INFO 08-13 16:39:39 [core.py:193] init engine (profile, create kv cache, warmup model) took 15.55 seconds |
| 模型加载完成。 |
| |
| ===== 场景：prefill640_decode1 | prefill=640, decode=1 ===== |
| |
| --- 批量大小 bs=1 --- |
| 预热中... |
|
Adding requests: 0%| | 0/1 [00:00<?, ?it/s]
Adding requests: 100%|██████████| 1/1 [00:00<00:00, 417.76it/s] |
|
Processed prompts: 0%| | 0/1 [00:00<?, ?it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 38.26it/s, est. speed input: 24537.51 toks/s, output: 38.31 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 38.00it/s, est. speed input: 24537.51 toks/s, output: 38.31 toks/s] |
|
Adding requests: 0%| | 0/1 [00:00<?, ?it/s]
Adding requests: 100%|██████████| 1/1 [00:00<00:00, 581.57it/s] |
|
Processed prompts: 0%| | 0/1 [00:00<?, ?it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 106.10it/s, est. speed input: 68245.15 toks/s, output: 106.40 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 104.38it/s, est. speed input: 68245.15 toks/s, output: 106.40 toks/s] |
| 执行时间: 0.0120 s |
| 实际平均输入 tokens: 640.00（目标 640） |
| 生成总 tokens: 1 |
| 吞吐(生成tokens/秒): 83.48 |
| TTFT (V1 metrics): 0.0111 s |
| 解码吞吐 (V1 metrics): nan tok/s |
| |
| --- 批量大小 bs=2 --- |
| 预热中... |
|
Adding requests: 0%| | 0/1 [00:00<?, ?it/s]
Adding requests: 100%|██████████| 1/1 [00:00<00:00, 480.61it/s] |
|
Processed prompts: 0%| | 0/1 [00:00<?, ?it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 116.04it/s, est. speed input: 74613.07 toks/s, output: 116.34 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 114.15it/s, est. speed input: 74613.07 toks/s, output: 116.34 toks/s] |
|
Adding requests: 0%| | 0/2 [00:00<?, ?it/s]
Adding requests: 100%|██████████| 2/2 [00:00<00:00, 537.28it/s] |
|
Processed prompts: 0%| | 0/2 [00:00<?, ?it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s]
Processed prompts: 100%|██████████| 2/2 [00:00<00:00, 151.35it/s, est. speed input: 97169.45 toks/s, output: 151.62 toks/s]
Processed prompts: 100%|██████████| 2/2 [00:00<00:00, 149.65it/s, est. speed input: 97169.45 toks/s, output: 151.62 toks/s] |
| 执行时间: 0.0177 s |
| 实际平均输入 tokens: 640.00（目标 640） |
| 生成总 tokens: 2 |
| 吞吐(生成tokens/秒): 113.07 |
| TTFT (V1 metrics): 0.0129 s |
| 解码吞吐 (V1 metrics): nan tok/s |
| |
| --- 批量大小 bs=4 --- |
| 预热中... |
|
Adding requests: 0%| | 0/1 [00:00<?, ?it/s]
Adding requests: 100%|██████████| 1/1 [00:00<00:00, 524.09it/s] |
|
Processed prompts: 0%| | 0/1 [00:00<?, ?it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 137.24it/s, est. speed input: 88237.28 toks/s, output: 137.59 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 135.12it/s, est. speed input: 88237.28 toks/s, output: 137.59 toks/s] |
|
Adding requests: 0%| | 0/4 [00:00<?, ?it/s]
Adding requests: 100%|██████████| 4/4 [00:00<00:00, 566.05it/s] |
|
Processed prompts: 0%| | 0/4 [00:00<?, ?it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s]
Processed prompts: 100%|██████████| 4/4 [00:00<00:00, 281.09it/s, est. speed input: 180209.43 toks/s, output: 281.41 toks/s]
Processed prompts: 100%|██████████| 4/4 [00:00<00:00, 278.15it/s, est. speed input: 180209.43 toks/s, output: 281.41 toks/s] |
| 执行时间: 0.0221 s |
| 实际平均输入 tokens: 640.00（目标 640） |
| 生成总 tokens: 4 |
| 吞吐(生成tokens/秒): 181.34 |
| TTFT (V1 metrics): 0.0165 s |
| 解码吞吐 (V1 metrics): nan tok/s |
| |
| --- 批量大小 bs=8 --- |
| 预热中... |
|
Adding requests: 0%| | 0/1 [00:00<?, ?it/s]
Adding requests: 100%|██████████| 1/1 [00:00<00:00, 525.14it/s] |
|
Processed prompts: 0%| | 0/1 [00:00<?, ?it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 112.89it/s, est. speed input: 72573.66 toks/s, output: 113.15 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 111.10it/s, est. speed input: 72573.66 toks/s, output: 113.15 toks/s] |
|
Adding requests: 0%| | 0/8 [00:00<?, ?it/s]
Adding requests: 100%|██████████| 8/8 [00:00<00:00, 571.72it/s] |
|
Processed prompts: 0%| | 0/8 [00:00<?, ?it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s]
Processed prompts: 100%|██████████| 8/8 [00:00<00:00, 460.76it/s, est. speed input: 295349.15 toks/s, output: 461.25 toks/s]
Processed prompts: 100%|██████████| 8/8 [00:00<00:00, 456.72it/s, est. speed input: 295349.15 toks/s, output: 461.25 toks/s] |
| 执行时间: 0.0322 s |
| 实际平均输入 tokens: 640.00（目标 640） |
| 生成总 tokens: 8 |
| 吞吐(生成tokens/秒): 248.79 |
| TTFT (V1 metrics): 0.0185 s |
| 解码吞吐 (V1 metrics): nan tok/s |
| |
| --- 批量大小 bs=16 --- |
| 预热中... |
|
Adding requests: 0%| | 0/1 [00:00<?, ?it/s]
Adding requests: 100%|██████████| 1/1 [00:00<00:00, 497.07it/s] |
|
Processed prompts: 0%| | 0/1 [00:00<?, ?it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 118.83it/s, est. speed input: 76422.91 toks/s, output: 119.15 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 116.85it/s, est. speed input: 76422.91 toks/s, output: 119.15 toks/s] |
|
Adding requests: 0%| | 0/16 [00:00<?, ?it/s]
Adding requests: 100%|██████████| 16/16 [00:00<00:00, 560.18it/s] |
|
Processed prompts: 0%| | 0/16 [00:00<?, ?it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s]
Processed prompts: 100%|██████████| 16/16 [00:00<00:00, 1175.47it/s, est. speed input: 753688.15 toks/s, output: 1176.94 toks/s]
Processed prompts: 100%|██████████| 16/16 [00:00<00:00, 1161.96it/s, est. speed input: 753688.15 toks/s, output: 1176.94 toks/s] |
| 执行时间: 0.0431 s |
| 实际平均输入 tokens: 640.00（目标 640） |
| 生成总 tokens: 16 |
| 吞吐(生成tokens/秒): 371.56 |
| TTFT (V1 metrics): 0.0217 s |
| 解码吞吐 (V1 metrics): nan tok/s |
| |
| --- 批量大小 bs=32 --- |
| 预热中... |
|
Adding requests: 0%| | 0/1 [00:00<?, ?it/s]
Adding requests: 100%|██████████| 1/1 [00:00<00:00, 524.35it/s] |
|
Processed prompts: 0%| | 0/1 [00:00<?, ?it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 114.17it/s, est. speed input: 73393.15 toks/s, output: 114.45 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 112.39it/s, est. speed input: 73393.15 toks/s, output: 114.45 toks/s] |
|
Adding requests: 0%| | 0/32 [00:00<?, ?it/s]
Adding requests: 100%|██████████| 32/32 [00:00<00:00, 573.82it/s] |
|
Processed prompts: 0%| | 0/32 [00:00<?, ?it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s]
Processed prompts: 100%|██████████| 32/32 [00:00<00:00, 2773.32it/s, est. speed input: 1778970.01 toks/s, output: 2777.68 toks/s]
Processed prompts: 100%|██████████| 32/32 [00:00<00:00, 2734.06it/s, est. speed input: 1778970.01 toks/s, output: 2777.68 toks/s] |
| 执行时间: 0.0683 s |
| 实际平均输入 tokens: 640.00（目标 640） |
| 生成总 tokens: 32 |
| 吞吐(生成tokens/秒): 468.80 |
| TTFT (V1 metrics): 0.0317 s |
| 解码吞吐 (V1 metrics): nan tok/s |
| |
| --- 批量大小 bs=64 --- |
| 预热中... |
|
Adding requests: 0%| | 0/1 [00:00<?, ?it/s]
Adding requests: 100%|██████████| 1/1 [00:00<00:00, 505.03it/s] |
|
Processed prompts: 0%| | 0/1 [00:00<?, ?it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 117.61it/s, est. speed input: 75628.40 toks/s, output: 117.92 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 115.72it/s, est. speed input: 75628.40 toks/s, output: 117.92 toks/s] |
|
Adding requests: 0%| | 0/64 [00:00<?, ?it/s]
Adding requests: 91%|█████████ | 58/64 [00:00<00:00, 571.63it/s]
Adding requests: 100%|██████████| 64/64 [00:00<00:00, 571.15it/s] |
|
Processed prompts: 0%| | 0/64 [00:00<?, ?it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s]
