Datasets:

Hamerlate
/

tts

	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 08-10 22:55:56 [__init__.py:244] Automatically detected platform cuda.
	INFO:__main__:FastTTS AIME Experiment
	INFO:__main__:==================================================
	INFO:__main__:Starting FastTTS AIME experiment
	INFO:__main__:Parameters: {'num_iterations': 2, 'n': 128, 'temperature': 2, 'beam_width': 4, 'generator_model': 'Qwen/Qwen2.5-Math-1.5B-Instruct', 'verifier_model': 'peiyi9979/math-shepherd-mistral-7b-prm', 'generator_gpu_memory': 0.3, 'verifier_gpu_memory': 0.62, 'offload_enabled': False, 'spec_beam_extension': False, 'prefix_aware_scheduling': False}
	INFO:__main__:Loaded AIME dataset with 30 samples
	INFO:__main__:Problem: Every morning Aya goes for a $9$-kilometer-long walk and stops at a coffee shop afterwards. When she walks at a constant speed of $s$ kilometers per hour, the walk takes her 4 hours, including $t$ minutes spent in the coffee shop. When she walks $s+2$ kilometers per hour, the walk takes her 2 hours and 24 minutes, including $t$ minutes spent in the coffee shop. Suppose Aya walks at $s+\frac{1}{2}$ kilometers per hour. Find the number of minutes the walk takes her, including the $t$ minutes spent in the coffee shop.
	INFO:__main__:Reference answer: 204
	INFO:__main__:Initializing FastTTS models...
	INFO:fasttts:Initializing FastTTS models...
	INFO:models.vllm_wrapper:Initializing generator model: Qwen/Qwen2.5-Math-1.5B-Instruct
	INFO 08-10 22:56:08 [__init__.py:244] Automatically detected platform cuda.
	INFO:models.tts_llm:Using V0 engine with speculative beam extension: False
	INFO:models.tts_llm:Prefix-aware scheduling enabled: False
	✅ Process PID: 3740901 \| CUDA Context Object: None
	INFO 08-10 22:56:19 [config.py:841] This model supports multiple tasks: {'classify', 'reward', 'generate', 'embed'}. Defaulting to 'generate'.
	INFO 08-10 22:56:19 [config.py:1472] Using max model len 4096
	INFO:models.generator_engine:Using GeneratorLLMEngine with vLLM version 0.9.2
	INFO 08-10 22:56:19 [llm_engine.py:230] Initializing a V0 LLM engine (v0.9.2) with config: model='Qwen/Qwen2.5-Math-1.5B-Instruct', speculative_config=None, tokenizer='Qwen/Qwen2.5-Math-1.5B-Instruct', skip_tokenizer_init=False, tokenizer_mode=auto, revision=None, override_neuron_config={}, tokenizer_revision=None, trust_remote_code=False, dtype=torch.bfloat16, max_seq_len=4096, 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='xgrammar', 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=42, served_model_name=Qwen/Qwen2.5-Math-1.5B-Instruct, num_scheduler_steps=1, multi_step_stream_outputs=True, enable_prefix_caching=True, chunked_prefill_enabled=False, use_async_output_proc=True, pooler_config=None, compilation_config={"level":0,"debug_dump_path":"","cache_dir":"","backend":"","custom_ops":[],"splitting_ops":[],"use_inductor":true,"compile_sizes":[],"inductor_compile_config":{"enable_auto_functionalized_v2":false},"inductor_passes":{},"use_cudagraph":false,"cudagraph_num_of_warmups":0,"cudagraph_capture_sizes":[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":256,"local_cache_dir":null}, use_cached_outputs=False,
	INFO 08-10 22:56:20 [cuda.py:363] Using Flash Attention backend.
	INFO 08-10 22:56:21 [parallel_state.py:1076] rank 0 in world size 1 is assigned as DP rank 0, PP rank 0, TP rank 0, EP rank 0
	INFO 08-10 22:56:21 [model_runner.py:1171] Starting to load model Qwen/Qwen2.5-Math-1.5B-Instruct...
	INFO 08-10 22:56:21 [weight_utils.py:292] Using model weights format ['*.safetensors']
	INFO 08-10 22:56:22 [weight_utils.py:345] 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, 1.87it/s]
	Loading safetensors checkpoint shards: 100% Completed \| 1/1 [00:00<00:00, 1.87it/s]

	INFO 08-10 22:56:23 [default_loader.py:272] Loading weights took 0.64 seconds
	INFO 08-10 22:56:23 [model_runner.py:1203] Model loading took 2.8798 GiB and 1.712160 seconds
	INFO 08-10 22:56:24 [worker.py:294] Memory profiling takes 0.58 seconds
	INFO 08-10 22:56:24 [worker.py:294] the current vLLM instance can use total_gpu_memory (23.64GiB) x gpu_memory_utilization (0.30) = 7.09GiB
	INFO 08-10 22:56:24 [worker.py:294] model weights take 2.88GiB; non_torch_memory takes 0.08GiB; PyTorch activation peak memory takes 1.40GiB; the rest of the memory reserved for KV Cache is 2.74GiB.
	INFO 08-10 22:56:24 [executor_base.py:113] # cuda blocks: 6412, # CPU blocks: 9362
	INFO 08-10 22:56:24 [executor_base.py:118] Maximum concurrency for 4096 tokens per request: 25.05x
	INFO 08-10 22:56:27 [model_runner.py:1513] Capturing cudagraphs for decoding. This may lead to unexpected consequences if the model is not static. To run the model in eager mode, set 'enforce_eager=True' or use '--enforce-eager' in the CLI. If out-of-memory error occurs during cudagraph capture, consider decreasing `gpu_memory_utilization` or switching to eager mode. You can also reduce the `max_num_seqs` as needed to decrease memory usage.
	Capturing CUDA graph shapes: 0%\| \| 0/35 [00:00<?, ?it/s] Capturing CUDA graph shapes: 3%\|▎ \| 1/35 [00:00<00:16, 2.03it/s] Capturing CUDA graph shapes: 6%\|▌ \| 2/35 [00:00<00:14, 2.26it/s] Capturing CUDA graph shapes: 9%\|▊ \| 3/35 [00:01<00:13, 2.35it/s] Capturing CUDA graph shapes: 11%\|█▏ \| 4/35 [00:01<00:12, 2.38it/s] Capturing CUDA graph shapes: 14%\|█▍ \| 5/35 [00:02<00:12, 2.42it/s] Capturing CUDA graph shapes: 17%\|█▋ \| 6/35 [00:02<00:11, 2.43it/s] Capturing CUDA graph shapes: 20%\|██ \| 7/35 [00:02<00:11, 2.43it/s] Capturing CUDA graph shapes: 23%\|██▎ \| 8/35 [00:03<00:11, 2.43it/s] Capturing CUDA graph shapes: 26%\|██▌ \| 9/35 [00:03<00:10, 2.43it/s] Capturing CUDA graph shapes: 29%\|██▊ \| 10/35 [00:04<00:10, 2.41it/s] Capturing CUDA graph shapes: 31%\|███▏ \| 11/35 [00:04<00:10, 2.37it/s] Capturing CUDA graph shapes: 34%\|███▍ \| 12/35 [00:05<00:09, 2.37it/s] Capturing CUDA graph shapes: 37%\|███▋ \| 13/35 [00:05<00:09, 2.39it/s] Capturing CUDA graph shapes: 40%\|████ \| 14/35 [00:05<00:08, 2.40it/s] Capturing CUDA graph shapes: 43%\|████▎ \| 15/35 [00:06<00:08, 2.42it/s] Capturing CUDA graph shapes: 46%\|████▌ \| 16/35 [00:06<00:07, 2.43it/s] Capturing CUDA graph shapes: 49%\|████▊ \| 17/35 [00:07<00:07, 2.43it/s] Capturing CUDA graph shapes: 51%\|█████▏ \| 18/35 [00:07<00:06, 2.44it/s] Capturing CUDA graph shapes: 54%\|█████▍ \| 19/35 [00:07<00:06, 2.43it/s] Capturing CUDA graph shapes: 57%\|█████▋ \| 20/35 [00:08<00:06, 2.43it/s] Capturing CUDA graph shapes: 60%\|██████ \| 21/35 [00:08<00:05, 2.39it/s] Capturing CUDA graph shapes: 63%\|██████▎ \| 22/35 [00:09<00:05, 2.39it/s] Capturing CUDA graph shapes: 66%\|██████▌ \| 23/35 [00:09<00:04, 2.40it/s] Capturing CUDA graph shapes: 69%\|██████▊ \| 24/35 [00:10<00:04, 2.40it/s] Capturing CUDA graph shapes: 71%\|███████▏ \| 25/35 [00:10<00:04, 2.40it/s] Capturing CUDA graph shapes: 74%\|███████▍ \| 26/35 [00:10<00:03, 2.39it/s] Capturing CUDA graph shapes: 77%\|███████▋ \| 27/35 [00:11<00:03, 2.40it/s] Capturing CUDA graph shapes: 80%\|████████ \| 28/35 [00:11<00:02, 2.40it/s] Capturing CUDA graph shapes: 83%\|████████▎ \| 29/35 [00:12<00:02, 2.40it/s] Capturing CUDA graph shapes: 86%\|████████▌ \| 30/35 [00:12<00:02, 2.38it/s] Capturing CUDA graph shapes: 89%\|████████▊ \| 31/35 [00:12<00:01, 2.35it/s] Capturing CUDA graph shapes: 91%\|█████████▏\| 32/35 [00:13<00:01, 2.37it/s] Capturing CUDA graph shapes: 94%\|█████████▍\| 33/35 [00:13<00:00, 2.38it/s] Capturing CUDA graph shapes: 97%\|█████████▋\| 34/35 [00:14<00:00, 2.36it/s] Capturing CUDA graph shapes: 100%\|██████████\| 35/35 [00:14<00:00, 2.36it/s] Capturing CUDA graph shapes: 100%\|██████████\| 35/35 [00:14<00:00, 2.39it/s]
	INFO 08-10 22:56:42 [model_runner.py:1671] Graph capturing finished in 15 secs, took 0.23 GiB
	INFO 08-10 22:56:42 [llm_engine.py:428] init engine (profile, create kv cache, warmup model) took 18.72 seconds
	INFO:models.custom_scheduler:Using CustomScheduler
	INFO:models.custom_scheduler:CustomScheduler initialized with config: SchedulerConfig(runner_type='generate', max_num_batched_tokens=4096, max_num_seqs=256, max_model_len=4096, max_num_partial_prefills=1, max_long_partial_prefills=1, long_prefill_token_threshold=0, num_lookahead_slots=0, cuda_graph_sizes=[512], delay_factor=0.0, enable_chunked_prefill=False, is_multimodal_model=False, max_num_encoder_input_tokens=4096, encoder_cache_size=4096, preemption_mode=None, num_scheduler_steps=1, multi_step_stream_outputs=True, send_delta_data=False, policy='fcfs', chunked_prefill_enabled=False, disable_chunked_mm_input=False, scheduler_cls=<class 'models.custom_scheduler.CustomScheduler'>, disable_hybrid_kv_cache_manager=False)
	INFO:models.vllm_wrapper:Generator model initialized successfully in separate process
	INFO:models.vllm_wrapper:Initializing verifier model: peiyi9979/math-shepherd-mistral-7b-prm
	INFO 08-10 22:56:46 [__init__.py:244] Automatically detected platform cuda.
	INFO:models.tts_llm:Prefix-aware scheduling enabled: False
	✅ Process PID: 3741297 \| CUDA Context Object: None
	INFO 08-10 22:56:58 [config.py:1472] Using max model len 4096
	INFO 08-10 22:56:58 [arg_utils.py:1596] (Disabling) chunked prefill by default
	INFO 08-10 22:56:58 [config.py:4601] Only "last" pooling supports chunked prefill and prefix caching; disabling both.
	You are using the default legacy behaviour of the <class 'transformers.models.llama.tokenization_llama.LlamaTokenizer'>. This is expected, and simply means that the `legacy` (previous) behavior will be used so nothing changes for you. If you want to use the new behaviour, set `legacy=False`. This should only be set if you understand what it means, and thoroughly read the reason why this was added as explained in https://github.com/huggingface/transformers/pull/24565 - if you loaded a llama tokenizer from a GGUF file you can ignore this message
	You are using the default legacy behaviour of the <class 'transformers.models.llama.tokenization_llama_fast.LlamaTokenizerFast'>. This is expected, and simply means that the `legacy` (previous) behavior will be used so nothing changes for you. If you want to use the new behaviour, set `legacy=False`. This should only be set if you understand what it means, and thoroughly read the reason why this was added as explained in https://github.com/huggingface/transformers/pull/24565 - if you loaded a llama tokenizer from a GGUF file you can ignore this message.
	INFO 08-10 22:57:00 [core.py:526] Waiting for init message from front-end.
	INFO 08-10 22:57:00 [core.py:69] Initializing a V1 LLM engine (v0.9.2) with config: model='peiyi9979/math-shepherd-mistral-7b-prm', speculative_config=None, tokenizer='peiyi9979/math-shepherd-mistral-7b-prm', skip_tokenizer_init=False, tokenizer_mode=auto, revision=None, override_neuron_config={}, tokenizer_revision=None, trust_remote_code=False, dtype=torch.bfloat16, max_seq_len=4096, 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='xgrammar', 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=42, served_model_name=peiyi9979/math-shepherd-mistral-7b-prm, num_scheduler_steps=1, multi_step_stream_outputs=True, enable_prefix_caching=False, chunked_prefill_enabled=False, use_async_output_proc=False, pooler_config=PoolerConfig(pooling_type='STEP', normalize=None, softmax=True, step_tag_id=12902, returned_token_ids=[648, 387]), compilation_config={"level":3,"debug_dump_path":"","cache_dir":"","backend":"","custom_ops":[],"splitting_ops":["vllm.unified_attention","vllm.unified_attention_with_output"],"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 08-10 22:57:01 [parallel_state.py:1076] rank 0 in world size 1 is assigned as DP rank 0, PP rank 0, TP rank 0, EP rank 0
	WARNING 08-10 22:57:01 [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 08-10 22:57:01 [gpu_model_runner.py:1770] Starting to load model peiyi9979/math-shepherd-mistral-7b-prm...
