[Hardware] Support platforms and plugin system (#774)

Signed-off-by: gcanlin <canlinguosdu@gmail.com>
This commit is contained in:
Canlin Guo authored and GitHub committed 2026-01-28 22:12:52 +08:00
1 parent b11d436e19
commit c9392dff01
103 files changed
+1770 -1218

No files matched your search

@@ -12,7 +12,7 @@ from vllm_omni.diffusion.data import DiffusionParallelConfig, logger
from vllm_omni.entrypoints.omni import Omni
from vllm_omni.inputs.data import OmniDiffusionSamplingParams
from vllm_omni.outputs import OmniRequestOutput
from vllm_omni.utils.platform_utils import detect_device_type, is_npu
from vllm_omni.platforms import current_omni_platform
def parse_args() -> argparse.Namespace:
@@ -113,17 +113,22 @@ def parse_args() -> argparse.Namespace:
default=1,
help="Number of GPUs used for tensor parallelism (TP) inside the DiT.",
)
parser.add_argument(
"--vae_use_slicing",
action="store_true",
help="Enable VAE slicing for memory optimization.",
)
parser.add_argument(
"--vae_use_tiling",
action="store_true",
help="Enable VAE tiling for memory optimization.",
)
return parser.parse_args()
def main():
args = parse_args()
device = detect_device_type()
generator = torch.Generator(device=device).manual_seed(args.seed)
# Enable VAE memory optimizations on NPU
vae_use_slicing = is_npu()
vae_use_tiling = is_npu()
generator = torch.Generator(device=current_omni_platform.device_type).manual_seed(args.seed)
# Configure cache based on backend type
cache_config = None
@@ -167,8 +172,8 @@ def main():
omni = Omni(
model=args.model,
vae_use_slicing=vae_use_slicing,
vae_use_tiling=vae_use_tiling,
vae_use_slicing=args.vae_use_slicing,
vae_use_tiling=args.vae_use_tiling,
cache_backend=args.cache_backend,
cache_config=cache_config,
enable_cache_dit_summary=args.enable_cache_dit_summary,