Processed prompts: 100%|██████████| 64/64 [00:00<00:00, 4103.45it/s, est. speed input: 2633212.63 toks/s, output: 4109.48 toks/s]
Processed prompts: 100%|██████████| 64/64 [00:00<00:00, 4064.12it/s, est. speed input: 2633212.63 toks/s, output: 4109.48 toks/s] |
| 执行时间: 0.1287 s |
| 实际平均输入 tokens: 640.00（目标 640） |
| 生成总 tokens: 64 |
| 吞吐(生成tokens/秒): 497.23 |
| TTFT (V1 metrics): 0.0598 s |
| 解码吞吐 (V1 metrics): nan tok/s |
| |
| --- 批量大小 bs=128 --- |
| 预热中... |
|
Adding requests: 0%| | 0/1 [00:00<?, ?it/s]
Adding requests: 100%|██████████| 1/1 [00:00<00:00, 518.07it/s] |
|
Processed prompts: 0%| | 0/1 [00:00<?, ?it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 115.68it/s, est. speed input: 74367.09 toks/s, output: 115.96 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 113.89it/s, est. speed input: 74367.09 toks/s, output: 115.96 toks/s] |
|
Adding requests: 0%| | 0/128 [00:00<?, ?it/s]
Adding requests: 45%|████▌ | 58/128 [00:00<00:00, 575.34it/s]
Adding requests: 91%|█████████ | 116/128 [00:00<00:00, 575.18it/s]
Adding requests: 100%|██████████| 128/128 [00:00<00:00, 575.53it/s] |
|
Processed prompts: 0%| | 0/128 [00:00<?, ?it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s]
Processed prompts: 100%|██████████| 128/128 [00:00<00:00, 10790.72it/s, est. speed input: 6920390.41 toks/s, output: 10805.49 toks/s]
Processed prompts: 100%|██████████| 128/128 [00:00<00:00, 10659.60it/s, est. speed input: 6920390.41 toks/s, output: 10805.49 toks/s] |
| 执行时间: 0.2355 s |
| 实际平均输入 tokens: 640.00（目标 640） |
| 生成总 tokens: 128 |
| 吞吐(生成tokens/秒): 543.42 |
| TTFT (V1 metrics): 0.1160 s |
| 解码吞吐 (V1 metrics): nan tok/s |
| |
| --- 批量大小 bs=256 --- |
| 预热中... |
|
Adding requests: 0%| | 0/1 [00:00<?, ?it/s]
Adding requests: 100%|██████████| 1/1 [00:00<00:00, 499.20it/s] |
|
Processed prompts: 0%| | 0/1 [00:00<?, ?it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 113.52it/s, est. speed input: 73240.96 toks/s, output: 113.86 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 111.61it/s, est. speed input: 73240.96 toks/s, output: 113.86 toks/s] |
|
Adding requests: 0%| | 0/256 [00:00<?, ?it/s]
Adding requests: 22%|██▏ | 57/256 [00:00<00:00, 567.10it/s]
Adding requests: 45%|████▍ | 114/256 [00:00<00:00, 567.42it/s]
Adding requests: 67%|██████▋ | 172/256 [00:00<00:00, 569.35it/s]
Adding requests: 89%|████████▉ | 229/256 [00:00<00:00, 562.66it/s]
Adding requests: 100%|██████████| 256/256 [00:00<00:00, 565.30it/s] |
|
Processed prompts: 0%| | 0/256 [00:00<?, ?it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s]
Processed prompts: 100%|██████████| 256/256 [00:00<00:00, 17208.50it/s, est. speed input: 11031298.94 toks/s, output: 17227.00 toks/s]
Processed prompts: 100%|██████████| 256/256 [00:00<00:00, 17028.92it/s, est. speed input: 11031298.94 toks/s, output: 17227.00 toks/s] |
| 执行时间: 0.4695 s |
| 实际平均输入 tokens: 640.00（目标 640） |
| 生成总 tokens: 256 |
| 吞吐(生成tokens/秒): 545.26 |
| TTFT (V1 metrics): 0.2358 s |
| 解码吞吐 (V1 metrics): nan tok/s |
| |
| ===== 场景：prefill1_decode512 | prefill=1, decode=512 ===== |
| |
| --- 批量大小 bs=1 --- |
| 预热中... |
|
Adding requests: 0%| | 0/1 [00:00<?, ?it/s]
Adding requests: 100%|██████████| 1/1 [00:00<00:00, 2880.70it/s] |
|
Processed prompts: 0%| | 0/1 [00:00<?, ?it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:01<00:00, 1.21s/it, est. speed input: 0.82 toks/s, output: 124.43 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:01<00:00, 1.21s/it, est. speed input: 0.82 toks/s, output: 124.43 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:01<00:00, 1.21s/it, est. speed input: 0.82 toks/s, output: 124.43 toks/s] |
|
Adding requests: 0%| | 0/1 [00:00<?, ?it/s]
Adding requests: 100%|██████████| 1/1 [00:00<00:00, 1002.22it/s] |
|
Processed prompts: 0%| | 0/1 [00:00<?, ?it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 1.10it/s, est. speed input: 1.10 toks/s, output: 166.68 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 1.10it/s, est. speed input: 1.10 toks/s, output: 166.68 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 1.10it/s, est. speed input: 1.10 toks/s, output: 166.68 toks/s] |
| 执行时间: 0.9129 s |
| 实际平均输入 tokens: 1.00（目标 1） |
| 生成总 tokens: 151 |
| 吞吐(生成tokens/秒): 165.41 |
| TTFT (V1 metrics): 0.0091 s |
| 解码吞吐 (V1 metrics): 166.96 tok/s |
| |
| --- 批量大小 bs=2 --- |
| 预热中... |
|
Adding requests: 0%| | 0/1 [00:00<?, ?it/s]
Adding requests: 100%|██████████| 1/1 [00:00<00:00, 1034.86it/s] |
|
Processed prompts: 0%| | 0/1 [00:00<?, ?it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 1.08it/s, est. speed input: 1.08 toks/s, output: 163.63 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 1.08it/s, est. speed input: 1.08 toks/s, output: 163.63 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 1.08it/s, est. speed input: 1.08 toks/s, output: 163.63 toks/s] |
|
Adding requests: 0%| | 0/2 [00:00<?, ?it/s]
Adding requests: 100%|██████████| 2/2 [00:00<00:00, 1249.05it/s] |
|
Processed prompts: 0%| | 0/2 [00:00<?, ?it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s]
Processed prompts: 50%|█████ | 1/2 [00:01<00:01, 1.00s/it, est. speed input: 1.00 toks/s, output: 150.94 toks/s]
Processed prompts: 100%|██████████| 2/2 [00:01<00:00, 1.00s/it, est. speed input: 1.99 toks/s, output: 299.87 toks/s]
Processed prompts: 100%|██████████| 2/2 [00:01<00:00, 1.99it/s, est. speed input: 1.99 toks/s, output: 299.87 toks/s] |
| 执行时间: 1.0105 s |
| 实际平均输入 tokens: 1.00（目标 1） |
| 生成总 tokens: 302 |
| 吞吐(生成tokens/秒): 298.86 |
| TTFT (V1 metrics): 0.0118 s |
| 解码吞吐 (V1 metrics): 150.97 tok/s |
| |
| --- 批量大小 bs=4 --- |
| 预热中... |
|
Adding requests: 0%| | 0/1 [00:00<?, ?it/s]
Adding requests: 100%|██████████| 1/1 [00:00<00:00, 1049.63it/s] |
|
Processed prompts: 0%| | 0/1 [00:00<?, ?it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 1.14it/s, est. speed input: 1.14 toks/s, output: 172.12 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 1.14it/s, est. speed input: 1.14 toks/s, output: 172.12 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 1.14it/s, est. speed input: 1.14 toks/s, output: 172.12 toks/s] |
|
Adding requests: 0%| | 0/4 [00:00<?, ?it/s]
Adding requests: 100%|██████████| 4/4 [00:00<00:00, 1518.03it/s] |
|
Processed prompts: 0%| | 0/4 [00:00<?, ?it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s]
Processed prompts: 25%|██▌ | 1/4 [00:00<00:02, 1.01it/s, est. speed input: 1.01 toks/s, output: 152.09 toks/s]
Processed prompts: 100%|██████████| 4/4 [00:01<00:00, 1.01it/s, est. speed input: 4.00 toks/s, output: 603.87 toks/s]
Processed prompts: 100%|██████████| 4/4 [00:01<00:00, 4.00it/s, est. speed input: 4.00 toks/s, output: 603.87 toks/s] |
| 执行时间: 1.0047 s |
| 实际平均输入 tokens: 1.00（目标 1） |
| 生成总 tokens: 604 |
| 吞吐(生成tokens/秒): 601.16 |
| TTFT (V1 metrics): 0.0132 s |
| 解码吞吐 (V1 metrics): 152.00 tok/s |
| |
| --- 批量大小 bs=8 --- |
| 预热中... |
|
Adding requests: 0%| | 0/1 [00:00<?, ?it/s]
Adding requests: 100%|██████████| 1/1 [00:00<00:00, 1030.04it/s] |
|
Processed prompts: 0%| | 0/1 [00:00<?, ?it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 1.12it/s, est. speed input: 1.12 toks/s, output: 169.87 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 1.12it/s, est. speed input: 1.12 toks/s, output: 169.87 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 1.12it/s, est. speed input: 1.12 toks/s, output: 169.87 toks/s] |