	INFO 08-10 22:57:01 [gpu_model_runner.py:1775] Loading model from scratch...
	INFO 08-10 22:57:01 [cuda.py:284] Using Flash Attention backend on V1 engine.
	INFO 08-10 22:57:02 [weight_utils.py:292] Using model weights format ['*.bin']
	Loading pt checkpoint shards: 0% Completed \| 0/2 [00:00<?, ?it/s]
	Loading pt checkpoint shards: 50% Completed \| 1/2 [00:05<00:05, 5.81s/it]
	Loading pt checkpoint shards: 100% Completed \| 2/2 [00:09<00:00, 4.80s/it]
	Loading pt checkpoint shards: 100% Completed \| 2/2 [00:09<00:00, 4.95s/it]

	INFO 08-10 22:57:12 [default_loader.py:272] Loading weights took 9.90 seconds
	INFO 08-10 22:57:13 [gpu_model_runner.py:1801] Model loading took 13.2457 GiB and 11.068114 seconds
	INFO 08-10 22:57:20 [backends.py:508] Using cache directory: /home/cy/.cache/vllm/torch_compile_cache/eae4db4fef/rank_0_0/backbone for vLLM's torch.compile
	INFO 08-10 22:57:20 [backends.py:519] Dynamo bytecode transform time: 7.08 s
	INFO 08-10 22:57:25 [backends.py:155] Directly load the compiled graph(s) for shape None from the cache, took 4.779 s
	INFO 08-10 22:57:26 [monitor.py:34] torch.compile takes 7.08 s in total
	INFO 08-10 22:57:27 [gpu_worker.py:232] Available KV cache memory: 0.88 GiB
	INFO 08-10 22:57:27 [kv_cache_utils.py:716] GPU KV cache size: 7,168 tokens
	INFO 08-10 22:57:27 [kv_cache_utils.py:720] Maximum concurrency for 4,096 tokens per request: 1.75x
	Capturing CUDA graph shapes: 0%\| \| 0/67 [00:00<?, ?it/s] Capturing CUDA graph shapes: 1%\|▏ \| 1/67 [00:00<00:20, 3.15it/s] Capturing CUDA graph shapes: 3%\|▎ \| 2/67 [00:00<00:20, 3.21it/s] Capturing CUDA graph shapes: 4%\|▍ \| 3/67 [00:00<00:19, 3.24it/s] Capturing CUDA graph shapes: 6%\|▌ \| 4/67 [00:01<00:19, 3.26it/s] Capturing CUDA graph shapes: 7%\|▋ \| 5/67 [00:01<00:19, 3.22it/s] Capturing CUDA graph shapes: 9%\|▉ \| 6/67 [00:01<00:18, 3.23it/s] Capturing CUDA graph shapes: 10%\|█ \| 7/67 [00:02<00:18, 3.22it/s] Capturing CUDA graph shapes: 12%\|█▏ \| 8/67 [00:02<00:18, 3.23it/s] Capturing CUDA graph shapes: 13%\|█▎ \| 9/67 [00:02<00:18, 3.20it/s] Capturing CUDA graph shapes: 15%\|█▍ \| 10/67 [00:03<00:17, 3.21it/s] Capturing CUDA graph shapes: 16%\|█▋ \| 11/67 [00:03<00:17, 3.23it/s] Capturing CUDA graph shapes: 18%\|█▊ \| 12/67 [00:03<00:17, 3.20it/s] Capturing CUDA graph shapes: 19%\|█▉ \| 13/67 [00:04<00:16, 3.19it/s] Capturing CUDA graph shapes: 21%\|██ \| 14/67 [00:04<00:16, 3.22it/s] Capturing CUDA graph shapes: 22%\|██▏ \| 15/67 [00:04<00:16, 3.23it/s] Capturing CUDA graph shapes: 24%\|██▍ \| 16/67 [00:04<00:15, 3.20it/s] Capturing CUDA graph shapes: 25%\|██▌ \| 17/67 [00:05<00:15, 3.22it/s] Capturing CUDA graph shapes: 27%\|██▋ \| 18/67 [00:05<00:15, 3.23it/s] Capturing CUDA graph shapes: 28%\|██▊ \| 19/67 [00:05<00:14, 3.24it/s] Capturing CUDA graph shapes: 30%\|██▉ \| 20/67 [00:06<00:14, 3.27it/s] Capturing CUDA graph shapes: 31%\|███▏ \| 21/67 [00:06<00:14, 3.28it/s] Capturing CUDA graph shapes: 33%\|███▎ \| 22/67 [00:06<00:13, 3.30it/s] Capturing CUDA graph shapes: 34%\|███▍ \| 23/67 [00:07<00:13, 3.31it/s] Capturing CUDA graph shapes: 36%\|███▌ \| 24/67 [00:07<00:12, 3.32it/s] Capturing CUDA graph shapes: 37%\|███▋ \| 25/67 [00:07<00:12, 3.28it/s] Capturing CUDA graph shapes: 39%\|███▉ \| 26/67 [00:08<00:12, 3.29it/s] Capturing CUDA graph shapes: 40%\|████ \| 27/67 [00:08<00:12, 3.30it/s] Capturing CUDA graph shapes: 42%\|████▏ \| 28/67 [00:08<00:11, 3.30it/s] Capturing CUDA graph shapes: 43%\|████▎ \| 29/67 [00:08<00:11, 3.32it/s] Capturing CUDA graph shapes: 45%\|████▍ \| 30/67 [00:09<00:11, 3.31it/s] Capturing CUDA graph shapes: 46%\|████▋ \| 31/67 [00:09<00:10, 3.31it/s] Capturing CUDA graph shapes: 48%\|████▊ \| 32/67 [00:09<00:10, 3.31it/s] Capturing CUDA graph shapes: 49%\|████▉ \| 33/67 [00:10<00:10, 3.35it/s] Capturing CUDA graph shapes: 51%\|█████ \| 34/67 [00:10<00:09, 3.39it/s] Capturing CUDA graph shapes: 52%\|█████▏ \| 35/67 [00:10<00:09, 3.38it/s] Capturing CUDA graph shapes: 54%\|█████▎ \| 36/67 [00:10<00:09, 3.40it/s] Capturing CUDA graph shapes: 55%\|█████▌ \| 37/67 [00:11<00:08, 3.42it/s] Capturing CUDA graph shapes: 57%\|█████▋ \| 38/67 [00:11<00:08, 3.44it/s] Capturing CUDA graph shapes: 58%\|█████▊ \| 39/67 [00:11<00:08, 3.45it/s] Capturing CUDA graph shapes: 60%\|█████▉ \| 40/67 [00:12<00:08, 3.36it/s] Capturing CUDA graph shapes: 61%\|██████ \| 41/67 [00:12<00:07, 3.39it/s] Capturing CUDA graph shapes: 63%\|██████▎ \| 42/67 [00:12<00:07, 3.41it/s] Capturing CUDA graph shapes: 64%\|██████▍ \| 43/67 [00:13<00:07, 3.43it/s] Capturing CUDA graph shapes: 66%\|██████▌ \| 44/67 [00:13<00:06, 3.44it/s] Capturing CUDA graph shapes: 67%\|██████▋ \| 45/67 [00:13<00:06, 3.42it/s] Capturing CUDA graph shapes: 69%\|██████▊ \| 46/67 [00:13<00:06, 3.44it/s] Capturing CUDA graph shapes: 70%\|███████ \| 47/67 [00:14<00:05, 3.46it/s] Capturing CUDA graph shapes: 72%\|███████▏ \| 48/67 [00:14<00:05, 3.46it/s] Capturing CUDA graph shapes: 73%\|███████▎ \| 49/67 [00:14<00:05, 3.47it/s] Capturing CUDA graph shapes: 75%\|███████▍ \| 50/67 [00:15<00:04, 3.47it/s] Capturing CUDA graph shapes: 76%\|███████▌ \| 51/67 [00:15<00:04, 3.47it/s] Capturing CUDA graph shapes: 78%\|███████▊ \| 52/67 [00:15<00:04, 3.49it/s] Capturing CUDA graph shapes: 79%\|███████▉ \| 53/67 [00:15<00:04, 3.49it/s] Capturing CUDA graph shapes: 81%\|████████ \| 54/67 [00:16<00:03, 3.50it/s] Capturing CUDA graph shapes: 82%\|████████▏ \| 55/67 [00:16<00:03, 3.50it/s] Capturing CUDA graph shapes: 84%\|████████▎ \| 56/67 [00:16<00:03, 3.47it/s] Capturing CUDA graph shapes: 85%\|████████▌ \| 57/67 [00:17<00:02, 3.45it/s] Capturing CUDA graph shapes: 87%\|████████▋ \| 58/67 [00:17<00:02, 3.46it/s] Capturing CUDA graph shapes: 88%\|████████▊ \| 59/67 [00:17<00:02, 3.47it/s] Capturing CUDA graph shapes: 90%\|████████▉ \| 60/67 [00:17<00:02, 3.49it/s] Capturing CUDA graph shapes: 91%\|█████████ \| 61/67 [00:18<00:01, 3.49it/s] Capturing CUDA graph shapes: 93%\|█████████▎\| 62/67 [00:18<00:01, 3.50it/s] Capturing CUDA graph shapes: 94%\|█████████▍\| 63/67 [00:18<00:01, 3.51it/s] Capturing CUDA graph shapes: 96%\|█████████▌\| 64/67 [00:19<00:00, 3.51it/s] Capturing CUDA graph shapes: 97%\|█████████▋\| 65/67 [00:19<00:00, 3.51it/s] Capturing CUDA graph shapes: 99%\|█████████▊\| 66/67 [00:19<00:00, 3.52it/s] Capturing CUDA graph shapes: 100%\|██████████\| 67/67 [00:19<00:00, 3.49it/s] Capturing CUDA graph shapes: 100%\|██████████\| 67/67 [00:19<00:00, 3.36it/s]
	INFO 08-10 22:57:47 [gpu_model_runner.py:2326] Graph capturing finished in 20 secs, took 0.53 GiB
	INFO 08-10 22:57:47 [core.py:172] init engine (profile, create kv cache, warmup model) took 34.54 seconds
	INFO 08-10 22:57:48 [config.py:4601] Only "last" pooling supports chunked prefill and prefix caching; disabling both.
	INFO:models.vllm_wrapper:Verifier model initialized successfully in separate process
	INFO:fasttts:FastTTS models initialized successfully
	INFO:__main__:Starting search...
	INFO:fasttts:Processing 1 problems at once
	INFO:search.beam_search:Starting beam search iterations
	Beam search iterations: 0%\| \| 0/2 [00:00<?, ?it/s] Adding requests: 0%\| \| 0/128 [00:00<?, ?it/s] Adding requests: 91%\|█████████▏\| 117/128 [00:00<00:00, 1160.83it/s] Adding requests: 100%\|██████████\| 128/128 [00:00<00:00, 1160.14it/s]
	Processed prompts: 0%\| \| 0/128 [00:00<?, ?it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s]INFO 08-10 22:57:48 [metrics.py:417] Avg prompt throughput: 61.0 tokens/s, Avg generation throughput: 0.2 tokens/s, Running: 128 reqs, Swapped: 0 reqs, Pending: 0 reqs, GPU KV cache usage: 2.2%, CPU KV cache usage: 0.0%.
