[WIP] Build C++ & Python tracing support for PyTorch/XLA ops #3348
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…intercept and generate traces
miladm
changed the title
|
Sample output of |
JackCaoG
reviewed
Feb 7, 2022
| @@ -0,0 +1,36 @@ | |||
| import torch | |||
| import torch_xla | |||
| #torch_xla._XLAC._ltc_init_ts_backend() #TOREMOVE | |||
| stack = traceback.format_list(stack) | ||
| if (len(stack) > 1): | ||
| second_frame = str(stack[-2]).replace( | ||
| "/usr/local/google/home/miladmo/anaconda3/envs/pytorch1/bin/ipython", |
There was a problem hiding this comment.
If you have to hardcode the path, maybe use something like /tmp instead of your home dir.
| @@ -128,6 +130,15 @@ int64_t XLATensorImpl::size(int64_t d) const { | |||
| } | |||
|
|
|||
| void XLATensorImpl::SetupSizeProperties() { | |||
| if (getPythonPrinter() && !disablePrinter()) { | |||
There was a problem hiding this comment.
so if I understand correctly, this will ensure that whenever tensor.size() is called, we log the python frame and C++ frame if printer is set. I think it is a useful tool but I am not sure if it is a good idea for us to merge the change in this file into the master. I think it make a lot of sense of add the binding and documents, but this part to me feel like a debugging feature for developer.
Unless we are at a stage where we expect user to dump this info and do debugging, I felt like it might be better to leave this in a separate branch?
| }; | ||
| getPythonPrinter() = printer; | ||
| }); | ||
| m.def("_print_text", |
There was a problem hiding this comment.
what's the use of this function?
miladm
changed the title
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This PR is a software tool to enable tracing PyTorch/XLA ops at Python and C++ levels combined.
As an immediate use case, the
sizeop is traced to showcase the approach. Understanding the utility ofsizeat a detailed level is an important aspect of supporting Dynamic Shape in PyTorch / XLA and in Lazy Tensor Core (LTC).