|
Adding requests: 0%| | 0/8 [00:00<?, ?it/s]
Adding requests: 100%|██████████| 8/8 [00:00<00:00, 1699.22it/s] |
|
Processed prompts: 0%| | 0/8 [00:00<?, ?it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s]
Processed prompts: 12%|█▎ | 1/8 [00:01<00:07, 1.01s/it, est. speed input: 0.99 toks/s, output: 149.95 toks/s]
Processed prompts: 100%|██████████| 8/8 [00:01<00:00, 1.01s/it, est. speed input: 7.88 toks/s, output: 1189.99 toks/s]
Processed prompts: 100%|██████████| 8/8 [00:01<00:00, 7.88it/s, est. speed input: 7.88 toks/s, output: 1189.99 toks/s] |
| 执行时间: 1.0217 s |
| 实际平均输入 tokens: 1.00（目标 1） |
| 生成总 tokens: 1208 |
| 吞吐(生成tokens/秒): 1182.30 |
| TTFT (V1 metrics): 0.0136 s |
| 解码吞吐 (V1 metrics): 149.60 tok/s |
| |
| --- 批量大小 bs=16 --- |
| 预热中... |
|
Adding requests: 0%| | 0/1 [00:00<?, ?it/s]
Adding requests: 100%|██████████| 1/1 [00:00<00:00, 1038.97it/s] |
|
Processed prompts: 0%| | 0/1 [00:00<?, ?it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 1.13it/s, est. speed input: 1.13 toks/s, output: 171.37 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 1.13it/s, est. speed input: 1.13 toks/s, output: 171.37 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 1.13it/s, est. speed input: 1.13 toks/s, output: 171.37 toks/s] |
|
Adding requests: 0%| | 0/16 [00:00<?, ?it/s]
Adding requests: 100%|██████████| 16/16 [00:00<00:00, 1906.56it/s] |
|
Processed prompts: 0%| | 0/16 [00:00<?, ?it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s]
Processed prompts: 6%|▋ | 1/16 [00:01<00:15, 1.02s/it, est. speed input: 0.98 toks/s, output: 148.27 toks/s]
Processed prompts: 100%|██████████| 16/16 [00:01<00:00, 1.02s/it, est. speed input: 15.57 toks/s, output: 2351.57 toks/s]
Processed prompts: 100%|██████████| 16/16 [00:01<00:00, 15.57it/s, est. speed input: 15.57 toks/s, output: 2351.57 toks/s] |
| 执行时间: 1.0377 s |
| 实际平均输入 tokens: 1.00（目标 1） |
| 生成总 tokens: 2416 |
| 吞吐(生成tokens/秒): 2328.14 |
| TTFT (V1 metrics): 0.0144 s |
| 解码吞吐 (V1 metrics): 147.72 tok/s |
| |
| --- 批量大小 bs=32 --- |
| 预热中... |
|
Adding requests: 0%| | 0/1 [00:00<?, ?it/s]
Adding requests: 100%|██████████| 1/1 [00:00<00:00, 1015.57it/s] |
|
Processed prompts: 0%| | 0/1 [00:00<?, ?it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 1.13it/s, est. speed input: 1.13 toks/s, output: 171.03 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 1.13it/s, est. speed input: 1.13 toks/s, output: 171.03 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 1.13it/s, est. speed input: 1.13 toks/s, output: 171.03 toks/s] |
|
Adding requests: 0%| | 0/32 [00:00<?, ?it/s]
Adding requests: 100%|██████████| 32/32 [00:00<00:00, 1845.12it/s] |
|
Processed prompts: 0%| | 0/32 [00:00<?, ?it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s]
Processed prompts: 3%|▎ | 1/32 [00:01<00:33, 1.09s/it, est. speed input: 0.92 toks/s, output: 138.48 toks/s]
Processed prompts: 100%|██████████| 32/32 [00:01<00:00, 1.09s/it, est. speed input: 28.87 toks/s, output: 4359.02 toks/s]
Processed prompts: 100%|██████████| 32/32 [00:01<00:00, 28.86it/s, est. speed input: 28.87 toks/s, output: 4359.02 toks/s] |
| 执行时间: 1.1279 s |
| 实际平均输入 tokens: 1.00（目标 1） |
| 生成总 tokens: 4832 |
| 吞吐(生成tokens/秒): 4283.95 |
| TTFT (V1 metrics): 0.0161 s |
| 解码吞吐 (V1 metrics): 136.92 tok/s |
| |
| --- 批量大小 bs=64 --- |
| 预热中... |
|
Adding requests: 0%| | 0/1 [00:00<?, ?it/s]
Adding requests: 100%|██████████| 1/1 [00:00<00:00, 1049.89it/s] |
|
Processed prompts: 0%| | 0/1 [00:00<?, ?it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 1.10it/s, est. speed input: 1.10 toks/s, output: 166.46 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 1.10it/s, est. speed input: 1.10 toks/s, output: 166.46 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 1.10it/s, est. speed input: 1.10 toks/s, output: 166.46 toks/s] |
|
Adding requests: 0%| | 0/64 [00:00<?, ?it/s]
Adding requests: 80%|███████▉ | 51/64 [00:00<00:00, 179.15it/s]
Adding requests: 100%|██████████| 64/64 [00:00<00:00, 220.68it/s] |
|
Processed prompts: 0%| | 0/64 [00:00<?, ?it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s]
Processed prompts: 2%|▏ | 1/64 [00:00<00:55, 1.14it/s, est. speed input: 1.14 toks/s, output: 171.61 toks/s]
Processed prompts: 80%|███████▉ | 51/64 [00:01<00:00, 57.75it/s, est. speed input: 44.67 toks/s, output: 6744.26 toks/s]
Processed prompts: 100%|██████████| 64/64 [00:01<00:00, 57.75it/s, est. speed input: 56.00 toks/s, output: 8455.56 toks/s]
Processed prompts: 100%|██████████| 64/64 [00:01<00:00, 55.98it/s, est. speed input: 56.00 toks/s, output: 8455.56 toks/s] |
| 执行时间: 1.4347 s |
| 实际平均输入 tokens: 1.00（目标 1） |
| 生成总 tokens: 9664 |
| 吞吐(生成tokens/秒): 6735.89 |
| TTFT (V1 metrics): 0.2242 s |
| 解码吞吐 (V1 metrics): 130.35 tok/s |
| |
| --- 批量大小 bs=128 --- |
| 预热中... |
|
Adding requests: 0%| | 0/1 [00:00<?, ?it/s]
Adding requests: 100%|██████████| 1/1 [00:00<00:00, 999.83it/s] |
|
Processed prompts: 0%| | 0/1 [00:00<?, ?it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 1.10it/s, est. speed input: 1.10 toks/s, output: 166.41 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 1.10it/s, est. speed input: 1.10 toks/s, output: 166.41 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 1.10it/s, est. speed input: 1.10 toks/s, output: 166.41 toks/s] |
|
Adding requests: 0%| | 0/128 [00:00<?, ?it/s]
Adding requests: 100%|██████████| 128/128 [00:00<00:00, 2066.13it/s] |
|
Processed prompts: 0%| | 0/128 [00:00<?, ?it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s]
Processed prompts: 1%| | 1/128 [00:01<02:36, 1.23s/it, est. speed input: 0.81 toks/s, output: 122.48 toks/s]
Processed prompts: 100%|██████████| 128/128 [00:01<00:00, 1.23s/it, est. speed input: 98.85 toks/s, output: 14926.10 toks/s]
Processed prompts: 100%|██████████| 128/128 [00:01<00:00, 98.82it/s, est. speed input: 98.85 toks/s, output: 14926.10 toks/s] |
| 执行时间: 1.3591 s |
| 实际平均输入 tokens: 1.00（目标 1） |
| 生成总 tokens: 19328 |
| 吞吐(生成tokens/秒): 14221.19 |
| TTFT (V1 metrics): 0.0341 s |
| 解码吞吐 (V1 metrics): 117.37 tok/s |
| |
| --- 批量大小 bs=256 --- |
| 预热中... |
|
Adding requests: 0%| | 0/1 [00:00<?, ?it/s]
Adding requests: 100%|██████████| 1/1 [00:00<00:00, 990.16it/s] |
|
Processed prompts: 0%| | 0/1 [00:00<?, ?it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 1.12it/s, est. speed input: 1.12 toks/s, output: 168.91 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 1.12it/s, est. speed input: 1.12 toks/s, output: 168.91 toks/s]
Processed prompts: 100%|██████████| 1/1 [00:00<00:00, 1.12it/s, est. speed input: 1.12 toks/s, output: 168.91 toks/s] |
|
Adding requests: 0%| | 0/256 [00:00<?, ?it/s]
Adding requests: 94%|█████████▍| 241/256 [00:00<00:00, 2402.35it/s]
Adding requests: 100%|██████████| 256/256 [00:00<00:00, 2452.22it/s] |
|
Processed prompts: 0%| | 0/256 [00:00<?, ?it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s]
Processed prompts: 0%| | 1/256 [00:01<07:15, 1.71s/it, est. speed input: 0.59 toks/s, output: 88.50 toks/s]
Processed prompts: 86%|████████▌ | 219/256 [00:01<00:00, 168.95it/s, est. speed input: 121.24 toks/s, output: 18307.26 toks/s]
Processed prompts: 100%|██████████| 256/256 [00:01<00:00, 168.95it/s, est. speed input: 141.22 toks/s, output: 21324.63 toks/s]
Processed prompts: 100%|██████████| 256/256 [00:01<00:00, 141.19it/s, est. speed input: 141.22 toks/s, output: 21324.63 toks/s] |
