-
To ensure the reliability of time-dependent tests and reduce their duration, I suggest mocking time in tests.
Currently, some tests implicitly rely on underlying platform performance by setting arbit…
-
Currently, the new profiler doesn't support HTML5 platform.
We should investigate the possibility to adding pthread support for the HTML5 build, and also
make the external profiling library build fo…
-
After perform the order of "cmake .. -DCUTLASS_NVCC_ARCHS=80" & "cmake .. -DCUTLASS_NVCC_ARCHS=80" , I got the following:
-- CMake Version: 3.20.5
-- CUTLASS 3.5.0
-- The CXX compiler identificatio…
-
### 🐛 Describe the bug
The following code:
```python
import torch
from torch.profiler import ProfilerActivity, profile, record_function, tensorboard_trace_handler
DEVICE = "cuda:1"
def…
-
Looks like it accepts an offset and passes it to GL as a pointer?
I think it needs to be 2 variations of this function: one where you pass a buffer with offset, and another where you pass a pointer. …
-
### Description
This is a meta issue to collect reported issues around Android Profiling and how we can improve profiling overall.
Refer to the individual issues to learn more about the status.
```…
-
### Software versions
The legate 24.06.00 conda package is missing legion_prof
### Jupyter notebook / Jupyter Lab version
_No response_
### Expected behavior
The conda packages should include leg…
-
Originally reported on Google Code with ID 6868
```
When we have blackboxing, internal functions of libraries (i.e., a function from a blackboxed
script called from a function of a blackboxed script)…
-
### 🐛 Describe the bug
Reproducing example:
```
>>> import torch
>>> from torch.fx.experimental.shape_inference.infer_shape import infer_shape
Traceback (most recent call last):
File "", lin…
-
I propose adding a DD_PROFILING_ENABLED environment variable to control profiling, similar to how DD_TRACE_ENABLED works for tracing.
Currently, there is no official environment variable to control…