-
I keep getting errors:
Section 'autotune_tmc stepper_x' is not a valid config section
Here is my config section:
```
[autotune_tmc stepper_x]
motor: ldo-42sth48-2004mah
[autotune_tmc ste…
-
os:windows cpu:13900ks and gpu:4090 torch:2.5.0-cuda
first try pip install with windows whl
get
ERROR: deepspeed-0.15.0-cp311-cp311-win_amd64.whl is not a supported wheel on this platform.
then try i…
-
```
CUDA_LAUNCH_BLOCKING=1 TORCHINDUCTOR_PERSISTENT_REDUCTIONS=0 pytest test/inductor/test_cuda_repro.py -k test_selecsls42b_misaligned_address
```
Fails with:
```
RuntimeError: Triton Error [CUD…
-
# Summary
FlexAttention differs from other TritonTemplates currently supported in inductor (mm variants). Traditional templates are typically only enabled when compiling with max-autotune and sweep…
-
**Describe the bug**
Kruize remote monitoring functional tests are failing with performance_profile not found from createExp and listRecommendations testcases on openshift with latest kruize 0.2 imag…
-
Hey!
I played around with your lib (thank a lot for the great job btw!) and nearly fried my stepper motor. When reading again you mentioned at several point to reduce the max current, when necessary,…
-
Hi, when I use autotune to train a text classification model in python with argument `autotune-modelsize`, the trained model size is always less than 1M when `autotune-modelsize` is greater than 35M(e…
-
Hi wanted to ask how tmc autotune works because it is greyed out
-
### 🚀 The feature, motivation and pitch
I'm trying to write an inductor lowering for `torch._cslt_sparse_mm`, an aten op that takes has two optional tensor kwargs, `bias` and `alpha`.
https://git…
jcaip updated
3 weeks ago
-
Hello Guys, change to version 3.0 and made a autotune. So they show me this numbers:
…