	INFO 08-10 22:57:48 [metrics.py:433] Prefix cache hit rate: GPU: 99.22%, CPU: 0.00%
	Processed prompts: 1%\| \| 1/128 [00:00<00:26, 4.86it/s, est. speed input: 1313.28 toks/s, output: 9.73 toks/s] Processed prompts: 5%\|▍ \| 6/128 [00:00<00:06, 19.21it/s, est. speed input: 4519.96 toks/s, output: 86.49 toks/s] Processed prompts: 9%\|▊ \| 11/128 [00:00<00:04, 29.07it/s, est. speed input: 6457.69 toks/s, output: 208.72 toks/s] Processed prompts: 14%\|█▍ \| 18/128 [00:00<00:03, 36.47it/s, est. speed input: 7994.28 toks/s, output: 376.67 toks/s] Processed prompts: 17%\|█▋ \| 22/128 [00:00<00:03, 32.05it/s, est. speed input: 7735.07 toks/s, output: 463.57 toks/s] Processed prompts: 20%\|██ \| 26/128 [00:01<00:04, 24.92it/s, est. speed input: 6939.79 toks/s, output: 525.91 toks/s] Processed prompts: 23%\|██▎ \| 29/128 [00:01<00:03, 24.87it/s, est. speed input: 6911.55 toks/s, output: 620.53 toks/s] Processed prompts: 27%\|██▋ \| 34/128 [00:01<00:03, 30.22it/s, est. speed input: 7421.44 toks/s, output: 820.55 toks/s] Processed prompts: 30%\|██▉ \| 38/128 [00:01<00:03, 23.60it/s, est. speed input: 6862.34 toks/s, output: 885.53 toks/s] Processed prompts: 32%\|███▏ \| 41/128 [00:01<00:03, 24.18it/s, est. speed input: 6877.88 toks/s, output: 990.35 toks/s] Processed prompts: 34%\|███▍ \| 44/128 [00:01<00:03, 23.94it/s, est. speed input: 6834.11 toks/s, output: 1086.08 toks/s] Processed prompts: 38%\|███▊ \| 49/128 [00:01<00:02, 29.18it/s, est. speed input: 7158.78 toks/s, output: 1301.87 toks/s] Processed prompts: 41%\|████▏ \| 53/128 [00:02<00:03, 19.69it/s, est. speed input: 6492.75 toks/s, output: 1316.23 toks/s] Processed prompts: 44%\|████▍ \| 56/128 [00:02<00:04, 16.02it/s, est. speed input: 6049.52 toks/s, output: 1346.32 toks/s] Processed prompts: 46%\|████▌ \| 59/128 [00:02<00:04, 14.29it/s, est. speed input: 5742.60 toks/s, output: 1404.81 toks/s] Processed prompts: 49%\|████▉ \| 63/128 [00:02<00:03, 17.91it/s, est. speed input: 5900.48 toks/s, output: 1611.26 toks/s] Processed prompts: 52%\|█████▏ \| 66/128 [00:02<00:03, 19.78it/s, est. speed input: 5960.38 toks/s, output: 1750.30 toks/s] Processed prompts: 54%\|█████▍ \| 69/128 [00:03<00:02, 20.33it/s, est. speed input: 5959.47 toks/s, output: 1875.48 toks/s] Processed prompts: 56%\|█████▋ \| 72/128 [00:03<00:03, 18.22it/s, est. speed input: 5829.53 toks/s, output: 1957.26 toks/s] Processed prompts: 59%\|█████▊ \| 75/128 [00:03<00:02, 20.09it/s, est. speed input: 5876.67 toks/s, output: 2106.01 toks/s] Processed prompts: 61%\|██████ \| 78/128 [00:03<00:02, 19.75it/s, est. speed input: 5843.43 toks/s, output: 2226.37 toks/s] Processed prompts: 63%\|██████▎ \| 81/128 [00:03<00:02, 15.95it/s, est. speed input: 5637.34 toks/s, output: 2288.18 toks/s] Processed prompts: 66%\|██████▋ \| 85/128 [00:04<00:03, 11.76it/s, est. speed input: 5239.90 toks/s, output: 2310.79 toks/s] Processed prompts: 69%\|██████▉ \| 88/128 [00:04<00:02, 13.94it/s, est. speed input: 5290.17 toks/s, output: 2492.78 toks/s] Processed prompts: 73%\|███████▎ \| 93/128 [00:04<00:02, 16.29it/s, est. speed input: 5320.19 toks/s, output: 2767.50 toks/s] Processed prompts: 74%\|███████▍ \| 95/128 [00:05<00:02, 11.88it/s, est. speed input: 5040.87 toks/s, output: 2738.37 toks/s]INFO 08-10 22:57:53 [metrics.py:417] Avg prompt throughput: 6095.1 tokens/s, Avg generation throughput: 5720.9 tokens/s, Running: 33 reqs, Swapped: 0 reqs, Pending: 0 reqs, GPU KV cache usage: 15.2%, CPU KV cache usage: 0.0%.
	INFO 08-10 22:57:53 [metrics.py:433] Prefix cache hit rate: GPU: 99.22%, CPU: 0.00%
	Processed prompts: 76%\|███████▌ \| 97/128 [00:05<00:02, 12.26it/s, est. speed input: 5005.74 toks/s, output: 2836.77 toks/s] Processed prompts: 77%\|███████▋ \| 99/128 [00:05<00:02, 13.27it/s, est. speed input: 5004.43 toks/s, output: 2952.10 toks/s] Processed prompts: 80%\|███████▉ \| 102/128 [00:05<00:01, 15.22it/s, est. speed input: 5026.02 toks/s, output: 3139.33 toks/s] Processed prompts: 83%\|████████▎ \| 106/128 [00:05<00:01, 16.27it/s, est. speed input: 5022.37 toks/s, output: 3369.64 toks/s] Processed prompts: 84%\|████████▍ \| 108/128 [00:06<00:02, 6.74it/s, est. speed input: 4401.64 toks/s, output: 3086.27 toks/s] Processed prompts: 86%\|████████▌ \| 110/128 [00:06<00:02, 7.94it/s, est. speed input: 4411.62 toks/s, output: 3232.95 toks/s] Processed prompts: 88%\|████████▊ \| 112/128 [00:06<00:01, 8.89it/s, est. speed input: 4397.15 toks/s, output: 3361.54 toks/s] Processed prompts: 90%\|████████▉ \| 115/128 [00:07<00:01, 7.17it/s, est. speed input: 4170.22 toks/s, output: 3392.85 toks/s] Processed prompts: 91%\|█████████▏\| 117/128 [00:07<00:01, 7.60it/s, est. speed input: 4123.61 toks/s, output: 3503.43 toks/s] Processed prompts: 93%\|█████████▎\| 119/128 [00:08<00:01, 4.79it/s, est. speed input: 3774.90 toks/s, output: 3366.62 toks/s] Processed prompts: 94%\|█████████▍\| 120/128 [00:09<00:02, 3.15it/s, est. speed input: 3444.15 toks/s, output: 3161.06 toks/s] Processed prompts: 95%\|█████████▍\| 121/128 [00:09<00:02, 3.36it/s, est. speed input: 3396.81 toks/s, output: 3207.68 toks/s] Processed prompts: 96%\|█████████▌\| 123/128 [00:09<00:01, 4.08it/s, est. speed input: 3352.00 toks/s, output: 3344.12 toks/s]INFO 08-10 22:57:58 [metrics.py:417] Avg prompt throughput: 0.0 tokens/s, Avg generation throughput: 2089.5 tokens/s, Running: 4 reqs, Swapped: 0 reqs, Pending: 0 reqs, GPU KV cache usage: 5.0%, CPU KV cache usage: 0.0%.
	INFO 08-10 22:57:58 [metrics.py:433] Prefix cache hit rate: GPU: 99.22%, CPU: 0.00%
	Processed prompts: 98%\|█████████▊\| 125/128 [00:10<00:00, 3.90it/s, est. speed input: 3225.26 toks/s, output: 3397.74 toks/s] Processed prompts: 99%\|█████████▉\| 127/128 [00:12<00:00, 1.98it/s, est. speed input: 2735.12 toks/s, output: 3065.90 toks/s] Processed prompts: 100%\|██████████\| 128/128 [00:12<00:00, 1.98it/s, est. speed input: 2753.94 toks/s, output: 3192.61 toks/s] Processed prompts: 100%\|██████████\| 128/128 [00:12<00:00, 10.20it/s, est. speed input: 2753.94 toks/s, output: 3192.61 toks/s]
	Adding requests: 0%\| \| 0/128 [00:00<?, ?it/s] Adding requests: 100%\|██████████\| 128/128 [00:00<00:00, 13102.72it/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 [00:00<00:26, 4.76it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 3%\|▎ \| 4/128 [00:00<00:17, 6.99it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 5%\|▌ \| 7/128 [00:00<00:16, 7.39it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 9%\|▊ \| 11/128 [00:01<00:13, 8.70it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 12%\|█▏ \| 15/128 [00:01<00:12, 9.36it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 14%\|█▍ \| 18/128 [00:02<00:12, 8.95it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 19%\|█▉ \| 24/128 [00:02<00:09, 11.14it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 20%\|██ \| 26/128 [00:02<00:11, 9.21it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 23%\|██▎ \| 30/128 [00:03<00:10, 9.50it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 25%\|██▌ \| 32/128 [00:03<00:11, 8.26it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 30%\|██▉ \| 38/128 [00:04<00:08, 10.37it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 32%\|███▏ \| 41/128 [00:04<00:09, 9.47it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 35%\|███▌ \| 45/128 [00:04<00:08, 9.84it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 39%\|███▉ \| 50/128 [00:05<00:07, 10.64it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 41%\|████ \| 52/128 [00:05<00:08, 9.00it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 45%\|████▌ \| 58/128 [00:06<00:06, 10.99it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 49%\|████▉ \| 63/128 [00:06<00:05, 11.56it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 52%\|█████▏ \| 67/128 [00:06<00:05, 11.23it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 56%\|█████▋ \| 72/128 [00:07<00:04, 11.78it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 59%\|█████▉ \| 76/128 [00:07<00:04, 11.49it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 65%\|██████▍ \| 83/128 [00:07<00:03, 13.46it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 67%\|██████▋ \| 86/128 [00:08<00:03, 11.81it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 70%\|██████▉ \| 89/128 [00:08<00:03, 10.70it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 73%\|███████▎ \| 93/128 [00:09<00:03, 10.60it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 77%\|███████▋ \| 99/128 [00:09<00:02, 12.10it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 81%\|████████▏ \| 104/128 [00:09<00:01, 12.19it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 83%\|████████▎ \| 106/128 [00:10<00:02, 10.23it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 87%\|████████▋ \| 111/128 [00:10<00:01, 11.08it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 91%\|█████████ \| 116/128 [00:10<00:01, 11.66it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 95%\|█████████▌\| 122/128 [00:11<00:00, 13.15it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 98%\|█████████▊\| 125/128 [00:11<00:00, 12.10it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 100%\|██████████\| 128/128 [00:11<00:00, 13.67it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 100%\|██████████\| 128/128 [00:11<00:00, 13.67it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 100%\|██████████\| 128/128 [00:11<00:00, 10.87it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s]
	INFO:search.beam_search:----------------------------------------------------------------------------------------------------
	INFO:search.beam_search:Iteration 0 completed beams: 0, skipped beams: 0, extended beams: 0, verifier beams: 0, total latency: 24.71s, length of agg_scores: [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1], num_steps: [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1], stop reasons: ['\n\n', '\n\n', '\n\n', '\n\n', '\n\n', '\n\n', '\n\n', '\n\n', '\n\n', '\n\n', '\n\n', '\n\n', '\n\n', '\n\n', '\n\n', '\n\n', '\n\n', '\n\n', '\n\n', '\n\n', '\n\n', '\n\n', '\n\n', '\n\n', '\n\n', '\n\n', '\n\n', '\n\n', '\n\n', '\n\n', '\n\n', '\n\n']
	Beam search iterations: 50%\|█████ \| 1/2 [00:24<00:24, 24.90s/it] Adding requests: 0%\| \| 0/128 [00:00<?, ?it/s] Adding requests: 23%\|██▎ \| 29/128 [00:00<00:00, 280.28it/s] Adding requests: 45%\|████▌ \| 58/128 [00:00<00:00, 269.95it/s] Adding requests: 67%\|██████▋ \| 86/128 [00:00<00:00, 269.67it/s] Adding requests: 88%\|████████▊ \| 113/128 [00:00<00:00, 256.90it/s] Adding requests: 100%\|██████████\| 128/128 [00:00<00:00, 255.67it/s]
	Processed prompts: 0%\| \| 0/128 [00:00<?, ?it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s]INFO 08-10 22:58:14 [metrics.py:417] Avg prompt throughput: 321.0 tokens/s, Avg generation throughput: 63.1 tokens/s, Running: 11 reqs, Swapped: 0 reqs, Pending: 117 reqs, GPU KV cache usage: 7.4%, CPU KV cache usage: 0.0%.
	INFO 08-10 22:58:14 [metrics.py:433] Prefix cache hit rate: GPU: 83.44%, CPU: 0.00%
	Processed prompts: 1%\| \| 1/128 [00:01<02:11, 1.03s/it, est. speed input: 1027.48 toks/s, output: 22.23 toks/s] Processed prompts: 2%\|▏ \| 2/128 [00:03<04:18, 2.06s/it, est. speed input: 658.71 toks/s, output: 46.26 toks/s] INFO 08-10 22:58:19 [metrics.py:417] Avg prompt throughput: 22905.8 tokens/s, Avg generation throughput: 5787.9 tokens/s, Running: 126 reqs, Swapped: 0 reqs, Pending: 0 reqs, GPU KV cache usage: 51.1%, CPU KV cache usage: 0.0%.
	INFO 08-10 22:58:19 [metrics.py:433] Prefix cache hit rate: GPU: 86.24%, CPU: 0.00%
	INFO 08-10 22:58:24 [metrics.py:417] Avg prompt throughput: 0.0 tokens/s, Avg generation throughput: 6870.3 tokens/s, Running: 126 reqs, Swapped: 0 reqs, Pending: 0 reqs, GPU KV cache usage: 84.6%, CPU KV cache usage: 0.0%.
	INFO 08-10 22:58:24 [metrics.py:433] Prefix cache hit rate: GPU: 86.24%, CPU: 0.00%
	WARNING 08-10 22:58:27 [scheduler.py:1834] Sequence group 255 is preempted by PreemptionMode.RECOMPUTE mode because there is not enough KV cache space. This can affect the end-to-end performance. Increase gpu_memory_utilization or tensor_parallel_size to provide more KV cache memory. total_num_cumulative_preemption=1
	Processed prompts: 2%\|▏ \| 3/128 [00:13<11:37, 5.58s/it, est. speed input: 293.74 toks/s, output: 61.52 toks/s]INFO 08-10 22:58:29 [metrics.py:417] Avg prompt throughput: 0.0 tokens/s, Avg generation throughput: 5810.7 tokens/s, Running: 106 reqs, Swapped: 0 reqs, Pending: 19 reqs, GPU KV cache usage: 99.4%, CPU KV cache usage: 0.0%.
	INFO 08-10 22:58:29 [metrics.py:433] Prefix cache hit rate: GPU: 86.24%, CPU: 0.00%
	INFO 08-10 22:58:34 [metrics.py:417] Avg prompt throughput: 0.0 tokens/s, Avg generation throughput: 5482.8 tokens/s, Running: 75 reqs, Swapped: 0 reqs, Pending: 50 reqs, GPU KV cache usage: 99.1%, CPU KV cache usage: 0.0%.
