[Bug Report] Regression: Out-of-memory in ttnn.conv2d #17787
Labels
bug
Something isn't working
CNN_bug
CNNs
community
op_cat: conv2D
2D convolution for CNNs
P0
StableDiffusion3.5_medium
Describe the bug
Since commit 5444b3c we get an OOM error in
ttnn.conv2d
for an operation required in the transformer of Stable Diffusion 3.5 for an image size of 1024x1024.There is a similar issue (#17489) regarding a requirement to perform conv2d on even larger tensors. The failing tests in the similar issue however did never succeed, as far as I know. The one in the current issue worked before commit 5444b3c.
Test:
Error log:
Environment:
OS: Linux
Distro: Ubuntu 20.04.6 LTS
Kernel: 5.4.0-196-generic
Platform: x86_64
Python: 3.8.10
Device: n150
FW Bundle Version: 80.10.0.0
TT-Flash Version: 0.2.2.0
CM FW Version: 2.27.0.0
ETH FW Version: 6.9.0
BM BL Version: 129.2.0.0
BM App Version: 5.9.0.2
The text was updated successfully, but these errors were encountered: