352 lines
11 KiB
JSON
352 lines
11 KiB
JSON
{
|
|
"metadata": {
|
|
"model": "Diffusion Server",
|
|
"hardware": "CI H100 80GB pool",
|
|
"description": "Reference numbers captured from the CI diffusion server baseline run"
|
|
},
|
|
"tolerances": {
|
|
"e2e": 0.05,
|
|
"stage": 0.05,
|
|
"denoise_step": 0.15,
|
|
"denoise_agg": 0.08
|
|
},
|
|
"improvement_reporting": {
|
|
"threshold": 0.2
|
|
},
|
|
"sampling": {
|
|
"step_fractions": [
|
|
0.0,
|
|
0.2,
|
|
0.4,
|
|
0.6,
|
|
0.8,
|
|
1.0
|
|
],
|
|
"warmup_requests": {
|
|
"text": 1,
|
|
"image_edit": 0
|
|
}
|
|
},
|
|
"scenarios": {
|
|
"qwen_image_t2i": {
|
|
"notes": "Single-image generation using the default prompt",
|
|
"expected_e2e_ms": 74500.0,
|
|
"expected_avg_denoise_ms": 422.42,
|
|
"expected_median_denoise_ms": 410.62,
|
|
"stages_ms": {
|
|
"InputValidationStage": 0.1,
|
|
"TextEncodingStage": 834.2,
|
|
"ConditioningStage": 0.1,
|
|
"TimestepPreparationStage": 10.6,
|
|
"LatentPreparationStage": 11.8,
|
|
"DenoisingStage": 21202.6,
|
|
"DecodingStage": 476.12
|
|
},
|
|
"denoise_step_ms": {
|
|
"0": 1077.77,
|
|
"1": 345.13,
|
|
"2": 413.8,
|
|
"3": 405.49,
|
|
"4": 408.14,
|
|
"5": 409.06,
|
|
"6": 408.85,
|
|
"7": 410.53,
|
|
"8": 407.51,
|
|
"9": 409.44,
|
|
"10": 408.65,
|
|
"11": 410.14,
|
|
"12": 411.74,
|
|
"13": 409.59,
|
|
"14": 409.17,
|
|
"15": 410.78,
|
|
"16": 410.66,
|
|
"17": 410.58,
|
|
"18": 411.27,
|
|
"19": 410.51,
|
|
"20": 409.03,
|
|
"21": 410.16,
|
|
"22": 409.42,
|
|
"23": 411.03,
|
|
"24": 410.18,
|
|
"25": 409.72,
|
|
"26": 410.26,
|
|
"27": 410.21,
|
|
"28": 410.71,
|
|
"29": 410.76,
|
|
"30": 411.06,
|
|
"31": 410.1,
|
|
"32": 410.55,
|
|
"33": 410.77,
|
|
"34": 410.74,
|
|
"35": 411.75,
|
|
"36": 410.78,
|
|
"37": 411.56,
|
|
"38": 410.85,
|
|
"39": 411.08,
|
|
"40": 411.12,
|
|
"41": 411.1,
|
|
"42": 411.09,
|
|
"43": 410.87,
|
|
"44": 411.37,
|
|
"45": 411.68,
|
|
"46": 411.0,
|
|
"47": 410.09,
|
|
"48": 412.72,
|
|
"49": 410.42
|
|
}
|
|
},
|
|
"flux_image_t2i": {
|
|
"stages_ms": {
|
|
"InputValidationStage": 0.08,
|
|
"TextEncodingStage": 87.36,
|
|
"ConditioningStage": 0.024,
|
|
"TimestepPreparationStage": 2.53,
|
|
"LatentPreparationStage": 6.21,
|
|
"DenoisingStage": 8161.97,
|
|
"DecodingStage": 378.21
|
|
},
|
|
"denoise_step_ms": {
|
|
"0": 56.16,
|
|
"10": 163.54,
|
|
"20": 165.94,
|
|
"29": 165.26,
|
|
"39": 165.16,
|
|
"49": 165.36
|
|
},
|
|
"expected_e2e_ms": 8764.73,
|
|
"expected_avg_denoise_ms": 160.79,
|
|
"expected_median_denoise_ms": 165.14
|
|
},
|
|
"qwen_image_edit_ti2i": {
|
|
"notes": "single uploaded reference image, Qwen/Qwen-Image-Edit",
