minke.yu/sglang
1
0
Fork 0
Code Issues Pull Requests Actions 12 Packages Projects Releases Wiki Activity

[codex] Add LTX-2.3 benchmark skill recipes (#22631)

Browse Source
This commit is contained in:
Xiaoyu Zhang
2026-04-13 12:23:32 +08:00
committed by GitHub
parent c775c31cb3
commit fae0a2fc3c
4 changed files with 130 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files
python/sglang/multimodal_gen/.claude/skills/sglang-diffusion-benchmark-profile/SKILL.md
+1 -1
View File
@@ -27,7 +27,7 @@ Before running any benchmark, profiler, or kernel-validation command:
## Main Reference
- [benchmark-and-profile.md](benchmark-and-profile.md) — canonical denoise benchmark, perf dump, and `torch.profiler` workflow; uses the checked-in nightly-aligned presets, including `LTX-2` two-stage
- [benchmark-and-profile.md](benchmark-and-profile.md) — canonical denoise benchmark, perf dump, and `torch.profiler` workflow; uses the checked-in nightly-aligned presets, plus `LTX-2`, `LTX-2.3` one-stage, and `LTX-2.3` two-stage benchmark recipes
- [existing-fast-paths.md](existing-fast-paths.md) — map bottlenecks to existing fused kernels, packed QKV paths, fused `QK norm + RoPE`, and distributed overlap patterns before proposing new code
- [scripts/diffusion_skill_env.py](scripts/diffusion_skill_env.py) — preflight helper: repo root discovery via `sglang.__file__`, write-access probe, benchmark/profile output directories, idle GPU selection
- [scripts/bench_diffusion_denoise.py](scripts/bench_diffusion_denoise.py) — end-to-end denoise benchmark preset runner via `sglang generate`; use `--list-models` to inspect preset order, then save perf dumps by label and compare them with `compare_perf.py`
python/sglang/multimodal_gen/.claude/skills/sglang-diffusion-benchmark-profile/benchmark-and-profile.md
+57
View File
@@ -100,6 +100,26 @@ PYTHONPATH=python python3 \
--output-dir "${BENCH_DIR}"
```
Run the `LTX-2.3` one-stage skill preset:
```bash
PYTHONPATH=python python3 \
python/sglang/multimodal_gen/.claude/skills/sglang-diffusion-benchmark-profile/scripts/bench_diffusion_denoise.py \
--model ltx23-one-stage \
--label baseline \
--output-dir "${BENCH_DIR}"
```
Run the `LTX-2.3` two-stage skill preset:
```bash
PYTHONPATH=python python3 \
python/sglang/multimodal_gen/.claude/skills/sglang-diffusion-benchmark-profile/scripts/bench_diffusion_denoise.py \
--model ltx23-two-stage \
--label baseline \
--output-dir "${BENCH_DIR}"
```
Run the full preset sweep:
```bash
@@ -123,6 +143,8 @@ Nightly-aligned presets come first; skill-only presets stay available after them
| `wan-ti2v` | `Wan-AI/Wan2.2-TI2V-5B-Diffusers` | Yes: `wan22_ti2v_5b_720p` | Uses the nightly cat image and motion prompt |
| `ltx2` | `Lightricks/LTX-2` | Yes: `ltx2_twostage_t2v` | Uses `LTX2TwoStagePipeline`; nightly-aligned prompt, negative prompt, 1536x1024, 121 frames, fps 24, seed 1234 |
| `wan-i2v` | `Wan-AI/Wan2.2-I2V-A14B-Diffusers` | Yes: `wan22_i2v_a14b_720p` | Aligned to nightly CFG-parallel 4-GPU launch |
| `ltx23-one-stage` | `Lightricks/LTX-2.3` | No | Skill-only extra preset for the native `LTX-2.3` one-stage baseline; 2 GPUs, 768x512, 121 frames, fps 24, 30 steps, guidance 3.0, seed 1234 |
| `ltx23-two-stage` | `Lightricks/LTX-2.3` | No | Skill-only extra preset for the native `LTX-2.3` two-stage path; uses `LTX2TwoStagePipeline`, 2 GPUs, 1536x1024, 121 frames, fps 24, 30 steps, guidance 3.0, seed 1234 |
| `hunyuanvideo` | `hunyuanvideo-community/HunyuanVideo` | No | Skill-only extra preset |
| `mova-720p` | `OpenMOSS-Team/MOVA-720p` | No | Skill-only extra preset |
| `helios` | `BestWishYsh/Helios-Base` | No | Skill-only extra preset |
@@ -151,6 +173,41 @@ After [PR #20707](https://github.com/sgl-project/sglang/pull/20707),
`LTX2TwoStagePipeline` is a native path. The spatial upsampler and distilled
LoRA are auto-resolved from the same model snapshot unless you override them.
