Files
vllm-omni/tests/dfx/perf/tests/test_ltx2_vllm_omni.json
T

319 lines
8.1 KiB
JSON

[
{
"test_name": "test_ltx2_baseline_eager",
"mark": [
{
"hardware_marks": {
"res": {
"cuda": "H100"
},
"num_cards": 1
}
},
"local_model",
"diffusion"
],
"description": "Single-device baseline with enforce-eager (no torch.compile)",
"server_params": {
"model": "Lightricks/LTX-2",
"serve_args": {
"enforce-eager": true,
"enable-diffusion-pipeline-profiler": true
}
},
"benchmark_params": [
{
"name": "256x256_145f_steps6",
"dataset_name": "random",
"tokenizer": "gpt2",
"endpoint": "/v1/videos",
"num_prompts": 3,
"max_concurrency": 1,
"random_input_len": 8,
"random_output_len": 1,
"percentile_metrics": "e2el",
"extra_body": {
"width": 256,
"height": 256,
"num_frames": 145,
"fps": 24,
"num_inference_steps": 6,
"negative_prompt": "Negative prompt for benchmarking diffusion models"
}
},
{
"name": "480x768_41f_steps20",
"dataset_name": "random",
"tokenizer": "gpt2",
"endpoint": "/v1/videos",
"num_prompts": 3,
"max_concurrency": 1,
"random_input_len": 8,
"random_output_len": 1,
"percentile_metrics": "e2el",
"extra_body": {
"width": 768,
"height": 480,
"num_frames": 41,
"fps": 24,
"num_inference_steps": 20,
"negative_prompt": "Negative prompt for benchmarking diffusion models"
}
}
]
},
{
"test_name": "test_ltx2_torch_compile",
"mark": [
{
"hardware_marks": {
"res": {
"cuda": "H100"
},
"num_cards": 1
}
},
"local_model",
"diffusion"
],
"description": "Single-device with torch.compile (default, no enforce-eager)",
"server_params": {
"model": "Lightricks/LTX-2",
"serve_args": {
"enable-diffusion-pipeline-profiler": true
}
},
"benchmark_params": [
{
"name": "256x256_145f_steps6",
"dataset_name": "random",
"tokenizer": "gpt2",
"endpoint": "/v1/videos",
"num_prompts": 3,
"max_concurrency": 1,
"random_input_len": 8,
"random_output_len": 1,
"percentile_metrics": "e2el",
"extra_body": {
"width": 256,
"height": 256,
"num_frames": 145,
"fps": 24,
"num_inference_steps": 6,
"negative_prompt": "Negative prompt for benchmarking diffusion models"
}
},
{
"name": "480x768_41f_steps20",
"dataset_name": "random",
"tokenizer": "gpt2",
"endpoint": "/v1/videos",
"num_prompts": 3,
"max_concurrency": 1,
"random_input_len": 8,
"random_output_len": 1,
"percentile_metrics": "e2el",
"extra_body": {
"width": 768,
"height": 480,
"num_frames": 41,
"fps": 24,
"num_inference_steps": 20,
"negative_prompt": "Negative prompt for benchmarking diffusion models"
}
}
]
},
{
"test_name": "test_ltx2_cfg2_eager",
"mark": [
{
"hardware_marks": {
"res": {
"cuda": "H100"
},
"num_cards": 2
}
},
"local_model",
"diffusion"
],
"description": "CFG-parallel=2 with enforce-eager",
"server_params": {
"model": "Lightricks/LTX-2",
"serve_args": {
"cfg-parallel-size": 2,
"enforce-eager": true,
"enable-diffusion-pipeline-profiler": true
}
},
"benchmark_params": [
{
"name": "256x256_145f_steps6",
"dataset_name": "random",
"tokenizer": "gpt2",
"endpoint": "/v1/videos",
"num_prompts": 3,
"max_concurrency": 1,
"random_input_len": 8,
"random_output_len": 1,
"percentile_metrics": "e2el",
"extra_body": {
"width": 256,
"height": 256,
"num_frames": 145,
"fps": 24,
"num_inference_steps": 6,
"negative_prompt": "Negative prompt for benchmarking diffusion models"
}
},
{
"name": "480x768_41f_steps20",
"dataset_name": "random",
"tokenizer": "gpt2",
"endpoint": "/v1/videos",
"num_prompts": 3,
"max_concurrency": 1,
"random_input_len": 8,
"random_output_len": 1,
"percentile_metrics": "e2el",
"extra_body": {
"width": 768,
"height": 480,
"num_frames": 41,
"fps": 24,
"num_inference_steps": 20,
"negative_prompt": "Negative prompt for benchmarking diffusion models"
}
}
]
},
{
"test_name": "test_ltx2_cfg2_compile",
"mark": [
{
"hardware_marks": {
"res": {
"cuda": "H100"
},
"num_cards": 2
}
},
"local_model",
"diffusion"
],
"description": "CFG-parallel=2 with torch.compile",
"server_params": {
"model": "Lightricks/LTX-2",
"serve_args": {
"cfg-parallel-size": 2,
"enable-diffusion-pipeline-profiler": true
}
},
"benchmark_params": [
{
"name": "256x256_145f_steps6",
"dataset_name": "random",
"tokenizer": "gpt2",
"endpoint": "/v1/videos",
"num_prompts": 3,
"max_concurrency": 1,
"random_input_len": 8,
"random_output_len": 1,
"percentile_metrics": "e2el",
"extra_body": {
"width": 256,
"height": 256,
"num_frames": 145,
"fps": 24,
"num_inference_steps": 6,
"negative_prompt": "Negative prompt for benchmarking diffusion models"
}
},
{
"name": "480x768_41f_steps20",
"dataset_name": "random",
"tokenizer": "gpt2",
"endpoint": "/v1/videos",
"num_prompts": 3,
"max_concurrency": 1,
"random_input_len": 8,
"random_output_len": 1,
"percentile_metrics": "e2el",
"extra_body": {
"width": 768,
"height": 480,
"num_frames": 41,
"fps": 24,
"num_inference_steps": 20,
"negative_prompt": "Negative prompt for benchmarking diffusion models"
}
}
]
},
{
"test_name": "test_ltx2_cache_dit_eager",
"mark": [
{
"hardware_marks": {
"res": {
"cuda": "H100"
},
"num_cards": 1
}
},
"local_model",
"diffusion"
],
"description": "CacheDiT with enforce-eager",
"server_params": {
"model": "Lightricks/LTX-2",
"serve_args": {
"cache-backend": "cache_dit",
"enforce-eager": true,
"enable-diffusion-pipeline-profiler": true
}
},
"benchmark_params": [
{
"name": "256x256_145f_steps6",
"dataset_name": "random",
"tokenizer": "gpt2",
"endpoint": "/v1/videos",
"num_prompts": 3,
"max_concurrency": 1,
"random_input_len": 8,
"random_output_len": 1,
"percentile_metrics": "e2el",
"extra_body": {
"width": 256,
"height": 256,
"num_frames": 145,
"fps": 24,
"num_inference_steps": 6,
"negative_prompt": "Negative prompt for benchmarking diffusion models"
}
},
{
"name": "480x768_41f_steps20",
"dataset_name": "random",
"tokenizer": "gpt2",
"endpoint": "/v1/videos",
"num_prompts": 3,
"max_concurrency": 1,
"random_input_len": 8,
"random_output_len": 1,
"percentile_metrics": "e2el",
"extra_body": {
"width": 768,
"height": 480,
"num_frames": 41,
"fps": 24,
"num_inference_steps": 20,
"negative_prompt": "Negative prompt for benchmarking diffusion models"
}
}
]
}
]