	INFO 08-10 22:58:34 [metrics.py:433] Prefix cache hit rate: GPU: 86.24%, CPU: 0.00%
	WARNING 08-10 22:58:34 [scheduler.py:1834] Sequence group 203 is preempted by PreemptionMode.RECOMPUTE mode because there is not enough KV cache space. This can affect the end-to-end performance. Increase gpu_memory_utilization or tensor_parallel_size to provide more KV cache memory. total_num_cumulative_preemption=51
	INFO 08-10 22:58:39 [metrics.py:417] Avg prompt throughput: 0.0 tokens/s, Avg generation throughput: 4252.6 tokens/s, Running: 56 reqs, Swapped: 0 reqs, Pending: 69 reqs, GPU KV cache usage: 98.8%, CPU KV cache usage: 0.0%.
	INFO 08-10 22:58:39 [metrics.py:433] Prefix cache hit rate: GPU: 86.24%, CPU: 0.00%
	INFO 08-10 22:58:44 [metrics.py:417] Avg prompt throughput: 0.0 tokens/s, Avg generation throughput: 3845.3 tokens/s, Running: 44 reqs, Swapped: 0 reqs, Pending: 81 reqs, GPU KV cache usage: 98.7%, CPU KV cache usage: 0.0%.
	INFO 08-10 22:58:44 [metrics.py:433] Prefix cache hit rate: GPU: 86.24%, CPU: 0.00%
	Processed prompts: 3%\|▎ \| 4/128 [00:33<23:25, 11.34s/it, est. speed input: 146.95 toks/s, output: 85.45 toks/s]INFO 08-10 22:58:49 [metrics.py:417] Avg prompt throughput: 27372.3 tokens/s, Avg generation throughput: 2144.1 tokens/s, Running: 65 reqs, Swapped: 0 reqs, Pending: 21 reqs, GPU KV cache usage: 98.0%, CPU KV cache usage: 0.0%.
	INFO 08-10 22:58:49 [metrics.py:433] Prefix cache hit rate: GPU: 65.96%, CPU: 0.00%
	Processed prompts: 34%\|███▎ \| 43/128 [00:36<00:46, 1.81it/s, est. speed input: 1135.06 toks/s, output: 2289.07 toks/s] Processed prompts: 34%\|███▍ \| 44/128 [00:36<00:46, 1.80it/s, est. speed input: 1134.22 toks/s, output: 2303.33 toks/s] Processed prompts: 35%\|███▌ \| 45/128 [00:37<00:46, 1.78it/s, est. speed input: 1133.92 toks/s, output: 2317.23 toks/s] Processed prompts: 36%\|███▌ \| 46/128 [00:38<00:46, 1.77it/s, est. speed input: 1153.92 toks/s, output: 2334.44 toks/s] Processed prompts: 37%\|███▋ \| 47/128 [00:38<00:46, 1.74it/s, est. speed input: 1169.83 toks/s, output: 2345.35 toks/s] Processed prompts: 38%\|███▊ \| 48/128 [00:39<00:45, 1.76it/s, est. speed input: 1173.05 toks/s, output: 2365.92 toks/s] Processed prompts: 38%\|███▊ \| 49/128 [00:39<00:44, 1.76it/s, est. speed input: 1179.12 toks/s, output: 2384.76 toks/s] Processed prompts: 39%\|███▉ \| 50/128 [00:40<00:42, 1.82it/s, est. speed input: 1180.12 toks/s, output: 2408.23 toks/s]INFO 08-10 22:58:54 [metrics.py:417] Avg prompt throughput: 812.7 tokens/s, Avg generation throughput: 4251.6 tokens/s, Running: 53 reqs, Swapped: 0 reqs, Pending: 25 reqs, GPU KV cache usage: 97.9%, CPU KV cache usage: 0.0%.
	INFO 08-10 22:58:54 [metrics.py:433] Prefix cache hit rate: GPU: 66.03%, CPU: 0.00%
	Processed prompts: 40%\|███▉ \| 51/128 [00:40<00:41, 1.87it/s, est. speed input: 1185.07 toks/s, output: 2430.77 toks/s] Processed prompts: 41%\|████ \| 52/128 [00:41<00:39, 1.95it/s, est. speed input: 1201.31 toks/s, output: 2455.19 toks/s] Processed prompts: 41%\|████▏ \| 53/128 [00:41<00:38, 1.97it/s, est. speed input: 1204.61 toks/s, output: 2475.47 toks/s] Processed prompts: 42%\|████▏ \| 54/128 [00:42<00:36, 2.03it/s, est. speed input: 1204.07 toks/s, output: 2498.00 toks/s] Processed prompts: 43%\|████▎ \| 55/128 [00:42<00:33, 2.17it/s, est. speed input: 1211.62 toks/s, output: 2524.21 toks/s] Processed prompts: 44%\|████▍ \| 56/128 [00:42<00:32, 2.25it/s, est. speed input: 1234.22 toks/s, output: 2548.33 toks/s] Processed prompts: 45%\|████▍ \| 57/128 [00:43<00:30, 2.31it/s, est. speed input: 1245.66 toks/s, output: 2571.99 toks/s] Processed prompts: 45%\|████▌ \| 58/128 [00:43<00:29, 2.40it/s, est. speed input: 1268.90 toks/s, output: 2596.86 toks/s] Processed prompts: 46%\|████▌ \| 59/128 [00:44<00:26, 2.60it/s, est. speed input: 1281.18 toks/s, output: 2625.15 toks/s] Processed prompts: 47%\|████▋ \| 60/128 [00:44<00:26, 2.57it/s, est. speed input: 1293.81 toks/s, output: 2647.86 toks/s] Processed prompts: 48%\|████▊ \| 61/128 [00:44<00:24, 2.78it/s, est. speed input: 1299.87 toks/s, output: 2676.47 toks/s] Processed prompts: 48%\|████▊ \| 62/128 [00:45<00:23, 2.78it/s, est. speed input: 1302.19 toks/s, output: 2700.50 toks/s] Processed prompts: 49%\|████▉ \| 63/128 [00:45<00:22, 2.87it/s, est. speed input: 1310.02 toks/s, output: 2726.47 toks/s]INFO 08-10 22:58:59 [metrics.py:417] Avg prompt throughput: 3518.1 tokens/s, Avg generation throughput: 3696.2 tokens/s, Running: 50 reqs, Swapped: 0 reqs, Pending: 15 reqs, GPU KV cache usage: 98.9%, CPU KV cache usage: 0.0%.
	INFO 08-10 22:58:59 [metrics.py:433] Prefix cache hit rate: GPU: 64.96%, CPU: 0.00%
	Processed prompts: 50%\|█████ \| 64/128 [00:45<00:22, 2.87it/s, est. speed input: 1321.13 toks/s, output: 2750.62 toks/s] Processed prompts: 51%\|█████ \| 65/128 [00:45<00:20, 3.15it/s, est. speed input: 1348.68 toks/s, output: 2780.44 toks/s] Processed prompts: 52%\|█████▏ \| 66/128 [00:46<00:19, 3.12it/s, est. speed input: 1355.75 toks/s, output: 2805.10 toks/s] Processed prompts: 52%\|█████▏ \| 67/128 [00:46<00:19, 3.20it/s, est. speed input: 1370.93 toks/s, output: 2831.26 toks/s] Processed prompts: 53%\|█████▎ \| 68/128 [00:46<00:17, 3.37it/s, est. speed input: 1384.08 toks/s, output: 2859.32 toks/s] Processed prompts: 54%\|█████▍ \| 69/128 [00:47<00:17, 3.46it/s, est. speed input: 1398.66 toks/s, output: 2886.30 toks/s] Processed prompts: 55%\|█████▍ \| 70/128 [00:47<00:17, 3.37it/s, est. speed input: 1415.57 toks/s, output: 2910.26 toks/s] Processed prompts: 55%\|█████▌ \| 71/128 [00:47<00:16, 3.53it/s, est. speed input: 1418.44 toks/s, output: 2937.94 toks/s] Processed prompts: 56%\|█████▋ \| 72/128 [00:47<00:15, 3.67it/s, est. speed input: 1424.60 toks/s, output: 2965.51 toks/s] Processed prompts: 57%\|█████▋ \| 73/128 [00:48<00:14, 3.71it/s, est. speed input: 1427.03 toks/s, output: 2991.88 toks/s] Processed prompts: 58%\|█████▊ \| 74/128 [00:48<00:14, 3.85it/s, est. speed input: 1448.83 toks/s, output: 3019.46 toks/s] Processed prompts: 59%\|█████▊ \| 75/128 [00:48<00:12, 4.11it/s, est. speed input: 1459.29 toks/s, output: 3048.87 toks/s] Processed prompts: 59%\|█████▉ \| 76/128 [00:48<00:11, 4.40it/s, est. speed input: 1468.44 toks/s, output: 3078.94 toks/s] Processed prompts: 60%\|██████ \| 77/128 [00:49<00:11, 4.54it/s, est. speed input: 1477.43 toks/s, output: 3107.89 toks/s] Processed prompts: 61%\|██████ \| 78/128 [00:49<00:10, 4.82it/s, est. speed input: 1500.96 toks/s, output: 3138.29 toks/s] Processed prompts: 62%\|██████▏ \| 79/128 [00:49<00:09, 5.05it/s, est. speed input: 1510.71 toks/s, output: 3168.62 toks/s] Processed prompts: 62%\|██████▎ \| 80/128 [00:49<00:09, 5.13it/s, est. speed input: 1522.85 toks/s, output: 3197.95 toks/s] Processed prompts: 63%\|██████▎ \| 81/128 [00:49<00:08, 5.36it/s, est. speed input: 1529.50 toks/s, output: 3228.38 toks/s] Processed prompts: 64%\|██████▍ \| 82/128 [00:49<00:08, 5.64it/s, est. speed input: 1539.87 toks/s, output: 3259.37 toks/s] Processed prompts: 65%\|██████▍ \| 83/128 [00:50<00:07, 6.12it/s, est. speed input: 1559.36 toks/s, output: 3291.77 toks/s] Processed prompts: 66%\|██████▌ \| 84/128 [00:50<00:07, 6.26it/s, est. speed input: 1569.12 toks/s, output: 3322.65 toks/s] Processed prompts: 66%\|██████▋ \| 85/128 [00:50<00:06, 6.39it/s, est. speed input: 1574.65 toks/s, output: 3353.51 toks/s] Processed prompts: 67%\|██████▋ \| 86/128 [00:50<00:06, 6.49it/s, est. speed input: 1585.29 toks/s, output: 3384.21 toks/s] Processed prompts: 68%\|██████▊ \| 87/128 [00:50<00:05, 7.24it/s, est. speed input: 1610.92 toks/s, output: 3417.95 toks/s]INFO 08-10 22:59:04 [metrics.py:417] Avg prompt throughput: 4812.9 tokens/s, Avg generation throughput: 3738.1 tokens/s, Running: 41 reqs, Swapped: 0 reqs, Pending: 0 reqs, GPU KV cache usage: 78.3%, CPU KV cache usage: 0.0%.
	INFO 08-10 22:59:04 [metrics.py:433] Prefix cache hit rate: GPU: 64.47%, CPU: 0.00%
	Processed prompts: 69%\|██████▉ \| 88/128 [00:50<00:05, 7.31it/s, est. speed input: 1626.21 toks/s, output: 3449.31 toks/s] Processed prompts: 70%\|██████▉ \| 89/128 [00:50<00:05, 7.41it/s, est. speed input: 1651.17 toks/s, output: 3480.76 toks/s] Processed prompts: 70%\|███████ \| 90/128 [00:50<00:04, 7.68it/s, est. speed input: 1665.62 toks/s, output: 3512.82 toks/s] Processed prompts: 71%\|███████ \| 91/128 [00:51<00:04, 8.13it/s, est. speed input: 1683.06 toks/s, output: 3545.59 toks/s] Processed prompts: 72%\|███████▏ \| 92/128 [00:51<00:04, 8.23it/s, est. speed input: 1691.82 toks/s, output: 3577.44 toks/s] Processed prompts: 73%\|███████▎ \| 94/128 [00:51<00:03, 9.01it/s, est. speed input: 1711.55 toks/s, output: 3643.40 toks/s] Processed prompts: 74%\|███████▍ \| 95/128 [00:51<00:03, 8.82it/s, est. speed input: 1726.21 toks/s, output: 3674.64 toks/s] Processed prompts: 76%\|███████▌ \| 97/128 [00:51<00:03, 9.75it/s, est. speed input: 1766.03 toks/s, output: 3741.50 toks/s] Processed prompts: 77%\|███████▋ \| 99/128 [00:51<00:02, 10.78it/s, est. speed input: 1800.95 toks/s, output: 3809.51 toks/s] Processed prompts: 79%\|███████▉ \| 101/128 [00:51<00:02, 11.16it/s, est. speed input: 1828.53 toks/s, output: 3876.05 toks/s] Processed prompts: 80%\|████████ \| 103/128 [00:52<00:02, 11.96it/s, est. speed input: 1845.30 toks/s, output: 3943.97 toks/s] Processed prompts: 82%\|████████▏ \| 105/128 [00:54<00:10, 2.09it/s, est. speed input: 1797.13 toks/s, output: 3829.91 toks/s] Processed prompts: 83%\|████████▎ \| 106/128 [00:55<00:10, 2.20it/s, est. speed input: 1803.49 toks/s, output: 3831.62 toks/s]INFO 08-10 22:59:09 [metrics.py:417] Avg prompt throughput: 0.0 tokens/s, Avg generation throughput: 2943.3 tokens/s, Running: 22 reqs, Swapped: 0 reqs, Pending: 0 reqs, GPU KV cache usage: 48.9%, CPU KV cache usage: 0.0%.
