tp: merge Wattson interconnect power into cpu_2d curves#5818
Draft
runpinglai-creator wants to merge 4 commits into
Draft
tp: merge Wattson interconnect power into cpu_2d curves#5818runpinglai-creator wants to merge 4 commits into
runpinglai-creator wants to merge 4 commits into
Conversation
310afb2 to
2627dc8
Compare
172cd45 to
0b4cf88
Compare
0b4cf88 to
a642bf5
Compare
wusamuel6
reviewed
May 14, 2026
wusamuel6
reviewed
May 14, 2026
| + iif(suspend.suspended = 1, 0, coalesce(base.interconnect_mw, 0)) AS dsu_scu_mw | ||
| FROM _w_independent_cpus_calc AS slices | ||
| JOIN _unique_estimates_mw AS base USING (config_hash) | ||
| LEFT JOIN _gapless_suspend_slices AS suspend |
Contributor
There was a problem hiding this comment.
This is weird logic, why does it need to check for time window overlap?
And if we do need time window overlap, we should use interval_intersect!()
Contributor
Author
There was a problem hiding this comment.
This was because upstream tables didn't pass "suspend" state down here. So I had to check the suspend overlap here to determine whether it's suspended here. Now I've passed the "suspend" states from upstream tables. Is this the concern?
Integrate the interconnect baseline power directly into cpu_2d.csv under a new interconnect column. This power represents the active baseline of the interconnect/bus system and is present whenever the system is awake (!suspended). - Add interconnect column to cpu_2d schema in gen_wattson_curves.py. - Register interconnect column in C++ plugin __intrinsic_wattson_curves_cpu_2d. - Migrate Tensor G4 & G5 deep idle power data into cpu_2d.csv under the interconnect column. Other devices are updated with 0. Bug: 338654268 Signed-off-by: Runping Lai <runpinglai@google.com>
d732cdb to
94df7e0
Compare
Extract interconnect curves from cpu_2d_raw and unconditionally look up interconnect_mw for Policy 0. Join CPU estimates with suspend slices for CPU 0, and add the interconnect baseline power to dsu_scu_mw whenever the system is awake (!suspended). Bug: 338654268 Test: UI test and TP test Signed-off-by: Runping Lai <runpinglai@google.com>
As DSU estimation adds "interconnect" power into it, dsu power will behigher Test: tools/diff_test_trace_processor.py Bug: 338654268 Signed-off-by: Runping Lai <runpinglai@google.com>
94df7e0 to
f7af353
Compare
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
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.
Background: In wattson, suspend is always mixed up with deep idle, both modeled as using 0 power. A lot of wattson regressions are due to suspend, either suspending longer or shorter than expected. Either way, wattson should be able to differentiate between suspend vs deep idle.
This change introduces active baseline power modeling for the interconnect and system bus in Wattson. This power is modeled under a new
interconnectcolumn incpu_2d.csvand integrated directly into the "DSU/SCU estimate"track.
The interconnect baseline represents the public power overhead of the SOC whenever the system is awake (!suspended), keeping the power domain model unified and physically accurate.