theluyuan/DiffSynth-Studio
1
0
Fork 0
mirror of https://github.com/modelscope/DiffSynth-Studio.git synced 2026-03-18 22:08:13 +00:00
Code Issues Packages Projects Releases Wiki Activity

[model][NPU]:Wan model rope use torch.complex64 in NPU

Browse Source
This commit is contained in:
feng0w0
2026-01-13 11:17:51 +08:00
parent 6be244233a
commit d16877e695
2 changed files with 3 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
diffsynth/models/wan_video_dit.py
View File

@@ -5,7 +5,6 @@ import math
from typing import Tuple, Optional
from einops import rearrange
from .wan_video_camera_controller import SimpleAdapter
from ..core.device.npu_compatible_device import IS_NPU_AVAILABLE
try:
import flash_attn_interface
@@ -94,7 +93,7 @@ def rope_apply(x, freqs, num_heads):
x = rearrange(x, "b s (n d) -> b s n d", n=num_heads)
x_out = torch.view_as_complex(x.to(torch.float64).reshape(
x.shape[0], x.shape[1], x.shape[2], -1, 2))
freqs = freqs.to(torch.complex64) if IS_NPU_AVAILABLE else freqs
freqs = freqs.to(torch.complex64) if freqs.device == "npu" else freqs
x_out = torch.view_as_real(x_out * freqs).flatten(2)
return x_out.to(x.dtype)

4
diffsynth/utils/xfuser/xdit_context_parallel.py
View File

@@ -5,7 +5,7 @@ from xfuser.core.distributed import (get_sequence_parallel_rank,
get_sequence_parallel_world_size,
get_sp_group)
from xfuser.core.long_ctx_attention import xFuserLongContextAttention
from ...core.device import parse_nccl_backend, parse_device_type, IS_NPU_AVAILABLE
from ...core.device import parse_nccl_backend, parse_device_type
def initialize_usp(device_type):
@@ -50,7 +50,7 @@ def rope_apply(x, freqs, num_heads):
sp_rank = get_sequence_parallel_rank()
freqs = pad_freqs(freqs, s_per_rank * sp_size)
freqs_rank = freqs[(sp_rank * s_per_rank):((sp_rank + 1) * s_per_rank), :, :]
freqs_rank = freqs_rank.to(torch.complex64) if IS_NPU_AVAILABLE else freqs_rank
freqs_rank = freqs_rank.to(torch.complex64) if freqs_rank.device == "npu" else freqs_rank
x_out = torch.view_as_real(x_out * freqs_rank).flatten(2)
return x_out.to(x.dtype)