| [rank0]:[W813 16:40:02.010999005 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) |
| 执行时间: 1.9196 s |
| 实际平均输入 tokens: 1.00（目标 1） |
| 生成总 tokens: 38656 |
| 吞吐(生成tokens/秒): 20137.93 |
| TTFT (V1 metrics): 0.0514 s |
| 解码吞吐 (V1 metrics): 83.37 tok/s |
| |
| 完成。提示：在 Nsight Systems 中可通过 NVTX 区间快速定位各场景/批量的调用。 |
| GPU 3: General Metrics for NVIDIA AD10x (any frequency) |
| Generating '/tmp/nsys-report-1490.qdstrm' |
|
[1/8] [0% ] test_nvtx.nsys-rep
[1/8] [0% ] test_nvtx.nsys-rep
[1/8] [0% ] test_nvtx.nsys-rep
[1/8] [5% ] test_nvtx.nsys-rep
[1/8] [6% ] test_nvtx.nsys-rep
[1/8] [5% ] test_nvtx.nsys-rep
[1/8] [6% ] test_nvtx.nsys-rep
[1/8] [8% ] test_nvtx.nsys-rep
[1/8] [7% ] test_nvtx.nsys-rep
[1/8] [9% ] test_nvtx.nsys-rep
[1/8] [8% ] test_nvtx.nsys-rep
[1/8] [7% ] test_nvtx.nsys-rep
[1/8] [8% ] test_nvtx.nsys-rep
[1/8] [7% ] test_nvtx.nsys-rep
[1/8] [8% ] test_nvtx.nsys-rep
[1/8] [7% ] test_nvtx.nsys-rep
[1/8] [8% ] test_nvtx.nsys-rep
[1/8] [7% ] test_nvtx.nsys-rep
[1/8] [8% ] test_nvtx.nsys-rep
[1/8] [7% ] test_nvtx.nsys-rep
[1/8] [8% ] test_nvtx.nsys-rep
[1/8] [7% ] test_nvtx.nsys-rep
[1/8] [8% ] test_nvtx.nsys-rep
[1/8] [7% ] test_nvtx.nsys-rep
[1/8] [6% ] test_nvtx.nsys-rep
[1/8] [7% ] test_nvtx.nsys-rep
[1/8] [6% ] test_nvtx.nsys-rep
[1/8] [7% ] test_nvtx.nsys-rep
[1/8] [6% ] test_nvtx.nsys-rep
[1/8] [7% ] test_nvtx.nsys-rep
[1/8] [6% ] test_nvtx.nsys-rep
[1/8] [7% ] test_nvtx.nsys-rep
[1/8] [6% ] test_nvtx.nsys-rep
[1/8] [7% ] test_nvtx.nsys-rep
[1/8] [6% ] test_nvtx.nsys-rep
[1/8] [7% ] test_nvtx.nsys-rep
[1/8] [6% ] test_nvtx.nsys-rep
[1/8] [5% ] test_nvtx.nsys-rep
[1/8] [6% ] test_nvtx.nsys-rep
[1/8] [5% ] test_nvtx.nsys-rep
[1/8] [6% ] test_nvtx.nsys-rep
[1/8] [5% ] test_nvtx.nsys-rep
[1/8] [6% ] test_nvtx.nsys-rep
[1/8] [5% ] test_nvtx.nsys-rep
[1/8] [6% ] test_nvtx.nsys-rep
[1/8] [7% ] test_nvtx.nsys-rep
[1/8] [6% ] test_nvtx.nsys-rep
[1/8] [7% ] test_nvtx.nsys-rep
[1/8] [6% ] test_nvtx.nsys-rep
[1/8] [7% ] test_nvtx.nsys-rep
[1/8] [6% ] test_nvtx.nsys-rep
[1/8] [7% ] test_nvtx.nsys-rep
[1/8] [8% ] test_nvtx.nsys-rep
[1/8] [9% ] test_nvtx.nsys-rep
[1/8] [10% ] test_nvtx.nsys-rep
[1/8] [11% ] test_nvtx.nsys-rep
[1/8] [12% ] test_nvtx.nsys-rep
[1/8] [13% ] test_nvtx.nsys-rep
[1/8] [14% ] test_nvtx.nsys-rep
[1/8] [=15% ] test_nvtx.nsys-rep
[1/8] [=16% ] test_nvtx.nsys-rep
[1/8] [=17% ] test_nvtx.nsys-rep
[1/8] [==18% ] test_nvtx.nsys-rep
[1/8] [==19% ] test_nvtx.nsys-rep
[1/8] [==20% ] test_nvtx.nsys-rep
[1/8] [==21% ] test_nvtx.nsys-rep
[1/8] [===22% ] test_nvtx.nsys-rep
[1/8] [===23% ] test_nvtx.nsys-rep
[1/8] [===24% ] test_nvtx.nsys-rep
[1/8] [====25% ] test_nvtx.nsys-rep
[1/8] [====26% ] test_nvtx.nsys-rep
[1/8] [====27% ] test_nvtx.nsys-rep
[1/8] [====28% ] test_nvtx.nsys-rep
[1/8] [=====29% ] test_nvtx.nsys-rep
[1/8] [=====30% ] test_nvtx.nsys-rep
[1/8] [=====31% ] test_nvtx.nsys-rep
[1/8] [=====32% ] test_nvtx.nsys-rep
[1/8] [======33% ] test_nvtx.nsys-rep
[1/8] [======34% ] test_nvtx.nsys-rep
[1/8] [======35% ] test_nvtx.nsys-rep
[1/8] [=======36% ] test_nvtx.nsys-rep
[1/8] [=======37% ] test_nvtx.nsys-rep
[1/8] [=======38% ] test_nvtx.nsys-rep
[1/8] [=======39% ] test_nvtx.nsys-rep
[1/8] [========40% ] test_nvtx.nsys-rep
[1/8] [========41% ] test_nvtx.nsys-rep
[1/8] [========42% ] test_nvtx.nsys-rep
[1/8] [=========43% ] test_nvtx.nsys-rep
[1/8] [=========44% ] test_nvtx.nsys-rep
[1/8] [=========45% ] test_nvtx.nsys-rep
[1/8] [=========46% ] test_nvtx.nsys-rep
[1/8] [==========47% ] test_nvtx.nsys-rep
[1/8] [==========48% ] test_nvtx.nsys-rep
[1/8] [==========49% ] test_nvtx.nsys-rep
[1/8] [===========50% ] test_nvtx.nsys-rep
[1/8] [===========51% ] test_nvtx.nsys-rep
[1/8] [===========52% ] test_nvtx.nsys-rep
[1/8] [===========53% ] test_nvtx.nsys-rep
[1/8] [============54% ] test_nvtx.nsys-rep
[1/8] [============55% ] test_nvtx.nsys-rep
[1/8] [============56% ] test_nvtx.nsys-rep
[1/8] [============57% ] test_nvtx.nsys-rep
[1/8] [=============58% ] test_nvtx.nsys-rep
[1/8] [=============59% ] test_nvtx.nsys-rep
[1/8] [=============60% ] test_nvtx.nsys-rep
[1/8] [==============61% ] test_nvtx.nsys-rep
[1/8] [==============62% ] test_nvtx.nsys-rep
[1/8] [==============63% ] test_nvtx.nsys-rep
[1/8] [==============64% ] test_nvtx.nsys-rep
[1/8] [===============65% ] test_nvtx.nsys-rep
[1/8] [===============66% ] test_nvtx.nsys-rep
[1/8] [===============67% ] test_nvtx.nsys-rep
[1/8] [================68% ] test_nvtx.nsys-rep
[1/8] [================69% ] test_nvtx.nsys-rep
[1/8] [================70% ] test_nvtx.nsys-rep
[1/8] [================71% ] test_nvtx.nsys-rep
[1/8] [=================72% ] test_nvtx.nsys-rep
[1/8] [=================73% ] test_nvtx.nsys-rep
[1/8] [========================100%] test_nvtx.nsys-rep
[1/8] [========================100%] test_nvtx.nsys-rep |
|
[2/8] [0% ] test_nvtx.sqlite
[2/8] [1% ] test_nvtx.sqlite
[2/8] [2% ] test_nvtx.sqlite
[2/8] [3% ] test_nvtx.sqlite
[2/8] [4% ] test_nvtx.sqlite
[2/8] [5% ] test_nvtx.sqlite
[2/8] [6% ] test_nvtx.sqlite
[2/8] [7% ] test_nvtx.sqlite
[2/8] [8% ] test_nvtx.sqlite
[2/8] [9% ] test_nvtx.sqlite
[2/8] [10% ] test_nvtx.sqlite
[2/8] [11% ] test_nvtx.sqlite
[2/8] [12% ] test_nvtx.sqlite
[2/8] [13% ] test_nvtx.sqlite
[2/8] [14% ] test_nvtx.sqlite
[2/8] [=15% ] test_nvtx.sqlite
[2/8] [=16% ] test_nvtx.sqlite
[2/8] [=17% ] test_nvtx.sqlite
[2/8] [==18% ] test_nvtx.sqlite
[2/8] [==19% ] test_nvtx.sqlite
[2/8] [==20% ] test_nvtx.sqlite
[2/8] [==21% ] test_nvtx.sqlite
[2/8] [===22% ] test_nvtx.sqlite
[2/8] [===23% ] test_nvtx.sqlite
[2/8] [===24% ] test_nvtx.sqlite
[2/8] [====25% ] test_nvtx.sqlite
[2/8] [====26% ] test_nvtx.sqlite
[2/8] [====27% ] test_nvtx.sqlite
[2/8] [====28% ] test_nvtx.sqlite
[2/8] [=====29% ] test_nvtx.sqlite
[2/8] [=====30% ] test_nvtx.sqlite
[2/8] [=====31% ] test_nvtx.sqlite
[2/8] [=====32% ] test_nvtx.sqlite
[2/8] [======33% ] test_nvtx.sqlite
[2/8] [======34% ] test_nvtx.sqlite
[2/8] [======35% ] test_nvtx.sqlite
[2/8] [=======36% ] test_nvtx.sqlite
[2/8] [=======37% ] test_nvtx.sqlite
[2/8] [=======38% ] test_nvtx.sqlite
[2/8] [=======39% ] test_nvtx.sqlite
[2/8] [========40% ] test_nvtx.sqlite
[2/8] [========41% ] test_nvtx.sqlite
[2/8] [========42% ] test_nvtx.sqlite
[2/8] [=========43% ] test_nvtx.sqlite
[2/8] [=========44% ] test_nvtx.sqlite
[2/8] [=========45% ] test_nvtx.sqlite
[2/8] [=========46% ] test_nvtx.sqlite
[2/8] [==========47% ] test_nvtx.sqlite
[2/8] [==========48% ] test_nvtx.sqlite
[2/8] [==========49% ] test_nvtx.sqlite
[2/8] [===========50% ] test_nvtx.sqlite
[2/8] [===========51% ] test_nvtx.sqlite
[2/8] [===========52% ] test_nvtx.sqlite
[2/8] [===========53% ] test_nvtx.sqlite
[2/8] [============54% ] test_nvtx.sqlite
[2/8] [============55% ] test_nvtx.sqlite
[2/8] [============56% ] test_nvtx.sqlite
[2/8] [============57% ] test_nvtx.sqlite
[2/8] [=============58% ] test_nvtx.sqlite
[2/8] [=============59% ] test_nvtx.sqlite
[2/8] [=============60% ] test_nvtx.sqlite
[2/8] [==============61% ] test_nvtx.sqlite
[2/8] [==============62% ] test_nvtx.sqlite
[2/8] [==============63% ] test_nvtx.sqlite
[2/8] [==============64% ] test_nvtx.sqlite
[2/8] [===============65% ] test_nvtx.sqlite
[2/8] [===============66% ] test_nvtx.sqlite
[2/8] [===============67% ] test_nvtx.sqlite
[2/8] [================68% ] test_nvtx.sqlite
[2/8] [================69% ] test_nvtx.sqlite
[2/8] [================70% ] test_nvtx.sqlite
[2/8] [================71% ] test_nvtx.sqlite
[2/8] [=================72% ] test_nvtx.sqlite
[2/8] [=================73% ] test_nvtx.sqlite
[2/8] [=================74% ] test_nvtx.sqlite
[2/8] [==================75% ] test_nvtx.sqlite
[2/8] [==================76% ] test_nvtx.sqlite
[2/8] [==================77% ] test_nvtx.sqlite
[2/8] [==================78% ] test_nvtx.sqlite
[2/8] [===================79% ] test_nvtx.sqlite
[2/8] [===================80% ] test_nvtx.sqlite
[2/8] [===================81% ] test_nvtx.sqlite
[2/8] [===================82% ] test_nvtx.sqlite
[2/8] [====================83% ] test_nvtx.sqlite
[2/8] [====================84% ] test_nvtx.sqlite
[2/8] [====================85% ] test_nvtx.sqlite
[2/8] [=====================86% ] test_nvtx.sqlite
[2/8] [=====================87% ] test_nvtx.sqlite
[2/8] [=====================88% ] test_nvtx.sqlite
[2/8] [=====================89% ] test_nvtx.sqlite
[2/8] [======================90% ] test_nvtx.sqlite
[2/8] [======================91% ] test_nvtx.sqlite
[2/8] [======================92% ] test_nvtx.sqlite
[2/8] [=======================93% ] test_nvtx.sqlite
[2/8] [=======================94% ] test_nvtx.sqlite
[2/8] [=======================95% ] test_nvtx.sqlite
[2/8] [=======================96% ] test_nvtx.sqlite
[2/8] [========================97% ] test_nvtx.sqlite
[2/8] [========================98% ] test_nvtx.sqlite
[2/8] [========================99% ] test_nvtx.sqlite
[2/8] [========================100%] test_nvtx.sqlite
[2/8] [========================100%] test_nvtx.sqlite |
| [3/8] Executing 'nvtx_sum' stats report |
| |
| Time (%) Total Time (ns) Instances Avg (ns) Med (ns) Min (ns) Max (ns) StdDev (ns) Style Range |
| -------- --------------- --------- ---------------- ---------------- -------------- -------------- ----------- ------- ------------------------------------- |
| 52.6 36,466,864,668 1 36,466,864,668.0 36,466,864,668.0 36,466,864,668 36,466,864,668 0.0 PushPop :LLM_init |
| 2.8 1,929,931,509 1 1,929,931,509.0 1,929,931,509.0 1,929,931,509 1,929,931,509 0.0 PushPop :RUN [prefill1_decode512] bs=256 |
| 2.8 1,919,413,195 1 1,919,413,195.0 1,919,413,195.0 1,919,413,195 1,919,413,195 0.0 PushPop :generate [prefill1_decode512] bs=256 |
| 2.1 1,444,940,085 1 1,444,940,085.0 1,444,940,085.0 1,444,940,085 1,444,940,085 0.0 PushPop :RUN [prefill1_decode512] bs=64 |
| 2.1 1,434,566,089 1 1,434,566,089.0 1,434,566,089.0 1,434,566,089 1,434,566,089 0.0 PushPop :generate [prefill1_decode512] bs=64 |
| 2.0 1,369,556,550 1 1,369,556,550.0 1,369,556,550.0 1,369,556,550 1,369,556,550 0.0 PushPop :RUN [prefill1_decode512] bs=128 |
| 2.0 1,358,959,952 1 1,358,959,952.0 1,358,959,952.0 1,358,959,952 1,358,959,952 0.0 PushPop :generate [prefill1_decode512] bs=128 |
| 1.8 1,215,380,245 1 1,215,380,245.0 1,215,380,245.0 1,215,380,245 1,215,380,245 0.0 PushPop :WARMUP [prefill1_decode512] bs=1 |
| 1.6 1,138,055,752 1 1,138,055,752.0 1,138,055,752.0 1,138,055,752 1,138,055,752 0.0 PushPop :RUN [prefill1_decode512] bs=32 |
| 1.6 1,127,795,108 1 1,127,795,108.0 1,127,795,108.0 1,127,795,108 1,127,795,108 0.0 PushPop :generate [prefill1_decode512] bs=32 |
| 1.5 1,047,787,619 1 1,047,787,619.0 1,047,787,619.0 1,047,787,619 1,047,787,619 0.0 PushPop :RUN [prefill1_decode512] bs=16 |
| 1.5 1,037,603,494 1 1,037,603,494.0 1,037,603,494.0 1,037,603,494 1,037,603,494 0.0 PushPop :generate [prefill1_decode512] bs=16 |
| 1.5 1,031,903,705 1 1,031,903,705.0 1,031,903,705.0 1,031,903,705 1,031,903,705 0.0 PushPop :RUN [prefill1_decode512] bs=8 |
| 1.5 1,021,601,603 1 1,021,601,603.0 1,021,601,603.0 1,021,601,603 1,021,601,603 0.0 PushPop :generate [prefill1_decode512] bs=8 |
| 1.5 1,020,765,322 1 1,020,765,322.0 1,020,765,322.0 1,020,765,322 1,020,765,322 0.0 PushPop :RUN [prefill1_decode512] bs=2 |
| 1.5 1,014,932,632 1 1,014,932,632.0 1,014,932,632.0 1,014,932,632 1,014,932,632 0.0 PushPop :RUN [prefill1_decode512] bs=4 |
| 1.5 1,010,379,350 1 1,010,379,350.0 1,010,379,350.0 1,010,379,350 1,010,379,350 0.0 PushPop :generate [prefill1_decode512] bs=2 |
| 1.5 1,004,598,155 1 1,004,598,155.0 1,004,598,155.0 1,004,598,155 1,004,598,155 0.0 PushPop :generate [prefill1_decode512] bs=4 |
| 1.3 925,984,111 1 925,984,111.0 925,984,111.0 925,984,111 925,984,111 0.0 PushPop :WARMUP [prefill1_decode512] bs=2 |
| 1.3 923,349,044 1 923,349,044.0 923,349,044.0 923,349,044 923,349,044 0.0 PushPop :RUN [prefill1_decode512] bs=1 |
| 1.3 912,751,117 1 912,751,117.0 912,751,117.0 912,751,117 912,751,117 0.0 PushPop :generate [prefill1_decode512] bs=1 |
| 1.3 910,760,967 1 910,760,967.0 910,760,967.0 910,760,967 910,760,967 0.0 PushPop :WARMUP [prefill1_decode512] bs=128 |
| 1.3 910,250,406 1 910,250,406.0 910,250,406.0 910,250,406 910,250,406 0.0 PushPop :WARMUP [prefill1_decode512] bs=64 |
| 1.3 897,446,923 1 897,446,923.0 897,446,923.0 897,446,923 897,446,923 0.0 PushPop :WARMUP [prefill1_decode512] bs=256 |
| 1.3 892,030,307 1 892,030,307.0 892,030,307.0 892,030,307 892,030,307 0.0 PushPop :WARMUP [prefill1_decode512] bs=8 |
| 1.3 886,074,412 1 886,074,412.0 886,074,412.0 886,074,412 886,074,412 0.0 PushPop :WARMUP [prefill1_decode512] bs=32 |
| 1.3 884,245,171 1 884,245,171.0 884,245,171.0 884,245,171 884,245,171 0.0 PushPop :WARMUP [prefill1_decode512] bs=16 |
| 1.3 880,382,660 1 880,382,660.0 880,382,660.0 880,382,660 880,382,660 0.0 PushPop :WARMUP [prefill1_decode512] bs=4 |
| 0.7 473,603,532 1 473,603,532.0 473,603,532.0 473,603,532 473,603,532 0.0 PushPop :RUN [prefill640_decode1] bs=256 |
| 0.7 469,420,575 1 469,420,575.0 469,420,575.0 469,420,575 469,420,575 0.0 PushPop :generate [prefill640_decode1] bs=256 |
| 0.7 463,984,907 1 463,984,907.0 463,984,907.0 463,984,907 463,984,907 0.0 PushPop :WARMUP [prefill640_decode1] bs=1 |
| 0.3 239,661,120 1 239,661,120.0 239,661,120.0 239,661,120 239,661,120 0.0 PushPop :RUN [prefill640_decode1] bs=128 |
| 0.3 235,462,753 1 235,462,753.0 235,462,753.0 235,462,753 235,462,753 0.0 PushPop :generate [prefill640_decode1] bs=128 |
| 0.2 132,881,776 1 132,881,776.0 132,881,776.0 132,881,776 132,881,776 0.0 PushPop :RUN [prefill640_decode1] bs=64 |
| 0.2 128,633,915 1 128,633,915.0 128,633,915.0 128,633,915 128,633,915 0.0 PushPop :generate [prefill640_decode1] bs=64 |
| 0.1 72,289,268 1 72,289,268.0 72,289,268.0 72,289,268 72,289,268 0.0 PushPop :RUN [prefill640_decode1] bs=32 |
| 0.1 68,187,202 1 68,187,202.0 68,187,202.0 68,187,202 68,187,202 0.0 PushPop :generate [prefill640_decode1] bs=32 |
| 0.1 47,146,091 1 47,146,091.0 47,146,091.0 47,146,091 47,146,091 0.0 PushPop :RUN [prefill640_decode1] bs=16 |
| 0.1 42,988,572 1 42,988,572.0 42,988,572.0 42,988,572 42,988,572 0.0 PushPop :generate [prefill640_decode1] bs=16 |
| 0.1 36,174,975 1 36,174,975.0 36,174,975.0 36,174,975 36,174,975 0.0 PushPop :RUN [prefill640_decode1] bs=8 |
| 0.0 32,093,917 1 32,093,917.0 32,093,917.0 32,093,917 32,093,917 0.0 PushPop :generate [prefill640_decode1] bs=8 |
| 0.0 26,218,980 1 26,218,980.0 26,218,980.0 26,218,980 26,218,980 0.0 PushPop :RUN [prefill640_decode1] bs=4 |
| 0.0 21,993,697 1 21,993,697.0 21,993,697.0 21,993,697 21,993,697 0.0 PushPop :generate [prefill640_decode1] bs=4 |