### Manual command example: LTX-2.3 One-Stage
```bash
sglang generate \
--model-path=Lightricks/LTX-2.3 \
--prompt="A beautiful sunset over the ocean" \
--negative-prompt="shaky, glitchy, low quality, worst quality, deformed, distorted, disfigured, motion smear, motion artifacts, fused fingers, bad anatomy, weird hand, ugly, transition, static." \
--width=768 --height=512 \
--num-frames=121 --fps=24 \
--num-inference-steps=30 --guidance-scale=3.0 \
--seed=1234 --num-gpus=2 \
--save-output --enable-torch-compile --warmup
```
Use this when you want the native `LTX2Pipeline` baseline for `LTX-2.3` at the
validated one-stage resolution.
### Manual command example: LTX-2.3 Two-Stage
```bash
sglang generate \
--model-path=Lightricks/LTX-2.3 \
--pipeline-class-name=LTX2TwoStagePipeline \
--prompt="A beautiful sunset over the ocean" \
--negative-prompt="shaky, glitchy, low quality, worst quality, deformed, distorted, disfigured, motion smear, motion artifacts, fused fingers, bad anatomy, weird hand, ugly, transition, static." \
--width=1536 --height=1024 \
--num-frames=121 --fps=24 \
--num-inference-steps=30 --guidance-scale=3.0 \
--seed=1234 --num-gpus=2 \
--save-output --enable-torch-compile --warmup
```
This matches the new `ltx23-two-stage` skill preset and is a good benchmark target for
the recently merged `LTX-2.3` two-stage path.
### Manual command example: Wan2.2-I2V-A14B 720P
```bash
python/sglang/multimodal_gen/.claude/skills/sglang-diffusion-benchmark-profile/scripts/bench_diffusion_denoise.py
+41 -6
View File
@@ -12,7 +12,7 @@ Usage:
# Tag the run for later compare_perf.py usage
python3 python/sglang/multimodal_gen/.claude/skills/sglang-diffusion-benchmark-profile/scripts/bench_diffusion_denoise.py --model flux --label tuned
# All 12 preset models
# All 14 preset models
python3 python/sglang/multimodal_gen/.claude/skills/sglang-diffusion-benchmark-profile/scripts/bench_diffusion_denoise.py --all
# Show preset order, model path, and nightly mapping
@@ -193,6 +193,39 @@ MODELS = {
],
},
# 10. Skill-only extra preset
"ltx23-one-stage": {
"path": "Lightricks/LTX-2.3",
"prompt": "A beautiful sunset over the ocean",
"negative_prompt": "shaky, glitchy, low quality, worst quality, deformed, distorted, disfigured, motion smear, motion artifacts, fused fingers, bad anatomy, weird hand, ugly, transition, static.",
"seed": 1234,
"extra_args": [
"--width=768",
"--height=512",
"--num-frames=121",
"--fps=24",
"--num-inference-steps=30",
"--guidance-scale=3.0",
"--num-gpus=2",
],
},
# 11. Skill-only extra preset
"ltx23-two-stage": {
"path": "Lightricks/LTX-2.3",
"prompt": "A beautiful sunset over the ocean",
"negative_prompt": "shaky, glitchy, low quality, worst quality, deformed, distorted, disfigured, motion smear, motion artifacts, fused fingers, bad anatomy, weird hand, ugly, transition, static.",
"seed": 1234,
"extra_args": [
"--pipeline-class-name=LTX2TwoStagePipeline",
"--width=1536",
"--height=1024",
"--num-frames=121",
"--fps=24",
"--num-inference-steps=30",
"--guidance-scale=3.0",
"--num-gpus=2",
],
},
# 12. Skill-only extra preset
"hunyuanvideo": {
"path": "hunyuanvideo-community/HunyuanVideo",
"prompt": "A cat and a dog baking a cake together in a kitchen. The cat is carefully measuring flour, while the dog is stirring the batter with a wooden spoon. The kitchen is cozy, with sunlight streaming through the window.",
@@ -205,7 +238,7 @@ MODELS = {
"--num-inference-steps=30",
],
},
# 11. Skill-only extra preset
# 13. Skill-only extra preset
# Requires: <repo>/inputs/diffusion_benchmark/figs/mova_single_person.jpg
"mova-720p": {
