Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Uniform behavior for measurements in FpsInfo #1495

Open
wants to merge 1 commit into
base: dev
Choose a base branch
from

Conversation

marius-pelegrin-arm
Copy link
Contributor

Having uniform/standardized log formats/behaviors allow for automation tools to be easier to implement and maintain.

As the measurements done by GFXReconstruct are performance free and were already done on the whole replay if no frame-range was specified, I propose to consider gfxrecon-replay as always acquiring measurements, either on the whole replay or on the specified frame-range.

Thus there is no need for special message/behavior depending if a frame-range has been manually specified. That's what this commit implements:

  • A measurement file is always created if the replay has been run successfully
  • The same log message at the end of the replay is displayed, no matter if a range has been specified or not and no matter how much time the (potentially non-existant) load has taken

In addition, as CLOCK_MONOTONIC has an arbitrary starting point, CLOCK_BOOTTIME start/end time have been added to the measurement file for time to be coherent between multiple runs of gfxrecon-replay

@ci-tester-lunarg
Copy link

CI gfxreconstruct build queued with queue ID 166370.

@ci-tester-lunarg
Copy link

CI gfxreconstruct build # 3999 running.

@marius-pelegrin-arm marius-pelegrin-arm force-pushed the uniform-measurements-behavior branch from 70bb92f to d081beb Compare April 12, 2024 09:41
@ci-tester-lunarg
Copy link

CI gfxreconstruct build queued with queue ID 166386.

@ci-tester-lunarg
Copy link

CI gfxreconstruct build # 4000 running.

@ci-tester-lunarg
Copy link

CI gfxreconstruct build # 4000 passed.

@marius-pelegrin-arm marius-pelegrin-arm marked this pull request as draft April 29, 2024 07:38
@marius-pelegrin-arm marius-pelegrin-arm force-pushed the uniform-measurements-behavior branch from d081beb to ca6b079 Compare April 29, 2024 08:58
@ci-tester-lunarg
Copy link

CI gfxreconstruct build queued with queue ID 175703.

@ci-tester-lunarg
Copy link

CI gfxreconstruct build # 4065 running.

@marius-pelegrin-arm marius-pelegrin-arm force-pushed the uniform-measurements-behavior branch from ca6b079 to 53b5445 Compare April 29, 2024 09:02
@ci-tester-lunarg
Copy link

CI gfxreconstruct build queued with queue ID 175715.

@marius-pelegrin-arm marius-pelegrin-arm marked this pull request as ready for review April 29, 2024 09:02
@ci-tester-lunarg
Copy link

CI gfxreconstruct build # 4066 running.

@ci-tester-lunarg
Copy link

CI gfxreconstruct build # 4066 passed.

@bradgrantham-lunarg bradgrantham-lunarg added P3 replay Issue with replay (capture was successful) labels Oct 10, 2024
@marius-pelegrin-arm marius-pelegrin-arm force-pushed the uniform-measurements-behavior branch from 53b5445 to ee92f48 Compare January 31, 2025 14:03
@ci-tester-lunarg
Copy link

CI gfxreconstruct build queued with queue ID 361709.

@ci-tester-lunarg
Copy link

CI gfxreconstruct build # 5999 running.

Having uniform/standardized log formats/behaviors allow for
automation tools to be easier to implement and maintain.

As the measurements done by GFXReconstruct are performance free
and were already done on the whole replay if no frame-range was
specified, I propose to consider gfxrecon-replay as always
acquiring measurements, either on the whole replay or on the
specified frame-range.

Thus there is no need for special message/behavior depending if a
frame-range has been manually specified. That's what this commit
implements:
- A measurement file is always created if the replay has been run
successfully
- The same log message at the end of the replay is displayed, no
matter if a range has been specified or not and no matter how
much time the (potentially inexistant) load has taken

In addition, as CLOCK_MONOTONIC has an arbitrary starting point,
CLOCK_BOOTTIME start/end time have been added to the measurement
file.
Also, for measurement of CPU load, it's better to have a clock
that measures process time instead of absolute time, so
CLOCK_PROCESS_CPUTIME_ID start/end time have been added to the
measurement file.

Change-Id: I79edf334cbb7382233be7c5a3d07251294d0565f
@marius-pelegrin-arm marius-pelegrin-arm force-pushed the uniform-measurements-behavior branch from ee92f48 to 8793414 Compare January 31, 2025 14:06
@ci-tester-lunarg
Copy link

CI gfxreconstruct build queued with queue ID 361724.

@ci-tester-lunarg
Copy link

CI gfxreconstruct build # 6000 running.

@ci-tester-lunarg
Copy link

CI gfxreconstruct build # 6000 passed.

@fabian-lunarg
Copy link
Contributor

fabian-lunarg commented Feb 5, 2025

A measurement file is always created if the replay has been run successfully

why is it necessary to set as default?
when measurements are not the scope for a particular use-case, this will still create measurement-files.
this might sprinkle unexpected files across the filesystem.
what when the filesystem is read-only etc.?
this feels definitely like an opt-in feature for me.

@fabian-lunarg fabian-lunarg self-assigned this Feb 5, 2025
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
P3 replay Issue with replay (capture was successful)
Projects
None yet
Development

Successfully merging this pull request may close these issues.

4 participants