|
|
"expected_e2e_ms": 138500.0,
|
|
"expected_avg_denoise_ms": 720.0,
|
|
"expected_median_denoise_ms": 718.0,
|
|
"stages_ms": {
|
|
"InputValidationStage": 23,
|
|
"ImageEncodingStage": 1350.0,
|
|
"ImageVAEEncodingStage": 340.0,
|
|
"ConditioningStage": 0.13,
|
|
"TimestepPreparationStage": 13.78,
|
|
"LatentPreparationStage": 15.0,
|
|
"DenoisingStage": 36000.0,
|
|
"DecodingStage": 850.0
|
|
},
|
|
"denoise_step_ms": {
|
|
"0": 720.0,
|
|
"1": 720.0,
|
|
"2": 720.0,
|
|
"3": 720.0,
|
|
"4": 720.0,
|
|
"5": 720.0,
|
|
"6": 720.0,
|
|
"7": 720.0,
|
|
"8": 720.0,
|
|
"9": 720.0,
|
|
"10": 720.0,
|
|
"11": 720.0,
|
|
"12": 720.0,
|
|
"13": 720.0,
|
|
"14": 720.0,
|
|
"15": 720.0,
|
|
"16": 720.0,
|
|
"17": 720.0,
|
|
"18": 720.0,
|
|
"19": 720.0,
|
|
"20": 720.0,
|
|
"21": 720.0,
|
|
"22": 720.0,
|
|
"23": 720.0,
|
|
"24": 720.0,
|
|
"25": 720.0,
|
|
"26": 720.0,
|
|
"27": 720.0,
|
|
"28": 720.0,
|
|
"29": 720.0,
|
|
"30": 720.0,
|
|
"31": 720.0,
|
|
"32": 720.0,
|
|
"33": 720.0,
|
|
"34": 720.0,
|
|
"35": 720.0,
|
|
"36": 720.0,
|
|
"37": 720.0,
|
|
"38": 720.0,
|
|
"39": 720.0,
|
|
"40": 720.0,
|
|
"41": 720.0,
|
|
"42": 720.0,
|
|
"43": 720.0,
|
|
"44": 720.0,
|
|
"45": 720.0,
|
|
"46": 720.0,
|
|
"47": 720.0,
|
|
"48": 720.0,
|
|
"49": 720.0
|
|
}
|
|
},
|
|
"fastwan2_1_t2v": {
|
|
"notes": "Single-video generation using the default prompt",
|
|
"expected_e2e_ms": 95616.59,
|
|
"expected_avg_denoise_ms": 1798.77,
|
|
"expected_median_denoise_ms": 1786.78,
|
|
"stages_ms": {
|
|
"InputValidationStage": 1.03,
|
|
"TextEncodingStage": 3450.0,
|
|
"ConditioningStage": 1.0,
|
|
"TimestepPreparationStage": 6.0,
|
|
"LatentPreparationStage": 15.0,
|
|
"DenoisingStage": 90100.0,
|
|
"DecodingStage": 3650.0
|
|
},
|
|
"denoise_step_ms": {
|
|
"0": 3500.0,
|
|
"10": 1800.0,
|
|
"20": 1800.0,
|
|
"29": 1800.0,
|
|
"39": 1800.0,
|
|
"49": 1800.0
|
|
},
|
|
"frames_per_second": 0.51,
|
|
"total_frames": 49,
|
|
"avg_frame_time_ms": 1951.36
|
|
},
|
|
"wan2_2_i2v_a14b": {
|
|
"notes": "Wan-AI/Wan2.2-I2V-A14B",
|
|
"expected_e2e_ms": 282500.0,
|
|
"expected_avg_denoise_ms": 7000.0,
|
|
"expected_median_denoise_ms": 7000.19,
|
|
"stages_ms": {
|
|
"InputValidationStage": 26.26,
|
|
"TextEncodingStage": 2749.6,
|
|
"ConditioningStage": 2.0,
|
|
"TimestepPreparationStage": 2.0,
|
|
"LatentPreparationStage": 10.0,
|
|
"ImageVAEEncodingStage": 2031.0,
|
|
"DenoisingStage": 278000.0,
|
|
"DecodingStage": 2849.6
|
|
},
|
|
"denoise_step_ms": {
|
|
"0": 24000.0,
|
|
"8": 7000.0,
|
|
"16": 7000.0,
|
|
"23": 7000.0,
|
|
"31": 7000.0,
|
|
