Arm Backend: Expose PMU trace output from FVP run #14401
Conversation
Change-Id: I2c7d202e81ac81f98561904ad14bfcf2b0071bb6
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/14401
Note: Links to docs will display an error until the docs builds have been completed. ❌ 1 New FailureAs of commit cbb9cea with merge base d43cde5 ( NEW FAILURE - The following job has failed:
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
meta-cla
bot
added
the
CLA Signed
pytorch-bot
bot
added
the
release notes: none
|
@pytorchbot label "release notes: arm" |
|
@pytorchbot label ciflow/trunk |
|
To add these label(s) (ciflow/trunk) to the PR, please first approve the workflows that are awaiting approval (scroll to the bottom of this page). This helps ensure we don't trigger CI on this PR until it is actually authorized to do so. Please ping one of the reviewers if you do not have access to approve and run workflows. |
|
Didn't find following labels among repository labels: partner:,arm |
|
Didn't find following labels among repository labels: partner:arm |
zingo
added
ciflow/trunk
and removed
release notes: none
|
To add the ciflow label This helps ensure we don't trigger CI on this PR until it is actually authorized to do so. Please ping one of the reviewers if you do not have access to approve and run workflows. |
zingo
added
partner: arm
|
To add the ciflow label This helps ensure we don't trigger CI on this PR until it is actually authorized to do so. Please ping one of the reviewers if you do not have access to approve and run workflows. |
|
Hi @jmahbs thanks for the PR. Can you add more details (perhaps a snippet of output) on what's in this trace file and how this will be consumed downstream? Thanks. |
|
FYI https://docs.pytorch.org/executorch/main/etdump.html This is going to be fun! :) |
Exposes PMU trace output from an FVP. This lays part of the foundation
to enable us to use this output to as a data overlay in Model Explorer
visualisations.
The end goal here is to be able to visualise some profiling data in
Model Explorer using our Tosa Flatbuffer adapter. To enable this we need
to implement a few changes:
1. Expose PMU trace output from a FVP. This gives us performance data
from an FVP run. (This PR)
2. Expose Vela's debug database. This gives us generic information on
operators in a our model, and can be combined with the trace output to
provide more detailed profiling analysis
3. Write a script to combine the trace output and the debug database so
we can visualise it in Model Explorer in Executorch.
Here's a snippet of the PMU trace output:
```
{
"name": "axi_enabled_cycles",
"ph": "X",
"ts": "1029",
"pid": "DMA",
"tid": "axi_enabled_cycles",
"dur": "1014"
}
```
cc @digantdesai @freddan80 @per @zingo @oscarandersson8218
Follow on from #14401 Enables dumping of Vela's debug database to a specified directory . This gives us generic information on operators in our model, and can be combined with the trace output to provide more detailed profiling analysis. Co-authored-by: Zingo Andersen <zingo.andersen@arm.com>
Exposes PMU trace output from an FVP. This lays part of the foundation to enable us to use this output to as a data overlay in Model Explorer visualisations.
The end goal here is to be able to visualise some profiling data in Model Explorer using our Tosa Flatbuffer adapter. To enable this we need to implement a few changes:
Here's a snippet of the PMU trace output:
cc @digantdesai @freddan80 @per @zingo @oscarandersson8218