"path": "OpenMOSS-Team/MOVA-720p",
@@ -221,7 +254,7 @@ MODELS = {
"--num-inference-steps=2",
],
},
# 12. Skill-only extra preset
# 14. Skill-only extra preset
"helios": {
"path": "BestWishYsh/Helios-Base",
"prompt": "A curious raccoon",
@@ -247,6 +280,8 @@ def required_gpus_for_model(model_key: str) -> int:
return 4
if model_key == "mova-720p":
return 4
if model_key in {"ltx23-one-stage", "ltx23-two-stage"}:
return 2
return 1
@@ -260,11 +295,11 @@ def print_model_catalog():
print("=" * 95)
print("MODEL PRESETS — Nightly-aligned first, skill-only extras after")
print("=" * 95)
print(f"{'Preset':<14} {'Nightly':<28} {'Model Path':<46} {'GPUs':>4}")
print(f"{'Preset':<17} {'Nightly':<28} {'Model Path':<46} {'GPUs':>4}")
print("-" * 95)
for model_key, cfg in MODELS.items():
print(
f"{model_key:<14} {model_nightly_case_id(model_key):<28} {cfg['path']:<46} {required_gpus_for_model(model_key):>4}"
f"{model_key:<17} {model_nightly_case_id(model_key):<28} {cfg['path']:<46} {required_gpus_for_model(model_key):>4}"
)
print("-" * 112)
print(
@@ -474,7 +509,7 @@ def main():
choices=list(MODELS.keys()),
help="Model to benchmark (default: flux)",
)
parser.add_argument("--all", action="store_true", help="Benchmark all 12 models")
parser.add_argument("--all", action="store_true", help="Benchmark all 14 models")
parser.add_argument(
"--list-models",
action="store_true",
python/sglang/multimodal_gen/.claude/skills/sglang-diffusion-performance/SKILL.md
+31
View File
@@ -88,6 +88,37 @@ sglang generate --model-path Lightricks/LTX-2 \
Note: this generate recipe is aligned with the nightly comparison case `ltx2_twostage_t2v`. After [PR #20707](https://github.com/sgl-project/sglang/pull/20707), `LTX2TwoStagePipeline` is a native path and auto-resolves the spatial upsampler plus distilled LoRA from the same model snapshot unless you override them.
### Native baseline, 2 GPUs: LTX-2.3 one-stage
```bash
sglang generate --model-path Lightricks/LTX-2.3 \
--prompt "A beautiful sunset over the ocean" \
--negative-prompt "shaky, glitchy, low quality, worst quality, deformed, distorted, disfigured, motion smear, motion artifacts, fused fingers, bad anatomy, weird hand, ugly, transition, static." \
--width 768 --height 512 \
--num-frames 121 --fps 24 \
--num-inference-steps 30 --guidance-scale 3.0 \
--seed 1234 --num-gpus 2 \
--enable-torch-compile --warmup --save-output
```
Note: use this as the native `LTX2Pipeline` baseline for `LTX-2.3`. It keeps the validated one-stage resolution and explicit `LTX-2.3` sampling defaults, and matches the `ltx23-one-stage` benchmark preset in `sglang-diffusion-benchmark-profile`.
### Benchmark target, 2 GPUs: LTX-2.3 two-stage
```bash
sglang generate --model-path Lightricks/LTX-2.3 \
--pipeline-class-name LTX2TwoStagePipeline \
--prompt "A beautiful sunset over the ocean" \
--negative-prompt "shaky, glitchy, low quality, worst quality, deformed, distorted, disfigured, motion smear, motion artifacts, fused fingers, bad anatomy, weird hand, ugly, transition, static." \
--width 1536 --height 1024 \
--num-frames 121 --fps 24 \
--num-inference-steps 30 --guidance-scale 3.0 \
--seed 1234 --num-gpus 2 \
--enable-torch-compile --warmup --save-output
```
Note: this is the recommended benchmark command for the new `LTX-2.3` two-stage path. It uses the native `LTX2TwoStagePipeline` and matches the `ltx23-two-stage` benchmark preset in `sglang-diffusion-benchmark-profile`.
### Maximum speed, image model, single GPU, lossless
```bash