| 0.0 21,629,311 1 21,629,311.0 21,629,311.0 21,629,311 21,629,311 0.0 PushPop :RUN [prefill640_decode1] bs=2 |
| 0.0 17,626,968 1 17,626,968.0 17,626,968.0 17,626,968 17,626,968 0.0 PushPop :generate [prefill640_decode1] bs=2 |
| 0.0 15,993,748 1 15,993,748.0 15,993,748.0 15,993,748 15,993,748 0.0 PushPop :RUN [prefill640_decode1] bs=1 |
| 0.0 11,903,648 1 11,903,648.0 11,903,648.0 11,903,648 11,903,648 0.0 PushPop :generate [prefill640_decode1] bs=1 |
| 0.0 11,605,063 1 11,605,063.0 11,605,063.0 11,605,063 11,605,063 0.0 PushPop :WARMUP [prefill640_decode1] bs=256 |
| 0.0 11,512,402 1 11,512,402.0 11,512,402.0 11,512,402 11,512,402 0.0 PushPop :WARMUP [prefill640_decode1] bs=8 |
| 0.0 11,471,499 1 11,471,499.0 11,471,499.0 11,471,499 11,471,499 0.0 PushPop :WARMUP [prefill640_decode1] bs=2 |
| 0.0 11,419,754 1 11,419,754.0 11,419,754.0 11,419,754 11,419,754 0.0 PushPop :WARMUP [prefill640_decode1] bs=32 |
| 0.0 11,332,293 1 11,332,293.0 11,332,293.0 11,332,293 11,332,293 0.0 PushPop :WARMUP [prefill640_decode1] bs=128 |
| 0.0 11,243,804 1 11,243,804.0 11,243,804.0 11,243,804 11,243,804 0.0 PushPop :WARMUP [prefill640_decode1] bs=64 |
| 0.0 11,202,965 1 11,202,965.0 11,202,965.0 11,202,965 11,202,965 0.0 PushPop :WARMUP [prefill640_decode1] bs=16 |
| 0.0 9,898,426 1 9,898,426.0 9,898,426.0 9,898,426 9,898,426 0.0 PushPop :WARMUP [prefill640_decode1] bs=4 |
| 0.0 93,987 2 46,993.5 46,993.5 46,501 47,486 696.5 PushPop CCCL:cub::DeviceSegmentedRadixSort |
| |
| [4/8] Executing 'osrt_sum' stats report |
| |
| Time (%) Total Time (ns) Num Calls Avg (ns) Med (ns) Min (ns) Max (ns) StdDev (ns) Name |
| -------- --------------- --------- ---------------- ------------ ---------- -------------- ---------------- ---------------------- |
| 30.6 553,799,179,419 26,379 20,993,941.4 24,544.0 1,005 41,145,549,920 568,749,122.9 pthread_cond_timedwait |
| 25.2 456,524,597,477 32,258 14,152,290.8 10,062,515.0 1,005 28,095,036,340 325,046,972.8 epoll_wait |
| 23.2 419,657,136,245 24 17,485,714,010.2 6,853,452.0 18,625 41,147,042,823 20,466,758,555.3 pthread_cond_wait |
| 8.3 149,675,043,244 19,152 7,815,113.0 1,336.0 999 10,010,048,184 124,556,008.7 poll |
| 7.4 132,923,168,941 34 3,909,504,968.9 144,822.0 9,009 10,000,125,741 4,806,294,275.1 sem_timedwait |
| 3.1 55,830,697,249 34,920 1,598,817.2 2,390.5 1,000 39,313,467,846 217,801,569.2 read |
| 2.1 38,110,376,305 5,712 6,671,984.6 6,152,994.0 26,772 446,137,921 8,466,478.4 sem_wait |
| 0.0 479,589,582 195,913 2,448.0 1,418.0 1,009 72,446,503 163,781.2 munmap |
| 0.0 280,105,820 8,569 32,688.3 9,260.0 1,000 19,340,884 337,148.5 ioctl |
| 0.0 274,198,309 369 743,084.8 2,641.0 1,038 39,931,974 3,903,059.8 fopen |
| 0.0 121,562,837 24 5,065,118.2 5,064,638.5 5,054,821 5,076,686 6,107.3 nanosleep |
| 0.0 110,913,222 30,661 3,617.4 2,579.0 1,000 17,383,970 99,281.9 open64 |
| 0.0 93,147,784 103 904,347.4 641,033.0 2,165 5,493,863 923,715.3 pthread_rwlock_wrlock |
| 0.0 79,222,074 18,186 4,356.2 2,624.0 1,000 15,491,872 115,723.5 mmap64 |
| 0.0 74,214,458 95 781,204.8 4,153.0 1,052 19,575,392 3,729,074.2 open |
| 0.0 68,777,883 38 1,809,944.3 559,072.0 2,473 10,484,462 3,403,548.7 pthread_join |
| 0.0 59,686,160 60 994,769.3 3,703.5 1,002 52,559,474 6,777,723.7 waitpid |
| 0.0 56,197,546 10 5,619,754.6 18,348.5 8,949 55,929,667 17,677,147.5 connect |
| 0.0 40,540,766 22,732 1,783.4 1,440.0 1,000 168,023 3,098.9 pthread_cond_signal |
| 0.0 31,445,904 4,009 7,843.8 4,379.0 1,021 2,581,792 44,779.3 recv |
| 0.0 29,454,235 4,011 7,343.4 5,614.0 1,527 123,965 6,719.8 send |
| 0.0 28,740,099 8,225 3,494.2 2,632.0 1,008 397,826 6,722.1 write |
| 0.0 26,739,322 1 26,739,322.0 26,739,322.0 26,739,322 26,739,322 0.0 dup |
| 0.0 22,364,638 2,309 9,685.9 7,475.0 1,004 325,671 12,960.7 pthread_mutex_lock |
| 0.0 9,502,826 147 64,645.1 66,764.0 42,993 76,116 5,322.4 sleep |
| 0.0 8,105,190 5,098 1,589.9 1,408.0 999 20,826 751.4 epoll_ctl |
| 0.0 6,578,455 918 7,166.1 3,576.0 1,032 86,513 10,070.0 fgets |
| 0.0 6,248,976 131 47,702.1 41,442.0 15,477 175,400 23,310.7 pthread_create |
| 0.0 1,683,191 344 4,893.0 4,679.5 1,679 36,411 2,249.0 fopen64 |
| 0.0 1,555,208 1,244 1,250.2 1,068.0 1,001 7,887 604.1 fclose |
| 0.0 1,352,694 64 21,135.8 3,173.0 1,026 273,553 55,326.6 futex |
| 0.0 911,955 9 101,328.3 120,424.0 17,639 211,997 69,204.3 pthread_rwlock_rdlock |
| 0.0 902,671 1 902,671.0 902,671.0 902,671 902,671 0.0 fork |
| 0.0 718,148 153 4,693.8 3,110.0 1,116 50,550 6,717.9 mmap |
| 0.0 346,586 65 5,332.1 4,480.0 1,747 14,920 2,964.1 pipe2 |
| 0.0 231,056 41 5,635.5 4,622.0 1,823 17,718 3,644.9 socket |
| 0.0 175,429 20 8,771.5 2,825.5 1,038 61,940 15,631.8 bind |
| 0.0 92,804 27 3,437.2 2,888.0 1,457 8,880 1,744.4 pthread_cond_broadcast |
| 0.0 65,749 7 9,392.7 8,219.0 3,323 25,707 7,706.3 fread |
| 0.0 63,870 25 2,554.8 1,665.0 1,127 23,880 4,456.7 sigaction |
| 0.0 44,816 15 2,987.7 2,660.0 1,456 6,625 1,483.2 stat |
| 0.0 41,840 5 8,368.0 9,259.0 3,918 12,392 3,246.9 accept4 |
| 0.0 33,951 16 2,121.9 2,142.0 1,024 3,237 774.5 dup2 |
| 0.0 32,795 23 1,425.9 1,106.0 1,001 3,368 702.5 fcntl |
| 0.0 25,269 8 3,158.6 3,403.0 1,076 5,474 1,908.0 fflush |
| 0.0 23,535 4 5,883.8 5,970.0 5,137 6,458 593.8 lstat |
| 0.0 20,052 4 5,013.0 4,278.0 3,149 8,347 2,375.9 flock |
| 0.0 18,050 8 2,256.3 2,245.5 1,371 3,381 627.6 pread |
| 0.0 17,774 5 3,554.8 2,538.0 2,132 7,351 2,158.8 fwrite |
| 0.0 16,831 4 4,207.8 2,538.5 1,645 10,109 3,984.7 flockfile |
| 0.0 13,157 9 1,461.9 1,294.0 1,016 2,240 411.6 listen |
| 0.0 13,156 3 4,385.3 4,347.0 4,233 4,576 174.7 fputs_unlocked |
| 0.0 12,335 5 2,467.0 2,531.0 1,995 2,998 373.7 mprotect |
| 0.0 7,923 1 7,923.0 7,923.0 7,923 7,923 0.0 kill |
| 0.0 6,372 3 2,124.0 1,678.0 1,437 3,257 988.6 fstat |
| 0.0 3,957 3 1,319.0 1,317.0 1,280 1,360 40.0 pthread_mutex_trylock |
| 0.0 3,724 2 1,862.0 1,862.0 1,371 2,353 694.4 openat64 |
| 0.0 3,666 1 3,666.0 3,666.0 3,666 3,666 0.0 fputs |
| |
| [5/8] Executing 'cuda_api_sum' stats report |
| |
| Time (%) Total Time (ns) Num Calls Avg (ns) Med (ns) Min (ns) Max (ns) StdDev (ns) Name |
| -------- --------------- --------- ----------- ----------- -------- ----------- ----------- ------------------------------------------ |
| 37.7 2,516,538,786 63,492 39,635.5 8,623.0 2,864 117,041,679 515,768.1 cudaMemcpyAsync |
| 34.0 2,269,294,014 320,520 7,080.0 6,163.0 770 59,911,570 139,189.9 cudaLaunchKernel |
| 13.9 930,312,518 84,042 11,069.6 9,579.0 7,462 5,525,562 25,867.7 cudaGraphLaunch_v10000 |
| 3.3 219,036,068 1,943 112,730.9 72,540.0 38,839 1,515,066 189,686.6 cudaGraphInstantiateWithFlags_v11040 |
| 2.9 196,494,942 3,568 55,071.5 2,873.0 2,329 142,447,344 2,389,532.5 cudaStreamSynchronize |
| 2.2 144,936,056 2,167 66,883.3 29,852.0 1,968 79,331,250 1,703,598.4 cudaDeviceSynchronize |
| 1.6 107,945,970 33,366 3,235.2 3,829.0 617 215,309 2,439.1 cuLaunchKernel |
| 1.0 67,009,064 86,339 776.1 737.0 290 10,039 193.7 cudaStreamIsCapturing_v10000 |
| 0.8 50,768,621 220 230,766.5 122,901.0 72,410 2,852,540 313,016.1 cudaFree |
| 0.6 38,574,364 10 3,857,436.4 4,474,218.5 76,094 7,076,984 2,235,730.5 cuLibraryLoadData |
| 0.6 37,836,237 344 109,989.1 106,571.5 7,004 462,118 35,041.1 cudaMalloc |