	INFO 08-10 22:59:09 [metrics.py:433] Prefix cache hit rate: GPU: 64.47%, CPU: 0.00%
	Processed prompts: 84%\|████████▎ \| 107/128 [00:56<00:11, 1.80it/s, est. speed input: 1786.26 toks/s, output: 3804.12 toks/s] Processed prompts: 84%\|████████▍ \| 108/128 [00:56<00:11, 1.71it/s, est. speed input: 1777.83 toks/s, output: 3794.44 toks/s] Processed prompts: 85%\|████████▌ \| 109/128 [00:57<00:13, 1.44it/s, est. speed input: 1771.10 toks/s, output: 3762.21 toks/s] Processed prompts: 86%\|████████▌ \| 110/128 [00:58<00:11, 1.56it/s, est. speed input: 1780.70 toks/s, output: 3765.99 toks/s] Processed prompts: 88%\|████████▊ \| 112/128 [00:58<00:06, 2.29it/s, est. speed input: 1798.68 toks/s, output: 3815.09 toks/s] Processed prompts: 88%\|████████▊ \| 113/128 [00:58<00:05, 2.60it/s, est. speed input: 1802.00 toks/s, output: 3835.85 toks/s] Processed prompts: 90%\|████████▉ \| 115/128 [00:59<00:04, 3.15it/s, est. speed input: 1821.45 toks/s, output: 3876.98 toks/s] Processed prompts: 91%\|█████████▏\| 117/128 [00:59<00:02, 4.54it/s, est. speed input: 1839.16 toks/s, output: 3939.31 toks/s] Processed prompts: 92%\|█████████▏\| 118/128 [00:59<00:02, 4.71it/s, est. speed input: 1846.53 toks/s, output: 3961.76 toks/s] Processed prompts: 93%\|█████████▎\| 119/128 [01:00<00:03, 2.97it/s, est. speed input: 1847.44 toks/s, output: 3945.86 toks/s] Processed prompts: 94%\|█████████▍\| 120/128 [01:00<00:02, 3.31it/s, est. speed input: 1856.87 toks/s, output: 3966.90 toks/s]INFO 08-10 22:59:14 [metrics.py:417] Avg prompt throughput: 0.0 tokens/s, Avg generation throughput: 1937.4 tokens/s, Running: 6 reqs, Swapped: 0 reqs, Pending: 0 reqs, GPU KV cache usage: 16.1%, CPU KV cache usage: 0.0%.
	INFO 08-10 22:59:14 [metrics.py:433] Prefix cache hit rate: GPU: 64.47%, CPU: 0.00%
	Processed prompts: 96%\|█████████▌\| 123/128 [01:00<00:00, 5.69it/s, est. speed input: 1889.53 toks/s, output: 4057.39 toks/s] Processed prompts: 97%\|█████████▋\| 124/128 [01:00<00:00, 5.86it/s, est. speed input: 1897.59 toks/s, output: 4081.02 toks/s] Processed prompts: 98%\|█████████▊\| 125/128 [01:00<00:00, 6.13it/s, est. speed input: 1909.15 toks/s, output: 4105.57 toks/s] Processed prompts: 98%\|█████████▊\| 126/128 [01:01<00:00, 6.25it/s, est. speed input: 1930.51 toks/s, output: 4128.99 toks/s] Processed prompts: 100%\|██████████\| 128/128 [01:01<00:00, 8.06it/s, est. speed input: 1956.64 toks/s, output: 4186.27 toks/s] Processed prompts: 100%\|██████████\| 128/128 [01:01<00:00, 8.06it/s, est. speed input: 1956.64 toks/s, output: 4186.27 toks/s] Processed prompts: 100%\|██████████\| 128/128 [01:01<00:00, 2.09it/s, est. speed input: 1956.64 toks/s, output: 4186.27 toks/s]
	Adding requests: 0%\| \| 0/128 [00:00<?, ?it/s] Adding requests: 100%\|██████████\| 128/128 [00:00<00:00, 5411.13it/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 [00:00<00:56, 2.25it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 2%\|▏ \| 2/128 [00:00<00:53, 2.38it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 2%\|▏ \| 3/128 [00:01<00:52, 2.38it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 3%\|▎ \| 4/128 [00:01<00:51, 2.40it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 4%\|▍ \| 5/128 [00:02<00:51, 2.40it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 5%\|▍ \| 6/128 [00:02<00:50, 2.40it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 5%\|▌ \| 7/128 [00:02<00:50, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 6%\|▋ \| 8/128 [00:03<00:49, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 7%\|▋ \| 9/128 [00:03<00:49, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 8%\|▊ \| 10/128 [00:04<00:48, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 9%\|▊ \| 11/128 [00:04<00:48, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 9%\|▉ \| 12/128 [00:04<00:48, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 10%\|█ \| 13/128 [00:05<00:47, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 11%\|█ \| 14/128 [00:05<00:47, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 12%\|█▏ \| 15/128 [00:06<00:46, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 12%\|█▎ \| 16/128 [00:06<00:46, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 13%\|█▎ \| 17/128 [00:07<00:45, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 14%\|█▍ \| 18/128 [00:07<00:45, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 15%\|█▍ \| 19/128 [00:07<00:45, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 16%\|█▌ \| 20/128 [00:08<00:44, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 16%\|█▋ \| 21/128 [00:08<00:44, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 17%\|█▋ \| 22/128 [00:09<00:43, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 18%\|█▊ \| 23/128 [00:09<00:43, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 19%\|█▉ \| 24/128 [00:09<00:43, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 20%\|█▉ \| 25/128 [00:10<00:42, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 20%\|██ \| 26/128 [00:10<00:42, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 21%\|██ \| 27/128 [00:11<00:41, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 22%\|██▏ \| 28/128 [00:11<00:41, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 23%\|██▎ \| 29/128 [00:12<00:41, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 23%\|██▎ \| 30/128 [00:12<00:40, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 24%\|██▍ \| 31/128 [00:12<00:40, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 25%\|██▌ \| 32/128 [00:13<00:39, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 26%\|██▌ \| 33/128 [00:13<00:39, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 27%\|██▋ \| 34/128 [00:14<00:38, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 27%\|██▋ \| 35/128 [00:14<00:38, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 28%\|██▊ \| 36/128 [00:14<00:38, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 29%\|██▉ \| 37/128 [00:15<00:37, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 30%\|██▉ \| 38/128 [00:15<00:37, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 30%\|███ \| 39/128 [00:16<00:36, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 31%\|███▏ \| 40/128 [00:16<00:36, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 32%\|███▏ \| 41/128 [00:17<00:36, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 33%\|███▎ \| 42/128 [00:17<00:35, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 34%\|███▎ \| 43/128 [00:17<00:35, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 34%\|███▍ \| 44/128 [00:18<00:34, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 35%\|███▌ \| 45/128 [00:18<00:34, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 36%\|███▌ \| 46/128 [00:19<00:33, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 37%\|███▋ \| 47/128 [00:19<00:33, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 38%\|███▊ \| 48/128 [00:19<00:33, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 38%\|███▊ \| 49/128 [00:20<00:32, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 39%\|███▉ \| 50/128 [00:20<00:32, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 40%\|███▉ \| 51/128 [00:21<00:31, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 41%\|████ \| 52/128 [00:21<00:31, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 41%\|████▏ \| 53/128 [00:21<00:31, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 42%\|████▏ \| 54/128 [00:22<00:30, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 43%\|████▎ \| 55/128 [00:22<00:30, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 44%\|████▍ \| 56/128 [00:23<00:29, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 45%\|████▍ \| 57/128 [00:23<00:29, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 45%\|████▌ \| 58/128 [00:24<00:29, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 46%\|████▌ \| 59/128 [00:24<00:28, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 47%\|████▋ \| 60/128 [00:24<00:28, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 48%\|████▊ \| 61/128 [00:25<00:27, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 48%\|████▊ \| 62/128 [00:25<00:27, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 49%\|████▉ \| 63/128 [00:26<00:26, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 50%\|█████ \| 64/128 [00:26<00:26, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 51%\|█████ \| 65/128 [00:26<00:26, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 52%\|█████▏ \| 66/128 [00:27<00:25, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 52%\|█████▏ \| 67/128 [00:27<00:25, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 53%\|█████▎ \| 68/128 [00:28<00:24, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 54%\|█████▍ \| 69/128 [00:28<00:24, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 55%\|█████▍ \| 70/128 [00:29<00:24, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 55%\|█████▌ \| 71/128 [00:29<00:23, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 56%\|█████▋ \| 72/128 [00:29<00:23, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 57%\|█████▋ \| 73/128 [00:30<00:22, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 58%\|█████▊ \| 74/128 [00:30<00:22, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 59%\|█████▊ \| 75/128 [00:31<00:20, 2.63it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 59%\|█████▉ \| 76/128 [00:31<00:20, 2.56it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 60%\|██████ \| 77/128 [00:31<00:20, 2.52it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 61%\|██████ \| 78/128 [00:32<00:20, 2.48it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 62%\|██████▏ \| 79/128 [00:32<00:19, 2.46it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 62%\|██████▎ \| 80/128 [00:33<00:19, 2.45it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 63%\|██████▎ \| 81/128 [00:33<00:19, 2.44it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 64%\|██████▍ \| 82/128 [00:33<00:18, 2.43it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 65%\|██████▍ \| 83/128 [00:34<00:18, 2.42it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 66%\|██████▌ \| 84/128 [00:34<00:18, 2.42it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 66%\|██████▋ \| 85/128 [00:35<00:17, 2.43it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 67%\|██████▋ \| 86/128 [00:35<00:17, 2.42it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 68%\|██████▊ \| 87/128 [00:35<00:16, 2.42it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 69%\|██████▉ \| 88/128 [00:36<00:16, 2.42it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 70%\|██████▉ \| 89/128 [00:36<00:16, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 70%\|███████ \| 90/128 [00:37<00:15, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 71%\|███████ \| 91/128 [00:37<00:15, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 72%\|███████▏ \| 92/128 [00:38<00:14, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 73%\|███████▎ \| 93/128 [00:38<00:14, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 73%\|███████▎ \| 94/128 [00:38<00:14, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 74%\|███████▍ \| 95/128 [00:39<00:13, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 75%\|███████▌ \| 96/128 [00:39<00:13, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 76%\|███████▌ \| 97/128 [00:40<00:12, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 77%\|███████▋ \| 98/128 [00:40<00:12, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 77%\|███████▋ \| 99/128 [00:40<00:12, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 78%\|███████▊ \| 100/128 [00:41<00:11, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 79%\|███████▉ \| 101/128 [00:41<00:11, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 80%\|███████▉ \| 102/128 [00:42<00:10, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 80%\|████████ \| 103/128 [00:42<00:10, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 81%\|████████▏ \| 104/128 [00:43<00:09, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 82%\|████████▏ \| 105/128 [00:43<00:09, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 83%\|████████▎ \| 106/128 [00:43<00:09, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 84%\|████████▎ \| 107/128 [00:44<00:08, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 84%\|████████▍ \| 108/128 [00:44<00:06, 2.87it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 85%\|████████▌ \| 109/128 [00:44<00:07, 2.71it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 86%\|████████▌ \| 110/128 [00:45<00:06, 2.61it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 87%\|████████▋ \| 111/128 [00:45<00:06, 2.55it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 88%\|████████▊ \| 112/128 [00:46<00:06, 2.51it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 88%\|████████▊ \| 113/128 [00:46<00:06, 2.48it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 89%\|████████▉ \| 114/128 [00:46<00:05, 2.46it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 90%\|████████▉ \| 115/128 [00:47<00:05, 2.44it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 91%\|█████████ \| 116/128 [00:47<00:04, 2.43it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 91%\|█████████▏\| 117/128 [00:48<00:04, 2.43it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 92%\|█████████▏\| 118/128 [00:48<00:04, 2.42it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 93%\|█████████▎\| 119/128 [00:49<00:03, 2.42it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 94%\|█████████▍\| 120/128 [00:49<00:03, 2.42it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 95%\|█████████▍\| 121/128 [00:49<00:02, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 95%\|█████████▌\| 122/128 [00:50<00:02, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 96%\|█████████▌\| 123/128 [00:50<00:02, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 97%\|█████████▋\| 124/128 [00:51<00:01, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 98%\|█████████▊\| 125/128 [00:51<00:01, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 98%\|█████████▊\| 126/128 [00:51<00:00, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 99%\|█████████▉\| 127/128 [00:52<00:00, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 100%\|██████████\| 128/128 [00:52<00:00, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 100%\|██████████\| 128/128 [00:52<00:00, 2.41it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s] Processed prompts: 100%\|██████████\| 128/128 [00:52<00:00, 2.43it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s]
	INFO:search.beam_search:Early exit: 0 active, 128 completed
	Beam search iterations: 50%\|█████ \| 1/2 [02:21<02:21, 141.78s/it]
	INFO:__main__:
	==================================================
	INFO:__main__:RESULTS
	INFO:__main__:==================================================
	INFO:__main__:Total num tokens: 266546
	INFO:__main__:Effective num tokens: 351750
	INFO:__main__:Effective num tokens per step: 2748.046875
	INFO:__main__:Number of tokens in 1 completion: 2748.046875
	INFO:__main__:N completion tokens: 266546
	INFO:__main__:Total generator latency: 74.48s
	INFO:__main__:Total verifier latency: 66.11s
	INFO:__main__:N generator latency: 74.48s
	INFO:__main__:N verifier latency: 66.11s
	INFO:__main__:Goodput: 2502.02
	INFO:__main__:Per-token generator goodput: 19.55
	INFO:__main__:Completions: 128
	INFO:__main__:Completion time: 86.50s
	INFO:__main__:Number of steps in 1 completion: 9.7109375
	INFO:__main__:Extended tokens: [[], []]
	INFO:__main__:Cleaning up...
	[rank0]:[W810 23:00:13.531267861 ProcessGroupNCCL.cpp:1476] 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())
	INFO:models.vllm_wrapper:Generator model shutdown complete
	INFO:models.vllm_wrapper:Verifier model shutdown complete
	INFO:fasttts:FastTTS shutdown complete
	INFO:__main__:Experiment completed successfully!
	GPU 3: General Metrics for NVIDIA AD10x (any frequency)
	Generating '/tmp/nsys-report-b158.qdstrm'
	[1/8] [0% ] vllm_tts_N128.nsys-rep [1/8] [0% ] vllm_tts_N128.nsys-rep [1/8] [0% ] vllm_tts_N128.nsys-rep [1/8] [5% ] vllm_tts_N128.nsys-rep [1/8] [6% ] vllm_tts_N128.nsys-rep [1/8] [7% ] vllm_tts_N128.nsys-rep [1/8] [8% ] vllm_tts_N128.nsys-rep [1/8] [9% ] vllm_tts_N128.nsys-rep [1/8] [8% ] vllm_tts_N128.nsys-rep [1/8] [9% ] vllm_tts_N128.nsys-rep [1/8] [8% ] vllm_tts_N128.nsys-rep [1/8] [9% ] vllm_tts_N128.nsys-rep [1/8] [8% ] vllm_tts_N128.nsys-rep [1/8] [9% ] vllm_tts_N128.nsys-rep [1/8] [8% ] vllm_tts_N128.nsys-rep [1/8] [7% ] vllm_tts_N128.nsys-rep [1/8] [8% ] vllm_tts_N128.nsys-rep [1/8] [7% ] vllm_tts_N128.nsys-rep [1/8] [8% ] vllm_tts_N128.nsys-rep [1/8] [7% ] vllm_tts_N128.nsys-rep [1/8] [8% ] vllm_tts_N128.nsys-rep [1/8] [7% ] vllm_tts_N128.nsys-rep [1/8] [8% ] vllm_tts_N128.nsys-rep [1/8] [7% ] vllm_tts_N128.nsys-rep [1/8] [8% ] vllm_tts_N128.nsys-rep [1/8] [7% ] vllm_tts_N128.nsys-rep [1/8] [8% ] vllm_tts_N128.nsys-rep [1/8] [7% ] vllm_tts_N128.nsys-rep [1/8] [8% ] vllm_tts_N128.nsys-rep [1/8] [7% ] vllm_tts_N128.nsys-rep [1/8] [8% ] vllm_tts_N128.nsys-rep [1/8] [7% ] vllm_tts_N128.nsys-rep [1/8] [8% ] vllm_tts_N128.nsys-rep [1/8] [7% ] vllm_tts_N128.nsys-rep [1/8] [8% ] vllm_tts_N128.nsys-rep [1/8] [9% ] vllm_tts_N128.nsys-rep [1/8] [10% ] vllm_tts_N128.nsys-rep [1/8] [9% ] vllm_tts_N128.nsys-rep [1/8] [10% ] vllm_tts_N128.nsys-rep [1/8] [9% ] vllm_tts_N128.nsys-rep [1/8] [8% ] vllm_tts_N128.nsys-rep [1/8] [9% ] vllm_tts_N128.nsys-rep [1/8] [8% ] vllm_tts_N128.nsys-rep [1/8] [9% ] vllm_tts_N128.nsys-rep [1/8] [8% ] vllm_tts_N128.nsys-rep [1/8] [9% ] vllm_tts_N128.nsys-rep [1/8] [8% ] vllm_tts_N128.nsys-rep [1/8] [7% ] vllm_tts_N128.nsys-rep [1/8] [8% ] vllm_tts_N128.nsys-rep [1/8] [7% ] vllm_tts_N128.nsys-rep [1/8] [8% ] vllm_tts_N128.nsys-rep [1/8] [7% ] vllm_tts_N128.nsys-rep [1/8] [8% ] vllm_tts_N128.nsys-rep [1/8] [9% ] vllm_tts_N128.nsys-rep [1/8] [10% ] vllm_tts_N128.nsys-rep [1/8] [11% ] vllm_tts_N128.nsys-rep [1/8] [12% ] vllm_tts_N128.nsys-rep [1/8] [13% ] vllm_tts_N128.nsys-rep [1/8] [14% ] vllm_tts_N128.nsys-rep [1/8] [=15% ] vllm_tts_N128.nsys-rep [1/8] [=16% ] vllm_tts_N128.nsys-rep [1/8] [=17% ] vllm_tts_N128.nsys-rep [1/8] [==18% ] vllm_tts_N128.nsys-rep [1/8] [==19% ] vllm_tts_N128.nsys-rep [1/8] [==20% ] vllm_tts_N128.nsys-rep [1/8] [==21% ] vllm_tts_N128.nsys-rep [1/8] [===22% ] vllm_tts_N128.nsys-rep [1/8] [===23% ] vllm_tts_N128.nsys-rep [1/8] [===24% ] vllm_tts_N128.nsys-rep [1/8] [====25% ] vllm_tts_N128.nsys-rep [1/8] [====26% ] vllm_tts_N128.nsys-rep [1/8] [====27% ] vllm_tts_N128.nsys-rep [1/8] [====28% ] vllm_tts_N128.nsys-rep [1/8] [=====29% ] vllm_tts_N128.nsys-rep [1/8] [=====30% ] vllm_tts_N128.nsys-rep [1/8] [=====31% ] vllm_tts_N128.nsys-rep [1/8] [=====32% ] vllm_tts_N128.nsys-rep [1/8] [======33% ] vllm_tts_N128.nsys-rep [1/8] [======34% ] vllm_tts_N128.nsys-rep [1/8] [======35% ] vllm_tts_N128.nsys-rep [1/8] [=======36% ] vllm_tts_N128.nsys-rep [1/8] [=======37% ] vllm_tts_N128.nsys-rep [1/8] [=======38% ] vllm_tts_N128.nsys-rep [1/8] [=======39% ] vllm_tts_N128.nsys-rep [1/8] [========40% ] vllm_tts_N128.nsys-rep [1/8] [========41% ] vllm_tts_N128.nsys-rep [1/8] [========42% ] vllm_tts_N128.nsys-rep [1/8] [=========43% ] vllm_tts_N128.nsys-rep [1/8] [=========44% ] vllm_tts_N128.nsys-rep [1/8] [=========45% ] vllm_tts_N128.nsys-rep [1/8] [=========46% ] vllm_tts_N128.nsys-rep [1/8] [==========47% ] vllm_tts_N128.nsys-rep [1/8] [==========48% ] vllm_tts_N128.nsys-rep [1/8] [==========49% ] vllm_tts_N128.nsys-rep [1/8] [===========50% ] vllm_tts_N128.nsys-rep [1/8] [===========51% ] vllm_tts_N128.nsys-rep [1/8] [===========52% ] vllm_tts_N128.nsys-rep [1/8] [===========53% ] vllm_tts_N128.nsys-rep [1/8] [============54% ] vllm_tts_N128.nsys-rep [1/8] [============55% ] vllm_tts_N128.nsys-rep [1/8] [============56% ] vllm_tts_N128.nsys-rep [1/8] [============57% ] vllm_tts_N128.nsys-rep [1/8] [=============58% ] vllm_tts_N128.nsys-rep [1/8] [=============59% ] vllm_tts_N128.nsys-rep [1/8] [=============60% ] vllm_tts_N128.nsys-rep [1/8] [==============61% ] vllm_tts_N128.nsys-rep [1/8] [==============62% ] vllm_tts_N128.nsys-rep [1/8] [========================100%] vllm_tts_N128.nsys-rep [1/8] [========================100%] vllm_tts_N128.nsys-rep
	[2/8] [0% ] vllm_tts_N128.sqlite [2/8] [1% ] vllm_tts_N128.sqlite [2/8] [2% ] vllm_tts_N128.sqlite [2/8] [3% ] vllm_tts_N128.sqlite [2/8] [4% ] vllm_tts_N128.sqlite [2/8] [5% ] vllm_tts_N128.sqlite [2/8] [6% ] vllm_tts_N128.sqlite [2/8] [7% ] vllm_tts_N128.sqlite [2/8] [8% ] vllm_tts_N128.sqlite [2/8] [9% ] vllm_tts_N128.sqlite [2/8] [10% ] vllm_tts_N128.sqlite [2/8] [11% ] vllm_tts_N128.sqlite [2/8] [12% ] vllm_tts_N128.sqlite [2/8] [13% ] vllm_tts_N128.sqlite [2/8] [14% ] vllm_tts_N128.sqlite [2/8] [=15% ] vllm_tts_N128.sqlite [2/8] [=16% ] vllm_tts_N128.sqlite [2/8] [=17% ] vllm_tts_N128.sqlite [2/8] [==18% ] vllm_tts_N128.sqlite [2/8] [==19% ] vllm_tts_N128.sqlite [2/8] [==20% ] vllm_tts_N128.sqlite [2/8] [==21% ] vllm_tts_N128.sqlite [2/8] [===22% ] vllm_tts_N128.sqlite [2/8] [===23% ] vllm_tts_N128.sqlite [2/8] [===24% ] vllm_tts_N128.sqlite [2/8] [====25% ] vllm_tts_N128.sqlite [2/8] [====26% ] vllm_tts_N128.sqlite [2/8] [====27% ] vllm_tts_N128.sqlite [2/8] [====28% ] vllm_tts_N128.sqlite [2/8] [=====29% ] vllm_tts_N128.sqlite [2/8] [=====30% ] vllm_tts_N128.sqlite [2/8] [=====31% ] vllm_tts_N128.sqlite [2/8] [=====32% ] vllm_tts_N128.sqlite [2/8] [======33% ] vllm_tts_N128.sqlite [2/8] [======34% ] vllm_tts_N128.sqlite [2/8] [======35% ] vllm_tts_N128.sqlite [2/8] [=======36% ] vllm_tts_N128.sqlite [2/8] [=======37% ] vllm_tts_N128.sqlite [2/8] [=======38% ] vllm_tts_N128.sqlite [2/8] [=======39% ] vllm_tts_N128.sqlite [2/8] [========40% ] vllm_tts_N128.sqlite [2/8] [========41% ] vllm_tts_N128.sqlite [2/8] [========42% ] vllm_tts_N128.sqlite [2/8] [=========43% ] vllm_tts_N128.sqlite [2/8] [=========44% ] vllm_tts_N128.sqlite [2/8] [=========45% ] vllm_tts_N128.sqlite [2/8] [=========46% ] vllm_tts_N128.sqlite [2/8] [==========47% ] vllm_tts_N128.sqlite [2/8] [==========48% ] vllm_tts_N128.sqlite [2/8] [==========49% ] vllm_tts_N128.sqlite [2/8] [===========50% ] vllm_tts_N128.sqlite [2/8] [===========51% ] vllm_tts_N128.sqlite [2/8] [===========52% ] vllm_tts_N128.sqlite [2/8] [===========53% ] vllm_tts_N128.sqlite [2/8] [============54% ] vllm_tts_N128.sqlite [2/8] [============55% ] vllm_tts_N128.sqlite [2/8] [============56% ] vllm_tts_N128.sqlite [2/8] [============57% ] vllm_tts_N128.sqlite [2/8] [=============58% ] vllm_tts_N128.sqlite [2/8] [=============59% ] vllm_tts_N128.sqlite [2/8] [=============60% ] vllm_tts_N128.sqlite [2/8] [==============61% ] vllm_tts_N128.sqlite [2/8] [==============62% ] vllm_tts_N128.sqlite [2/8] [==============63% ] vllm_tts_N128.sqlite [2/8] [==============64% ] vllm_tts_N128.sqlite [2/8] [===============65% ] vllm_tts_N128.sqlite [2/8] [===============66% ] vllm_tts_N128.sqlite [2/8] [===============67% ] vllm_tts_N128.sqlite [2/8] [================68% ] vllm_tts_N128.sqlite [2/8] [================69% ] vllm_tts_N128.sqlite [2/8] [================70% ] vllm_tts_N128.sqlite [2/8] [================71% ] vllm_tts_N128.sqlite [2/8] [=================72% ] vllm_tts_N128.sqlite [2/8] [=================73% ] vllm_tts_N128.sqlite [2/8] [=================74% ] vllm_tts_N128.sqlite [2/8] [==================75% ] vllm_tts_N128.sqlite [2/8] [==================76% ] vllm_tts_N128.sqlite [2/8] [==================77% ] vllm_tts_N128.sqlite [2/8] [==================78% ] vllm_tts_N128.sqlite [2/8] [===================79% ] vllm_tts_N128.sqlite [2/8] [===================80% ] vllm_tts_N128.sqlite [2/8] [===================81% ] vllm_tts_N128.sqlite [2/8] [===================82% ] vllm_tts_N128.sqlite [2/8] [====================83% ] vllm_tts_N128.sqlite [2/8] [====================84% ] vllm_tts_N128.sqlite [2/8] [====================85% ] vllm_tts_N128.sqlite [2/8] [=====================86% ] vllm_tts_N128.sqlite [2/8] [=====================87% ] vllm_tts_N128.sqlite [2/8] [=====================88% ] vllm_tts_N128.sqlite [2/8] [=====================89% ] vllm_tts_N128.sqlite [2/8] [======================90% ] vllm_tts_N128.sqlite [2/8] [======================91% ] vllm_tts_N128.sqlite [2/8] [======================92% ] vllm_tts_N128.sqlite [2/8] [=======================93% ] vllm_tts_N128.sqlite [2/8] [=======================94% ] vllm_tts_N128.sqlite [2/8] [=======================95% ] vllm_tts_N128.sqlite [2/8] [=======================96% ] vllm_tts_N128.sqlite [2/8] [========================97% ] vllm_tts_N128.sqlite [2/8] [========================98% ] vllm_tts_N128.sqlite [2/8] [========================99% ] vllm_tts_N128.sqlite [2/8] [========================100%] vllm_tts_N128.sqlite [2/8] [========================100%] vllm_tts_N128.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
	-------- --------------- --------- ----------------- ----------------- --------------- --------------- ---------------- ------- ----------------------------------
	50.5 141,803,724,946 1 141,803,724,946.0 141,803,724,946.0 141,803,724,946 141,803,724,946 0.0 PushPop :Total
	26.5 74,476,022,907 2 37,238,011,453.5 37,238,011,453.5 12,678,630,307 61,797,392,600 34,732,209,900.9 PushPop :generate
	23.0 64,583,725,604 2 32,291,862,802.0 32,291,862,802.0 11,793,244,788 52,790,480,816 28,989,423,605.3 PushPop :encode