"39": 7000.0
|
|
}
|
|
},
|
|
|
|
"wan2_1_i2v_14b_480P": {
|
|
"stages_ms": {
|
|
"per_frame_generation": null
|
|
},
|
|
"denoise_step_ms": {
|
|
"0": 5432.37,
|
|
"1": 4861.66,
|
|
"2": 4890.23,
|
|
"3": 4919.9,
|
|
"4": 4929.49,
|
|
"5": 4927.35,
|
|
"6": 4927.6,
|
|
"7": 4932.02,
|
|
"8": 4897.83,
|
|
"9": 4897.16,
|
|
"10": 4899.59,
|
|
"11": 4901.51,
|
|
"12": 4896.38,
|
|
"13": 4914.3,
|
|
"14": 4901.91,
|
|
"15": 4898.58,
|
|
"16": 4903.18,
|
|
"17": 4899.54,
|
|
"18": 4910.5,
|
|
"19": 4898.18,
|
|
"20": 4901.56,
|
|
"21": 4903.12,
|
|
"22": 4896.82,
|
|
"23": 4901.01,
|
|
"24": 4897.42,
|
|
"25": 4911.94,
|
|
"26": 4901.29,
|
|
"27": 4894.79,
|
|
"28": 4899.57,
|
|
"29": 4900.63,
|
|
"30": 4904.78,
|
|
"31": 4899.79,
|
|
"32": 4898.01,
|
|
"33": 4910.81,
|
|
"34": 4901.16,
|
|
"35": 4898.48,
|
|
"36": 4894.97,
|
|
"37": 4905.9,
|
|
"38": 4901.52,
|
|
"39": 4898.56,
|
|
"40": 4900.19,
|
|
"41": 4896.5,
|
|
"42": 4899.89,
|
|
"43": 4898.45,
|
|
"44": 4897.27,
|
|
"45": 4895.47,
|
|
"46": 4889.26,
|
|
"47": 4887.44,
|
|
"48": 4894.07,
|
|
"49": 4888.59
|
|
},
|
|
"expected_e2e_ms": 254552.63,
|
|
"expected_avg_denoise_ms": 4912.17,
|
|
"expected_median_denoise_ms": 4899.69
|
|
},
|
|
"wan2_2_i2v_14b_720P": {
|
|
"stages_ms": {
|
|
"InputValidationStage": 47.04,
|
|
"TextEncodingStage": 2321.27,
|
|
"ImageEncodingStage": 3244.34,
|
|
"ConditioningStage": 0.0234,
|
|
"TimestepPreparationStage": 2.88,
|
|
"LatentPreparationStage": 5.24,
|
|
"ImageVAEEncodingStage": 1887.64,
|
|
"DenoisingStage": 245826.78,
|
|
"DecodingStage": 2882.45,
|
|
"per_frame_generation": null
|
|
},
|
|
"denoise_step_ms": {
|
|
"0": 5429.38,
|
|
"10": 4901.39,
|
|
"20": 4912.89,
|
|
"29": 4900.75,
|
|
"39": 4906.23,
|
|
"49": 4892.55
|
|
},
|
|
"expected_e2e_ms": 254850.94,
|
|
"expected_avg_denoise_ms": 4914.73,
|
|
"expected_median_denoise_ms": 4903.27
|
|
},
|
|
"wan2_2_ti2v_5b": {
|
|
"notes": "Text-and-Image-to-Video generation baseline for Wan2.2-TI2V-5B",
|
|
"expected_e2e_ms": 178300.0,
|
|
"expected_avg_denoise_ms": 3250.0,
|
|
"expected_median_denoise_ms": 3260.0,
|
|
"stages_ms": {
|
|
"InputValidationStage": 150.0,
|
|
"TextEncodingStage": 3000.0,
|
|
"ConditioningStage": 1.0,
|
|
"TimestepPreparationStage": 6.0,
|
|
"LatentPreparationStage": 30.0,
|
|
"DenoisingStage": 162900.0,
|
|
"DecodingStage": 14767.0
|
|
},
|
|
"denoise_step_ms": {
|
|
"0": 3700.0,
|
|
"10": 3300.0,
|
|
"20": 3300.0,
|
|
"29": 3300.0,
|
|
"39": 3300.0,
|
|
"49": 3300.0
|
|
},
|
|
"frames_per_second": null,
|
|
"total_frames": null,
|
|
"avg_frame_time_ms": null
|
|
}
|
|
}
|
|
}
|