| 0.5 34,049,787 8,194 4,155.5 3,966.5 193 1,228,658 13,779.5 cudaMemsetAsync |
| 0.2 12,827,793 169 75,904.1 82,342.0 27,226 289,473 37,429.7 cuModuleLoadData |
| 0.2 10,909,227 10,382 1,050.8 406.0 261 4,054,681 41,640.5 cuKernelGetFunction |
| 0.1 8,851,758 18,895 468.5 441.0 305 1,822 112.0 cudaStreamGetCaptureInfo_v2_v11030 |
| 0.1 7,834,959 1,943 4,032.4 3,938.0 3,099 11,065 667.7 cudaStreamBeginCapture_v10000 |
| 0.1 7,272,478 1,943 3,742.9 3,685.0 2,359 9,509 564.5 cudaGraphDestroy_v10000 |
| 0.0 2,856,444 128 22,316.0 2,314.5 1,234 940,303 115,214.0 cudaStreamCreateWithPriority |
| 0.0 2,538,948 1,943 1,306.7 1,287.0 1,003 2,449 128.2 cudaStreamEndCapture_v10000 |
| 0.0 1,556,849 1,943 801.3 733.0 596 2,550 240.5 cudaGraphGetNodes_v10000 |
| 0.0 1,246,939 14 89,067.1 5,496.5 3,635 1,165,570 309,848.1 cudaHostAlloc |
| 0.0 215,524 8 26,940.5 26,659.0 8,354 57,833 18,251.4 cudaMemGetInfo |
| 0.0 213,153 810 263.2 220.5 114 2,997 176.3 cuGetProcAddress_v2 |
| 0.0 24,622 19 1,295.9 545.0 435 4,953 1,356.7 cudaEventCreateWithFlags |
| 0.0 19,052 16 1,190.8 1,089.0 477 2,900 585.0 cuLibraryGetKernel |
| 0.0 8,007 14 571.9 547.5 300 1,096 199.3 cudaThreadExchangeStreamCaptureMode_v10010 |
| 0.0 6,691 3 2,230.3 1,975.0 1,913 2,803 496.9 cuInit |
| 0.0 5,648 1 5,648.0 5,648.0 5,648 5,648 0.0 cudaEventRecord |
| 0.0 3,830 1 3,830.0 3,830.0 3,830 3,830 0.0 cudaStreamWaitEvent |
| 0.0 3,418 4 854.5 728.0 177 1,785 760.5 cuModuleGetLoadingMode |
| 0.0 1,748 2 874.0 874.0 414 1,334 650.5 cudaGetDriverEntryPoint_v11030 |
| 0.0 1,364 1 1,364.0 1,364.0 1,364 1,364 0.0 cudaEventDestroy |
| |
| [6/8] Executing 'cuda_gpu_kern_sum' stats report |
| |
| Time (%) Total Time (ns) Instances Avg (ns) Med (ns) Min (ns) Max (ns) StdDev (ns) Name |
| -------- --------------- --------- ----------- ----------- --------- --------- ----------- ---------------------------------------------------------------------------------------------------- |
| 22.5 752,836,979 1,623 463,855.2 488,001.0 6,976 498,433 102,769.6 std::enable_if<!T7, void>::type internal::gemvx::kernel<int, int, __nv_bfloat16, __nv_bfloat16, __n… |
| 18.0 603,213,004 66,472 9,074.7 6,272.0 5,695 60,576 8,044.3 void flash::flash_fwd_splitkv_kernel<Flash_fwd_kernel_traits<(int)128, (int)64, (int)128, (int)4, (… |
| 10.8 362,298,762 4,154 87,216.8 50,816.0 7,872 549,921 134,119.9 void cutlass::Kernel2<cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_32x6_tn_align8>(T1::Param… |
| 8.7 289,989,837 1,001 289,700.1 493,760.0 10,592 507,681 232,609.8 void cutlass::Kernel2<cutlass_80_wmma_tensorop_bf16_s161616gemm_bf16_16x16_128x2_tn_align8>(T1::Par… |
| 6.4 214,006,131 381 561,695.9 569,666.0 124,352 812,257 208,209.5 ampere_bf16_s1688gemm_bf16_64x128_sliced1x2_ldg8_f2f_tn |
| 4.7 158,345,108 924 171,369.2 163,104.5 42,656 1,415,425 225,393.3 ampere_bf16_s1688gemm_bf16_128x64_sliced1x2_ldg8_f2f_tn |
| 4.0 133,349,533 70,504 1,891.4 1,888.0 1,663 2,592 183.7 void at::native::elementwise_kernel<(int)128, (int)4, void at::native::gpu_kernel_impl_nocast<at::n… |
| 3.0 101,619,901 81,144 1,252.3 1,183.0 928 2,304 162.7 void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0… |
| 2.8 94,924,575 187 507,618.0 507,681.0 505,890 511,329 671.0 void cutlass::Kernel2<cutlass_80_wmma_tensorop_bf16_s161616gemm_bf16_16x16_128x1_tn_align8>(T1::Par… |
| 2.6 86,211,442 11,312 7,621.2 6,496.0 6,240 17,536 2,446.0 void flash::flash_fwd_splitkv_kernel<Flash_fwd_kernel_traits<(int)128, (int)64, (int)128, (int)4, (… |
| 2.5 82,005,212 34 2,411,918.0 2,804,771.0 557,314 2,810,659 859,703.3 ampere_bf16_s16816gemm_bf16_128x64_ldg8_f2f_tn |
| 2.3 77,639,582 3,360 23,107.0 22,625.0 21,792 25,504 948.5 void flash::flash_fwd_splitkv_kernel<Flash_fwd_kernel_traits<(int)128, (int)64, (int)128, (int)4, (… |
| 1.7 57,584,064 2,900 19,856.6 2,080.0 1,983 218,273 48,632.2 void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIterator… |
| 1.6 54,796,208 2,902 18,882.2 5,280.0 5,056 204,353 44,079.2 void at::native::reduce_kernel<(int)512, (int)1, at::native::ReduceOp<float, at::native::ArgMaxOps<… |
| 1.4 46,186,182 35,252 1,310.2 1,312.0 1,183 1,664 89.1 void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::n… |
| 1.0 32,914,874 11,312 2,909.7 2,880.0 2,816 3,168 66.5 void flash::flash_fwd_splitkv_combine_kernel<Flash_fwd_kernel_traits<(int)128, (int)64, (int)128, (… |
| 0.7 23,852,851 784 30,424.6 13,728.0 12,032 66,240 21,441.9 ampere_bf16_s16816gemm_bf16_64x64_ldg8_f2f_stages_64x5_tn |
| 0.6 20,673,719 1,932 10,700.7 4,288.0 1,120 462,656 54,689.2 triton_poi_fused_mul_silu_1 |
| 0.6 18,926,708 28 675,953.9 677,296.5 635,969 679,296 7,864.1 void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<signed char>, std::a… |
| 0.3 10,025,160 1,932 5,189.0 3,648.0 1,504 111,328 12,654.2 triton_red_fused__to_copy_add_mean_mul_pow_rsqrt_2 |
| 0.3 9,894,698 8 1,236,837.3 1,232,289.5 1,208,961 1,274,593 28,056.5 void at_cuda_detail::cub::DeviceSegmentedRadixSortKernel<at_cuda_detail::cub::DeviceRadixSortPolicy… |
| 0.3 9,154,217 28 326,936.3 327,040.5 324,481 330,881 1,159.8 ampere_bf16_s1688gemm_bf16_128x128_ldg8_relu_f2f_stages_32x1_tn |
| 0.3 9,054,888 3,034 2,984.5 2,896.0 2,689 4,352 251.2 void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void … |
| 0.3 8,499,119 224 37,942.5 37,888.0 36,800 39,360 563.1 void cutlass::Kernel2<cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x128_32x6_tn_align8>(T1::Para… |
| 0.2 7,759,275 140 55,423.4 55,424.5 53,824 56,864 533.3 ampere_bf16_s1688gemm_bf16_128x128_ldg8_f2f_stages_32x1_tn |
| 0.2 7,224,015 336 21,500.0 21,408.0 21,056 22,560 343.1 ampere_bf16_s16816gemm_bf16_128x64_ldg8_relu_f2f_stages_64x3_tn |
| 0.2 6,500,865 476 13,657.3 13,536.0 12,160 15,872 1,120.2 ampere_bf16_s16816gemm_bf16_64x64_ldg8_relu_f2f_stages_64x5_tn |
| 0.2 6,429,453 1,932 3,327.9 2,144.0 1,504 79,744 9,090.3 triton_red_fused__to_copy_add_mean_mul_pow_rsqrt_0 |
| 0.2 5,123,524 4 1,280,881.0 1,282,577.0 1,232,705 1,325,665 48,453.1 void at_cuda_detail::cub::DeviceSegmentedRadixSortKernel<at_cuda_detail::cub::DeviceRadixSortPolicy… |
| 0.1 4,520,927 5,243 862.3 864.0 767 1,344 78.8 void at::native::vectorized_elementwise_kernel<(int)2, at::native::FillFunctor<long>, std::array<ch… |
| 0.1 4,101,276 1,863 2,201.4 1,889.0 1,344 21,824 2,256.9 triton_poi_fused_cat_3 |
| 0.1 3,803,408 2,898 1,312.4 1,185.0 1,120 2,272 210.5 void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIterator… |
| 0.1 3,599,400 56 64,275.0 64,288.0 63,264 65,216 465.8 void cutlass::Kernel2<cutlass_80_wmma_tensorop_bf16_s161616gemm_bf16_32x32_64x1_tn_align8>(T1::Para… |
| 0.1 3,352,211 2,900 1,155.9 1,088.0 992 1,728 139.6 void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIterator… |
| 0.1 3,073,044 1,512 2,032.4 1,856.0 1,312 3,105 561.0 void cublasLt::splitKreduce_kernel<(int)32, (int)16, int, __nv_bfloat16, __nv_bfloat16, float, (boo… |
| 0.1 3,033,286 112 27,082.9 26,960.5 25,824 41,824 1,477.6 ampere_bf16_s1688gemm_bf16_128x64_sliced1x2_ldg8_relu_f2f_tn |
| 0.1 2,702,248 1,863 1,450.5 1,216.0 832 16,831 1,789.0 triton_poi_fused_view_5 |
| 0.1 2,696,789 1,863 1,447.6 1,376.0 1,280 6,400 540.6 triton_poi_fused_cat_4 |
| 0.1 2,625,338 2,898 905.9 896.0 863 1,216 34.7 void at::native::unrolled_elementwise_kernel<at::native::CUDAFunctorOnSelf_add<int>, std::array<cha… |