	0.0 53,367 1 53,367.0 53,367.0 53,367 53,367 0.0 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
	-------- ----------------- --------- ---------------- ---------------- --------- --------------- --------------- ----------------------
	25.7 1,941,479,454,755 116,719 16,633,791.0 10,063,035.0 1,023 148,499,572,210 675,419,007.0 epoll_wait
	23.5 1,772,887,224,253 14,730 120,358,942.6 100,066,455.0 1,037 1,000,148,689 124,537,506.9 pthread_cond_timedwait
	20.9 1,574,149,969,306 99 15,900,504,740.5 13,961,263,322.0 16,812 51,678,490,869 5,611,442,800.5 pthread_cond_wait
	13.3 1,001,362,924,708 112 8,940,740,399.2 10,000,078,406.0 9,083 10,000,142,064 3,013,334,036.6 sem_timedwait
	9.6 723,946,854,221 41,491 17,448,286.5 3,264.0 1,000 130,626,983,539 1,084,287,776.0 read
	6.1 460,740,128,068 3,764 122,407,047.8 100,126,934.5 1,000 64,555,894,578 1,144,152,435.7 poll
	0.9 64,797,137,160 187 346,508,754.9 413,941,739.0 18,482 443,355,547 143,486,448.6 sem_wait
	0.0 3,176,441,893 3,863 822,273.3 6,893.0 1,007 108,363,729 8,962,038.9 ioctl
	0.0 967,884,946 76 12,735,328.2 6,059.0 1,079 895,352,237 102,805,784.1 waitpid
	0.0 448,005,107 523 856,606.3 2,662.0 1,079 45,001,465 4,149,922.8 fopen
	0.0 393,087,798 148,819 2,641.4 1,374.0 1,000 99,523,973 258,012.7 munmap
	0.0 202,633,892 40 5,065,847.3 5,062,881.0 5,055,007 5,087,267 8,269.3 nanosleep
	0.0 159,140,345 46,552 3,418.6 2,742.0 1,000 10,079,327 48,467.0 open64
	0.0 127,566,372 150 850,442.5 3,822.0 1,016 19,516,083 3,848,267.1 open
	0.0 59,141,239 10 5,914,123.9 34,776.5 7,512 58,810,892 18,586,041.9 connect
	0.0 57,542,754 3 19,180,918.0 568,836.0 532,266 56,441,652 32,268,747.4 fork
	0.0 51,806,716 374 138,520.6 4,753.5 1,972 17,440,224 1,486,212.1 fopen64
	0.0 42,809,293 99 432,417.1 12,738.0 7,909 8,558,636 1,312,769.1 pthread_join
	0.0 41,848,972 413 101,329.2 100,567.0 1,078 2,409,262 144,466.1 recv
	0.0 35,322,952 1,889 18,699.3 6,568.0 1,002 4,547,696 113,289.8 write
	0.0 32,264,515 245 131,691.9 66,638.0 53,003 8,056,575 624,679.7 sleep
	0.0 31,607,471 10,011 3,157.3 1,619.0 1,000 1,099,117 12,506.8 mmap64
	0.0 22,840,430 56 407,864.8 248,715.0 3,536 2,419,489 457,381.4 pthread_rwlock_wrlock
	0.0 16,481,996 622 26,498.4 8,016.0 1,566 90,344 27,850.6 send
	0.0 15,987,582 215 74,360.8 56,107.0 16,237 612,014 69,392.5 pthread_create
	0.0 10,101,707 1,521 6,641.5 2,201.0 1,017 101,266 10,869.6 fgets
	0.0 3,476,630 16 217,289.4 141,790.0 11,490 643,941 238,326.0 pthread_rwlock_rdlock
	0.0 2,867,942 2,338 1,226.7 1,062.0 1,002 8,678 584.7 fclose
	0.0 2,577,901 56 46,033.9 26,536.5 1,010 546,159 92,788.8 pthread_mutex_lock
	0.0 2,503,703 147 17,032.0 3,020.0 1,715 366,815 53,028.7 futex
	0.0 1,272,757 196 6,493.7 3,136.0 1,097 49,671 6,187.3 mmap
	0.0 1,002,479 318 3,152.4 2,686.0 1,006 23,403 2,197.9 pthread_cond_signal
	0.0 834,077 426 1,957.9 1,776.0 1,001 8,404 868.9 epoll_ctl
	0.0 525,888 102 5,155.8 3,960.5 1,784 20,000 3,034.4 pipe2
	0.0 241,411 42 5,747.9 4,102.5 1,484 17,172 4,415.0 socket
	0.0 189,797 34 5,582.3 2,770.0 1,066 20,582 5,825.2 pthread_cond_broadcast
	0.0 178,112 18 9,895.1 3,059.0 1,035 55,142 15,511.4 bind
	0.0 79,208 30 2,640.3 2,312.0 1,047 7,135 1,464.2 stat
	0.0 66,170 38 1,741.3 1,783.0 1,000 2,655 406.9 sigaction
	0.0 65,782 32 2,055.7 2,310.5 1,020 3,704 736.9 dup2
	0.0 59,236 14 4,231.1 4,867.0 1,068 6,335 1,802.7 fflush
	0.0 56,086 5 11,217.2 8,529.0 3,937 19,102 6,362.9 accept4
	0.0 46,967 29 1,619.6 1,323.0 1,024 5,384 846.5 fcntl
	0.0 44,892 8 5,611.5 5,555.5 4,480 6,739 908.7 lstat
	0.0 43,260 16 2,703.8 2,010.0 1,299 6,197 1,485.1 pthread_mutex_trylock
	0.0 39,551 18 2,197.3 1,735.5 1,028 4,507 963.0 pread
	0.0 33,643 2 16,821.5 16,821.5 13,867 19,776 4,178.3 socketpair
	0.0 31,239 7 4,462.7 4,464.0 3,874 5,090 454.1 fputs_unlocked
	0.0 29,488 8 3,686.0 3,236.0 2,146 6,132 1,342.2 flock
	0.0 25,590 5 5,118.0 3,799.0 3,396 7,755 2,114.6 fread
	0.0 19,988 8 2,498.5 2,367.0 1,953 3,141 440.9 mprotect
	0.0 17,774 3 5,924.7 4,924.0 1,788 11,062 4,717.3 fwrite
	0.0 12,119 9 1,346.6 1,199.0 1,027 2,223 397.4 listen
	0.0 11,228 6 1,871.3 1,685.5 1,184 3,264 726.5 fstat
	0.0 10,185 2 5,092.5 5,092.5 4,284 5,901 1,143.4 fputs
	0.0 9,908 1 9,908.0 9,908.0 9,908 9,908 0.0 kill
	0.0 5,360 3 1,786.7 1,629.0 1,243 2,488 637.3 openat64

	[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
	-------- --------------- --------- ------------ ----------- --------- ----------- ------------ ------------------------------------------
	82.3 27,420,409,463 139,503 196,557.8 4,922.0 2,784 99,659,704 1,188,464.0 cudaMemcpyAsync
	9.1 3,028,081,952 88 34,410,022.2 15,982.0 3,080 111,028,841 50,684,262.7 cudaHostAlloc
	3.3 1,112,336,088 112,601 9,878.6 5,998.0 760 63,708,752 315,579.4 cudaLaunchKernel
	2.9 965,884,935 6,003 160,900.4 166,436.0 60,695 1,303,107 74,679.3 cudaGraphLaunch_v10000
	0.4 145,465,680 10 14,546,568.0 36,689.0 8,205 145,150,716 45,889,628.7 cudaMemGetInfo
	0.4 122,827,552 96,849 1,268.2 1,026.0 579 566,356 3,507.5 cudaEventRecord
	0.3 110,809,697 96,842 1,144.2 814.0 369 36,341 973.5 cudaEventQuery
	0.2 79,445,889 14,582 5,448.2 5,501.5 652 121,802 2,584.5 cuLaunchKernel
	0.2 77,826,886 35 2,223,625.3 2,001,631.0 1,516,338 3,541,954 596,925.2 cudaGraphInstantiateWithFlags_v11040
	0.2 61,306,892 11,317 5,417.2 5,888.0 201 286,592 4,004.7 cudaMemsetAsync
	0.1 42,385,448 35 1,211,012.8 1,152,981.0 982,181 1,559,071 164,576.0 cudaGraphExecDestroy_v10000
	0.1 35,999,483 65 553,838.2 247,448.0 69,858 2,163,341 589,145.5 cudaFree
	0.1 27,096,581 6,636 4,083.3 3,511.0 1,465 58,130 3,570.4 cudaStreamSynchronize
	0.1 25,818,395 10 2,581,839.5 2,549,325.5 64,926 4,669,972 1,473,880.1 cuLibraryLoadData
	0.1 23,168,509 14,582 1,588.8 650.0 288 9,306,829 78,106.1 cuKernelGetFunction
	0.1 21,200,719 177 119,778.1 107,198.0 4,093 454,763 54,950.7 cudaMalloc
	0.0 14,647,093 18,320 799.5 923.0 274 10,787 389.8 cudaStreamIsCapturing_v10000
	0.0 5,349,186 35 152,833.9 150,021.0 127,819 187,165 15,387.6 cudaGraphDestroy_v10000
	0.0 4,929,954 8,785 561.2 549.0 312 7,429 182.4 cudaStreamGetCaptureInfo_v2_v11030
	0.0 4,558,131 128 35,610.4 3,766.5 2,805 1,477,648 179,500.2 cudaStreamCreateWithPriority
	0.0 4,315,032 35 123,286.6 117,122.0 95,415 211,554 20,641.4 cudaStreamEndCapture_v10000
	0.0 2,234,046 106 21,075.9 20,488.0 2,892 161,495 22,181.3 cudaDeviceSynchronize
	0.0 944,906 35 26,997.3 26,922.0 16,629 31,524 2,658.6 cudaGraphGetNodes_v10000
	0.0 462,996 35 13,228.5 10,520.0 8,175 25,261 4,946.4 cudaStreamBeginCapture_v10000
	0.0 134,382 810 165.9 131.0 69 2,107 128.8 cuGetProcAddress_v2
	0.0 41,343 26 1,590.1 366.0 298 19,293 3,799.6 cudaEventCreateWithFlags
	0.0 23,421 16 1,463.8 962.0 472 4,866 1,412.0 cuLibraryGetKernel
	0.0 4,605 8 575.6 517.5 386 1,087 218.7 cudaThreadExchangeStreamCaptureMode_v10010
	0.0 4,236 1 4,236.0 4,236.0 4,236 4,236 0.0 cudaStreamWaitEvent
	0.0 3,920 3 1,306.7 1,189.0 1,006 1,725 373.7 cuInit
	0.0 1,910 1 1,910.0 1,910.0 1,910 1,910 0.0 cudaEventDestroy
	0.0 1,326 3 442.0 112.0 108 1,106 575.0 cuModuleGetLoadingMode
	0.0 960 2 480.0 480.0 320 640 226.3 cudaGetDriverEntryPoint_v11030

	[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
	-------- --------------- --------- ----------- ----------- --------- --------- ----------- ----------------------------------------------------------------------------------------------------
	27.7 2,366,277,220 8,991 263,182.9 119,520.0 7,776 566,661 236,020.4 void cutlass::Kernel2<cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_32x6_tn_align8>(T1::Param…
	9.5 809,656,476 880 920,064.2 1,047,944.5 416,964 1,331,243 305,956.1 ampere_bf16_s16816gemm_bf16_128x64_ldg8_f2f_tn
	8.2 698,929,994 1,646 424,623.3 565,188.0 33,440 803,619 201,452.7 ampere_bf16_s1688gemm_bf16_64x128_sliced1x2_ldg8_f2f_tn
	6.8 581,528,436 2,002 290,473.7 493,923.0 10,496 526,212 233,548.2 void cutlass::Kernel2<cutlass_80_wmma_tensorop_bf16_s161616gemm_bf16_16x16_128x2_tn_align8>(T1::Par…
	6.1 521,624,103 1,876 278,051.2 128,096.5 41,312 713,251 241,965.8 ampere_bf16_s1688gemm_bf16_128x64_sliced1x2_ldg8_f2f_tn
	5.4 463,779,829 6,053 76,619.8 40,864.0 25,120 580,322 83,406.8 void at::native::<unnamed>::cunn_SoftMaxForward<(int)4, float, float, float, at::native::<unnamed>:…
	4.8 406,203,527 644 630,750.8 715,126.0 40,449 1,411,461 571,104.4 ampere_bf16_s1688gemm_bf16_128x128_ldg8_f2f_stages_32x1_tn
	4.3 370,145,293 6,053 61,150.7 27,840.0 1,440 497,922 73,912.8 void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, floa…
	4.2 357,709,876 6,054 59,086.5 32,160.0 29,568 633,923 50,489.5 void at::native::<unnamed>::cunn_SoftMaxForward<(int)4, float, float, float, at::native::<unnamed>:…
	4.0 341,552,571 6,053 56,427.0 52,384.0 2,880 333,217 51,601.7 void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void …
	3.4 288,580,611 1,344 214,717.7 197,345.5 54,977 427,652 100,049.1 void flash::flash_fwd_splitkv_kernel<Flash_fwd_kernel_traits<(int)128, (int)64, (int)128, (int)4, (…
	2.8 238,880,719 6,053 39,464.8 26,880.0 2,815 334,337 46,280.7 void at::native::elementwise_kernel<(int)128, (int)4, void at::native::gpu_kernel_impl<at::native::…
	2.5 211,356,505 3,360 62,903.7 6,145.0 3,232 248,513 84,563.1 void vllm::act_and_mul_kernel<c10::BFloat16, &vllm::silu_kernel<c10::BFloat16>, (bool)1>(T1 *, cons…
	2.2 188,333,659 6,053 31,114.1 29,952.0 2,335 177,569 26,431.2 void at::native::<unnamed>::distribution_elementwise_grid_stride_kernel<float, (int)4, void at::nat…
