Public/automatic
1
0
Fork 0
mirror of https://github.com/vladmandic/automatic synced 2026-09-20 01:31:13 +02:00
Code Issues Packages Projects Releases Wiki Activity

Fix PI_ERROR_DEVICE_NOT_FOUND with UniPC and IPEX

Browse Source
This commit is contained in:
Disty0
2023-06-07 19:17:55 +03:00
parent 3bef3e3eee
commit 8e3a8fb474
1 changed files with 13 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files
modules/models/diffusion/uni_pc/uni_pc.py
+13 -2
View File
@@ -3,6 +3,7 @@ import torch.nn.functional as F
import math
import time
from rich.progress import Progress, TextColumn, BarColumn, TaskProgressColumn, TimeRemainingColumn, TimeElapsedColumn
from modules import shared
class NoiseScheduleVP:
@@ -684,7 +685,12 @@ class UniPC:
if order == 2:
rhos_p = torch.tensor([0.5], device=b.device)
else:
rhos_p = torch.linalg.solve(R[:-1, :-1], b[:-1])
if shared.cmd_opts.use_ipex:
#Running torch.linalg.solve on XPU crashes the GPU.
rhos_p = torch.linalg.solve(R[:-1, :-1].to("cpu"), b[:-1].to("cpu"))
rhos_p = rhos_p.to(b.device)
else:
rhos_p = torch.linalg.solve(R[:-1, :-1], b[:-1])
else:
D1s = None
@@ -694,7 +700,12 @@ class UniPC:
if order == 1:
rhos_c = torch.tensor([0.5], device=b.device)
else:
rhos_c = torch.linalg.solve(R, b)
if shared.cmd_opts.use_ipex:
#Running torch.linalg.solve on XPU crashes the GPU.
rhos_c = torch.linalg.solve(R.to("cpu"), b.to("cpu"))
rhos_c = rhos_c.to(b.device)
else:
rhos_c = torch.linalg.solve(R, b)
model_t = None
if self.predict_x0: