|
About the mps category
|
|
3
|
3758
|
May 18, 2022
|
|
PyTorch support for Intel GPUs on Mac
|
|
11
|
14808
|
August 8, 2026
|
|
Mps seems much slower than cpu on an M1 Mac
|
|
2
|
92
|
August 4, 2026
|
|
Moving tensors to MPS device taking extremely long time
|
|
0
|
69
|
July 20, 2026
|
|
Float64 tensors on MPS
|
|
1
|
116
|
June 23, 2026
|
|
Faster Attenion with MPS backend
|
|
0
|
84
|
April 30, 2026
|
|
Request to join slack channel for better coordination on open PRs
|
|
0
|
54
|
March 13, 2026
|
|
Garbage collection for MPS
|
|
0
|
83
|
January 3, 2026
|
|
Macbook M2 Pro MPS issue
|
|
0
|
132
|
November 15, 2025
|
|
[MPS Crash] Assertion failed in _getLSTMGradKernelDAGObject using LSTM on macOS with MPS backend
|
|
8
|
411
|
August 26, 2025
|
|
Slower processing using Metal on PyTorch versus CPU for specific calculations
|
|
2
|
458
|
August 22, 2025
|
|
Loss.backward() assertion failures on mps
|
|
0
|
162
|
July 22, 2025
|
|
"no-op" (?!) set_default_device("mps") makes optimizer creation crash
|
|
0
|
134
|
May 11, 2025
|
|
MPS Deployment Target Mismatch
|
|
1
|
190
|
April 30, 2025
|
|
Questions about Pytorch 2.0 and mps
|
|
4
|
3181
|
April 30, 2025
|
|
MPS issue with Apple M4
|
|
0
|
714
|
March 5, 2025
|
|
Loss.backward() dies
|
|
0
|
179
|
February 28, 2025
|
|
How can I implement `aten::kthvalue.values`
|
|
0
|
108
|
January 16, 2025
|
|
Tanh issue: RuntimeError: view size is not compatible with input tensor's size and stride (at least one dimension spans across two contiguous subspaces). Use .reshape(...) instead
|
|
6
|
342
|
January 7, 2025
|
|
Mps large tensor handling bug
|
|
0
|
230
|
December 26, 2024
|
|
Can MPS profiles be inspected by non closed-source software? Where are they stored?
|
|
0
|
85
|
December 17, 2024
|
|
UniPC and `linalg_solve_ex` with AuraFlow and ComfyUI
|
|
0
|
261
|
December 7, 2024
|
|
Trying to move everything to MPS on M2 mac
|
|
1
|
796
|
November 27, 2024
|
|
Torch.fft.fftfreq returning wrong array
|
|
0
|
182
|
November 13, 2024
|
|
Torch.backends.mps.is_available = false in M1 sillicon
|
|
2
|
1308
|
October 31, 2024
|
|
Different results, and -sometimes- all nans
|
|
4
|
849
|
October 4, 2024
|
|
Training doesn't converge when running on M1 pro GPU (MPS device)
|
|
12
|
3165
|
August 19, 2024
|
|
Torch sparse tensor unsupported at mps
|
|
2
|
1828
|
August 19, 2024
|
|
MPS Back End Out of Memory on GitHub Action
|
|
6
|
2604
|
July 12, 2024
|
|
RuntimeError: required rank 4 tensor to use channels_last format
|
|
3
|
1891
|
July 8, 2024
|