| 0.1 2,420,864 112 21,614.9 21,488.0 9,504 34,336 12,012.3 void cutlass::Kernel2<cutlass_80_wmma_tensorop_bf16_s161616gemm_bf16_32x32_128x2_tn_align8>(T1::Par… |
| 0.1 2,362,563 4 590,640.8 592,177.0 550,272 627,937 40,524.0 void at::native::<unnamed>::cunn_SoftMaxForward<(int)4, float, float, float, at::native::<unnamed>:… |
| 0.1 2,229,105 2,757 808.5 800.0 767 1,056 18.4 void at::native::unrolled_elementwise_kernel<at::native::FillFunctor<int>, std::array<char *, (unsi… |
| 0.1 1,966,467 2 983,233.5 983,233.5 936,865 1,029,602 65,575.0 void at::native::_scatter_gather_elementwise_kernel<(int)128, (int)8, void at::native::_cuda_scatte… |
| 0.0 1,538,596 1,868 823.7 832.0 800 1,056 18.7 void at::native::vectorized_elementwise_kernel<(int)2, at::native::FillFunctor<int>, std::array<cha… |
| 0.0 1,390,264 1,546 899.3 896.0 864 929 10.1 void at::native::unrolled_elementwise_kernel<at::native::FillFunctor<long>, std::array<char *, (uns… |
| 0.0 1,362,401 4 340,600.3 340,768.0 339,520 341,345 831.2 void at::native::vectorized_elementwise_kernel<(int)4, at::native::<unnamed>::masked_fill_kernel(at… |
| 0.0 990,433 2 495,216.5 495,216.5 494,817 495,616 565.0 void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, floa… |
| 0.0 956,446 28 34,158.8 34,784.0 17,408 35,200 3,289.6 std::enable_if<!T7, void>::type internal::gemvx::kernel<int, int, __nv_bfloat16, float, float, floa… |
| 0.0 895,777 4 223,944.3 223,488.5 213,440 235,360 12,011.4 void at::native::elementwise_kernel<(int)128, (int)4, void at::native::gpu_kernel_impl_nocast<at::n… |
| 0.0 741,373 893 830.2 831.0 800 1,632 52.8 void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<int>, std::array<cha… |
| 0.0 707,616 2 353,808.0 353,808.0 353,312 354,304 701.4 void at::native::tensor_kernel_scan_innermost_dim<float, std::plus<float>>(T1 *, const T1 *, unsign… |
| 0.0 682,590 28 24,378.2 24,367.5 24,031 24,768 237.4 ampere_bf16_s16816gemm_bf16_128x64_ldg8_f2f_stages_32x6_tn |
| 0.0 637,344 2 318,672.0 318,672.0 318,528 318,816 203.6 at::native::<unnamed>::fill_reverse_indices_kernel(long *, int, at::cuda::detail::IntDivider<unsign… |
| 0.0 537,760 2 268,880.0 268,880.0 268,640 269,120 339.4 void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::n… |
| 0.0 328,896 2 164,448.0 164,448.0 159,552 169,344 6,924.0 void at::native::<unnamed>::distribution_elementwise_grid_stride_kernel<float, (int)4, void at::nat… |
| 0.0 298,467 168 1,776.6 1,776.0 1,600 2,080 114.0 void cublasLt::splitKreduce_kernel<(int)32, (int)16, int, __nv_bfloat16, __nv_bfloat16, float, (boo… |
| 0.0 260,097 69 3,769.5 3,200.0 1,984 36,416 4,039.0 triton_red_fused__to_copy_add_embedding_mean_mul_pow_rsqrt_0 |
| 0.0 190,973 69 2,767.7 2,208.0 1,632 40,225 4,583.0 triton_poi_fused_cat_1 |
| 0.0 156,416 1 156,416.0 156,416.0 156,416 156,416 0.0 void at::native::<unnamed>::CatArrayBatchedCopy_aligned16_contig<at::native::<unnamed>::OpaqueType<… |
| 0.0 102,436 69 1,484.6 1,376.0 1,248 8,672 879.2 triton_poi_fused_cat_2 |
| 0.0 99,200 69 1,437.7 1,280.0 832 14,016 1,556.1 triton_poi_fused_view_3 |
| 0.0 79,232 1 79,232.0 79,232.0 79,232 79,232 0.0 void at::native::vectorized_elementwise_kernel<(int)4, at::native::bfloat16_copy_kernel_cuda(at::Te… |
| 0.0 64,066 58 1,104.6 896.0 863 11,360 1,371.7 void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<c10::BFloat16>, std:… |
| 0.0 43,616 1 43,616.0 43,616.0 43,616 43,616 0.0 void at::native::vectorized_elementwise_kernel<(int)4, at::native::sin_kernel_cuda(at::TensorIterat… |
| 0.0 36,638 28 1,308.5 1,312.0 1,280 1,344 13.3 void cublasLt::splitKreduce_kernel<(int)32, (int)16, int, float, __nv_bfloat16, float, (bool)0, __n… |
| 0.0 26,528 1 26,528.0 26,528.0 26,528 26,528 0.0 void at::native::vectorized_elementwise_kernel<(int)4, at::native::cos_kernel_cuda(at::TensorIterat… |
| 0.0 19,617 1 19,617.0 19,617.0 19,617 19,617 0.0 void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::n… |
| 0.0 11,488 11 1,044.4 865.0 863 1,472 234.1 void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<float>, std::array<c… |
| 0.0 6,623 2 3,311.5 3,311.5 3,263 3,360 68.6 void at::native::<unnamed>::distribution_elementwise_grid_stride_kernel<float, (int)4, void at::nat… |
| 0.0 4,863 2 2,431.5 2,431.5 2,335 2,528 136.5 void at::native::_scatter_gather_elementwise_kernel<(int)128, (int)8, void at::native::_cuda_scatte… |
| 0.0 3,614 2 1,807.0 1,807.0 1,663 1,951 203.6 void at::native::vectorized_elementwise_kernel<(int)2, at::native::CUDAFunctorOnOther_add<long>, st… |
| 0.0 3,200 2 1,600.0 1,600.0 1,376 1,824 316.8 void at::native::elementwise_kernel<(int)128, (int)4, void at::native::gpu_kernel_impl_nocast<at::n… |
| 0.0 3,199 2 1,599.5 1,599.5 1,471 1,728 181.7 void at::native::vectorized_elementwise_kernel<(int)4, void at::native::compare_scalar_kernel<float… |
| 0.0 3,072 2 1,536.0 1,536.0 1,504 1,568 45.3 void at::native::vectorized_elementwise_kernel<(int)2, at::native::<unnamed>::where_kernel_impl(at:… |
| 0.0 2,975 2 1,487.5 1,487.5 991 1,984 702.2 void <unnamed>::elementwise_kernel_with_index<int, at::native::arange_cuda_out(const c10::Scalar &,… |
| 0.0 2,944 2 1,472.0 1,472.0 1,344 1,600 181.0 void at::native::vectorized_elementwise_kernel<(int)4, at::native::CUDAFunctorOnOther_add<float>, s… |
| 0.0 2,304 1 2,304.0 2,304.0 2,304 2,304 0.0 void at::native::elementwise_kernel<(int)128, (int)4, void at::native::gpu_kernel_impl<at::native::… |
| 0.0 1,183 1 1,183.0 1,183.0 1,183 1,183 0.0 void at::native::vectorized_elementwise_kernel<(int)4, at::native::reciprocal_kernel_cuda(at::Tenso… |
| 0.0 1,024 1 1,024.0 1,024.0 1,024 1,024 0.0 void at::native::vectorized_elementwise_kernel<(int)4, at::native::AUnaryFunctor<float, float, floa… |
| 0.0 1,024 1 1,024.0 1,024.0 1,024 1,024 0.0 void at::native::vectorized_elementwise_kernel<(int)4, at::native::BUnaryFunctor<float, float, floa… |
| 0.0 896 1 896.0 896.0 896 896 0.0 void at::native::vectorized_elementwise_kernel<(int)2, at::native::FillFunctor<double>, std::array<… |
| |
| [7/8] Executing 'cuda_gpu_mem_time_sum' stats report |
| |
| Time (%) Total Time (ns) Count Avg (ns) Med (ns) Min (ns) Max (ns) StdDev (ns) Operation |
| -------- --------------- ------ -------- -------- -------- ----------- ----------- ------------------------------ |
| 91.9 516,928,129 18,058 28,626.0 351.0 287 116,556,811 913,700.2 [CUDA memcpy Host-to-Device] |
| 7.1 39,961,230 42,536 939.5 896.0 895 344,640 3,316.6 [CUDA memcpy Device-to-Device] |
| 0.5 3,068,371 2,898 1,058.8 1,025.0 863 1,568 105.2 [CUDA memcpy Device-to-Host] |
| 0.4 2,471,409 5,282 467.9 352.0 288 1,792 211.8 [CUDA memset] |
| |
| [8/8] Executing 'cuda_gpu_mem_size_sum' stats report |
| |
| Total (MB) Count Avg (MB) Med (MB) Min (MB) Max (MB) StdDev (MB) Operation |
| ---------- ------ -------- -------- -------- -------- ----------- ------------------------------ |
| 3,251.899 18,058 0.180 0.000 0.000 466.747 3.954 [CUDA memcpy Host-to-Device] |
| 752.988 42,536 0.018 0.003 0.003 155.582 1.509 [CUDA memcpy Device-to-Device] |
| 2.319 5,282 0.000 0.000 0.000 0.006 0.001 [CUDA memset] |
| 0.316 2,898 0.000 0.000 0.000 0.001 0.000 [CUDA memcpy Device-to-Host] |
| |
| Generated: |
| /data/cy/kv_cache_vs_util/test_nvtx.nsys-rep |
| /data/cy/kv_cache_vs_util/test_nvtx.sqlite |
| |
