Skip to content

Commit 9b2ee3c

Browse files
committed
Minor fixes
Summary: 1. Move FILE option to "Optional settings" section. 2. Add a link. 3. Clarify a sentence. Test Plan: Reviewers: Subscribers: Tasks: Tags:
1 parent cd7f684 commit 9b2ee3c

File tree

1 file changed

+4
-4
lines changed

1 file changed

+4
-4
lines changed

prototype_source/flight_recorder_tutorial.rst

Lines changed: 4 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -48,8 +48,6 @@ Enabling Flight Recorder
4848
------------------------
4949
There are two required environment variables to get the initial version of Flight Recorder working.
5050

51-
- ``TORCH_NCCL_DEBUG_INFO_TEMP_FILE``: Setting the path where the flight recorder will be dumped with file prefix. One file per
52-
rank. The default value is ``/tmp/nccl_trace_rank_``.
5351
- ``TORCH_NCCL_TRACE_BUFFER_SIZE = (0, N)``: Setting ``N`` to a positive number enables collection.
5452
``N`` represents the number of entries that will be kept internally in a circular buffer.
5553
We recommended to set this value at *2000*.
@@ -58,6 +56,8 @@ There are two required environment variables to get the initial version of Fligh
5856

5957
**Optional settings:**
6058

59+
- ``TORCH_NCCL_DEBUG_INFO_TEMP_FILE``: Setting the path where the flight recorder will be dumped with file prefix. One file per
60+
rank. The default value is ``/tmp/nccl_trace_rank_``.
6161
- ``TORCH_NCCL_TRACE_CPP_STACK = (true, false)``: Setting this to true enables C++ stack traces to be captured in Flight Recorder.
6262
C++ stack traces can be useful in providing the exact code path from a PyTorch Python call down to the primitive
6363
C++ implementation. Also see ``TORCH_SYMBOLIZE_MODE`` in additional settings.
@@ -74,7 +74,7 @@ Additional Settings
7474
``fast`` is a new experimental mode that is shown to be much faster than the traditional ``addr2line``.
7575
Use this setting in conjunction with ``TORCH_NCCL_TRACE_CPP_STACK`` to collect C++ traces in the Flight Recorder data.
7676
- If you prefer not to have the flight recorder data dumped into the local disk but rather onto your own storage, you can define your own writer class.
77-
This class should inherit from class ``::c10d::DebugInfoWriter`` and then register the new writer using ``::c10d::DebugInfoWriter::registerWriter``
77+
This class should inherit from class ``::c10d::DebugInfoWriter`` `(code) <https://github.com/pytorch/pytorch/blob/release/2.5/torch/csrc/distributed/c10d/NCCLUtils.hpp#L237>`__ and then register the new writer using ``::c10d::DebugInfoWriter::registerWriter``
7878
before we initiate PyTorch distributed.
7979

8080
Retrieving Flight Recorder Data via an API
@@ -189,7 +189,7 @@ command directly:
189189
Currently, we support two modes for the analyzer script. The first mode allows the script to apply some heuristics to the parsed flight
190190
recorder dumps to generate a report identifying potential culprits for the timeout. The second mode is simply outputs the raw dumps.
191191
By default, the script prints flight recoder dumps for all ranks and all ``ProcessGroups``(PGs). This can be narrowed down to certain
192-
ranks and PGs. An example command is:
192+
ranks and PGs using the *--selected-ranks* argument. An example command is:
193193
194194
Caveat: tabulate module is needed, so you might need pip install it first.
195195

0 commit comments

Comments
 (0)