	1.9 162,336,044 6,053 26,819.1 20,384.0 3,423 220,896 23,923.0 void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIterator…
	1.3 111,387,694 6,053 18,402.1 14,432.0 5,152 203,136 15,527.9 void at::native::reduce_kernel<(int)512, (int)1, at::native::ReduceOp<float, at::native::ArgMaxOps<…
	1.0 87,789,134 173 507,451.6 507,107.0 506,339 534,755 2,184.9 void cutlass::Kernel2<cutlass_80_wmma_tensorop_bf16_s161616gemm_bf16_16x16_128x1_tn_align8>(T1::Par…
	0.6 47,877,355 6,720 7,124.6 3,135.0 1,664 34,432 7,371.8 std::enable_if<T2>(int)0&&vllm::_typeConvert<T1>::exists, void>::type vllm::fused_add_rms_norm_kern…
	0.5 41,880,516 308 135,975.7 151,761.0 76,064 168,064 27,431.8 ampere_bf16_s1688gemm_bf16_128x128_ldg8_relu_f2f_stages_32x1_tn
	0.3 26,672,559 280 95,259.1 100,048.5 26,432 145,921 46,062.0 ampere_bf16_s1688gemm_bf16_128x64_sliced1x2_ldg8_relu_f2f_tn
	0.2 21,054,279 3,360 6,266.2 2,304.0 1,695 37,472 6,476.6 void vllm::rotary_embedding_kernel<c10::BFloat16, (bool)1>(const long , T1 , T1 , const T1 , in…
	0.2 20,395,947 201 101,472.4 8,608.0 6,943 503,108 174,599.9 std::enable_if<!T7, void>::type internal::gemvx::kernel<int, int, __nv_bfloat16, __nv_bfloat16, __n…
	0.2 20,354,197 1,120 18,173.4 17,120.0 11,840 23,136 3,865.0 void flash::flash_fwd_splitkv_kernel<Flash_fwd_kernel_traits<(int)128, (int)64, (int)128, (int)4, (…
	0.2 16,763,120 448 37,417.7 37,376.0 35,681 38,976 530.3 void cutlass::Kernel2<cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x128_32x6_tn_align8>(T1::Para…
	0.2 14,523,485 4,851 2,993.9 2,720.0 2,144 32,896 2,200.7 void at::native::<unnamed>::indexSelectLargeIndex<c10::BFloat16, long, unsigned int, (int)2, (int)2…
	0.2 13,950,066 672 20,759.0 13,440.5 13,088 37,215 10,430.7 ampere_bf16_s16816gemm_bf16_64x64_ldg8_f2f_stages_64x5_tn
	0.2 12,894,344 952 13,544.5 13,408.0 12,224 15,744 973.0 ampere_bf16_s16816gemm_bf16_64x64_ldg8_relu_f2f_stages_64x5_tn
	0.1 11,024,273 3,332 3,308.6 1,152.0 960 13,312 3,375.4 void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0…
	0.1 10,447,935 3,446 3,031.9 1,056.0 832 109,920 9,518.1 void at::native::vectorized_elementwise_kernel<(int)8, at::native::FillFunctor<c10::BFloat16>, std:…
	0.1 10,203,785 6,054 1,685.5 1,632.0 1,248 2,656 168.9 void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIterator…
	0.1 8,252,852 6,003 1,374.8 1,376.0 1,120 1,920 95.8 void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::n…
	0.1 8,095,384 840 9,637.4 8,128.0 6,304 15,520 2,992.3 void flash::flash_fwd_splitkv_kernel<Flash_fwd_kernel_traits<(int)128, (int)64, (int)128, (int)4, (…
	0.1 7,168,283 112 64,002.5 63,920.5 62,752 65,952 757.4 void cutlass::Kernel2<cutlass_80_wmma_tensorop_bf16_s161616gemm_bf16_32x32_64x1_tn_align8>(T1::Para…
	0.1 6,180,439 3,024 2,043.8 1,887.0 1,344 3,072 549.4 void cublasLt::splitKreduce_kernel<(int)32, (int)16, int, __nv_bfloat16, __nv_bfloat16, float, (boo…
	0.1 4,969,066 1,322 3,758.7 3,616.5 1,568 6,336 1,312.2 void at::native::<unnamed>::indexSelectSmallIndex<c10::BFloat16, long, unsigned int, (int)2, (int)2…
	0.1 4,934,645 4 1,233,661.3 1,224,836.5 1,204,774 1,280,198 33,705.8 void at_cuda_detail::cub::DeviceSegmentedRadixSortKernel<at_cuda_detail::cub::DeviceRadixSortPolicy…
	0.1 4,845,822 224 21,633.1 21,456.0 9,504 34,496 11,939.4 void cutlass::Kernel2<cutlass_80_wmma_tensorop_bf16_s161616gemm_bf16_32x32_128x2_tn_align8>(T1::Par…
	0.0 2,879,343 56 51,416.8 51,424.0 49,505 53,088 701.3 void flash::flash_fwd_kernel<Flash_fwd_kernel_traits<(int)128, (int)64, (int)64, (int)4, (bool)0, (…
	0.0 2,559,051 2 1,279,525.5 1,279,525.5 1,239,814 1,319,237 56,160.5 void at_cuda_detail::cub::DeviceSegmentedRadixSortKernel<at_cuda_detail::cub::DeviceRadixSortPolicy…
	0.0 1,937,996 56 34,607.1 34,927.5 17,377 35,840 2,378.4 std::enable_if<!T7, void>::type internal::gemvx::kernel<int, int, __nv_bfloat16, float, float, floa…
	0.0 1,218,949 120 10,157.9 3,488.0 1,888 33,248 10,514.7 void vllm::rms_norm_kernel<c10::BFloat16>(T1 , const T1 , const T1 *, float, int, int)
	0.0 1,100,873 168 6,552.8 6,528.0 6,432 6,816 80.6 void flash::flash_fwd_splitkv_combine_kernel<Flash_fwd_kernel_traits<(int)128, (int)64, (int)128, (…
	0.0 920,420 1 920,420.0 920,420.0 920,420 920,420 0.0 void at::native::_scatter_gather_elementwise_kernel<(int)128, (int)8, void at::native::_cuda_scatte…
	0.0 882,082 280 3,150.3 3,136.0 2,785 3,425 211.0 void flash::flash_fwd_splitkv_combine_kernel<Flash_fwd_kernel_traits<(int)128, (int)64, (int)128, (…
	0.0 741,505 224 3,310.3 3,327.0 3,136 3,553 88.0 void flash::flash_fwd_splitkv_combine_kernel<Flash_fwd_kernel_traits<(int)128, (int)64, (int)128, (…
	0.0 679,618 2 339,809.0 339,809.0 339,201 340,417 859.8 void at::native::vectorized_elementwise_kernel<(int)4, at::native::<unnamed>::masked_fill_kernel(at…
	0.0 608,609 336 1,811.3 1,792.0 1,631 2,112 118.5 void cublasLt::splitKreduce_kernel<(int)32, (int)16, int, __nv_bfloat16, __nv_bfloat16, float, (boo…
	0.0 601,444 28 21,480.1 21,472.0 21,344 21,888 107.9 ampere_bf16_s16816gemm_bf16_128x64_ldg8_f2f_stages_32x6_tn
	0.0 354,689 1 354,689.0 354,689.0 354,689 354,689 0.0 void at::native::tensor_kernel_scan_innermost_dim<float, std::plus<float>>(T1 , const T1 , unsign…
	0.0 318,305 1 318,305.0 318,305.0 318,305 318,305 0.0 at::native::<unnamed>::fill_reverse_indices_kernel(long *, int, at::cuda::detail::IntDivider<unsign…
	0.0 317,864 112 2,838.1 2,848.0 2,689 3,040 44.2 void flash::flash_fwd_splitkv_combine_kernel<Flash_fwd_kernel_traits<(int)128, (int)64, (int)128, (…
	0.0 251,522 56 4,491.5 4,480.0 4,480 4,640 26.8 void flash::flash_fwd_splitkv_combine_kernel<Flash_fwd_kernel_traits<(int)128, (int)64, (int)128, (…
	0.0 233,505 1 233,505.0 233,505.0 233,505 233,505 0.0 void at::native::elementwise_kernel<(int)128, (int)4, void at::native::gpu_kernel_impl_nocast<at::n…
	0.0 222,561 1 222,561.0 222,561.0 222,561 222,561 0.0 void at::native::elementwise_kernel<(int)128, (int)4, void at::native::gpu_kernel_impl<at::native::…
	0.0 74,782 56 1,335.4 1,344.0 1,311 1,345 14.3 void cublasLt::splitKreduce_kernel<(int)32, (int)16, int, float, __nv_bfloat16, float, (bool)0, __n…
	0.0 65,629 73 899.0 896.0 863 1,408 74.0 void at::native::vectorized_elementwise_kernel<(int)2, at::native::FillFunctor<long>, std::array<ch…
	0.0 3,552 1 3,552.0 3,552.0 3,552 3,552 0.0 void at::native::<unnamed>::CatArrayBatchedCopy_aligned16_contig<at::native::<unnamed>::OpaqueType<…
	0.0 2,432 1 2,432.0 2,432.0 2,432 2,432 0.0 void at::native::_scatter_gather_elementwise_kernel<(int)128, (int)8, void at::native::_cuda_scatte…
	0.0 2,271 2 1,135.5 1,135.5 1,056 1,215 112.4 void <unnamed>::elementwise_kernel_with_index<int, at::native::arange_cuda_out(const c10::Scalar &,…
	0.0 2,208 1 2,208.0 2,208.0 2,208 2,208 0.0 void at::native::elementwise_kernel<(int)128, (int)4, void at::native::gpu_kernel_impl<at::native::…
	0.0 2,176 1 2,176.0 2,176.0 2,176 2,176 0.0 void at::native::vectorized_elementwise_kernel<(int)4, at::native::cos_kernel_cuda(at::TensorIterat…
	0.0 2,048 1 2,048.0 2,048.0 2,048 2,048 0.0 void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::n…
	0.0 1,856 1 1,856.0 1,856.0 1,856 1,856 0.0 void at::native::vectorized_elementwise_kernel<(int)4, at::native::sin_kernel_cuda(at::TensorIterat…
	0.0 1,728 1 1,728.0 1,728.0 1,728 1,728 0.0 void at::native::vectorized_elementwise_kernel<(int)8, at::native::bfloat16_copy_kernel_cuda(at::Te…
	0.0 1,473 1 1,473.0 1,473.0 1,473 1,473 0.0 void at::native::vectorized_elementwise_kernel<(int)4, at::native::BUnaryFunctor<float, float, floa…
	0.0 1,408 1 1,408.0 1,408.0 1,408 1,408 0.0 void at::native::elementwise_kernel<(int)128, (int)4, void at::native::gpu_kernel_impl_nocast<at::n…
	0.0 1,408 1 1,408.0 1,408.0 1,408 1,408 0.0 void at::native::vectorized_elementwise_kernel<(int)2, at::native::CUDAFunctorOnOther_add<long>, st…
	0.0 1,376 1 1,376.0 1,376.0 1,376 1,376 0.0 void at::native::vectorized_elementwise_kernel<(int)8, at::native::CUDAFunctorOnOther_add<c10::BFlo…
	0.0 1,184 1 1,184.0 1,184.0 1,184 1,184 0.0 void at::native::vectorized_elementwise_kernel<(int)4, at::native::reciprocal_kernel_cuda(at::Tenso…
	0.0 1,025 1 1,025.0 1,025.0 1,025 1,025 0.0 void at::native::vectorized_elementwise_kernel<(int)4, at::native::AUnaryFunctor<float, float, floa…
	0.0 895 1 895.0 895.0 895 895 0.0 void at::native::vectorized_elementwise_kernel<(int)2, at::native::FillFunctor<double>, std::array<…
	0.0 895 1 895.0 895.0 895 895 0.0 void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<int>, std::array<cha…

	[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
	-------- --------------- ------- -------- -------- -------- ---------- ----------- ------------------------------
	93.5 555,978,098 103,434 5,375.2 353.0 288 99,255,421 328,740.1 [CUDA memcpy Host-to-Device]
	4.6 27,417,635 30,016 913.4 896.0 831 343,297 1,976.7 [CUDA memcpy Device-to-Device]
	1.2 7,184,862 6,053 1,187.0 1,184.0 863 1,760 95.0 [CUDA memcpy Device-to-Host]
	0.7 4,237,758 9,217 459.8 352.0 320 1,632 219.9 [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
	---------- ------- -------- -------- -------- -------- ----------- ------------------------------
	3,434.549 103,434 0.033 0.000 0.000 466.747 1.653 [CUDA memcpy Host-to-Device]
	487.216 30,016 0.016 0.000 0.000 155.582 0.898 [CUDA memcpy Device-to-Device]
	6.572 9,217 0.001 0.000 0.000 0.003 0.001 [CUDA memset]
	2.375 6,053 0.000 0.000 0.000 0.002 0.000 [CUDA memcpy Device-to-Host]

	Generated:
	/data/cy/vllm_tts_N128.nsys-rep
	/data/cy/vllm_tts_N128.sqlite