Performance regression with many poses

[teh-cmc]

Via @Danvil in #7222 (comment):

With branch master (as of 8/27, 3PM PDT) it feels a little bit faster but performance is still lower than rerun 0.16.

I tried some bisection of the rerun visualizations in my application and the bottleneck now seems to be this code:

rr.set_time_seconds("timestamp", my_timestamp);
rr.log("path", &rerun::Clear::recursive())?;
for (i, pose) in path.iter().enumerate() {
    rr.log(format!("{tag}/{i}"), &rerun::Transform3D::from_translation_mat3x3(pose.t3, pose.r3))?;
}

(there are 250 poses)

Without this visualization 0.18 also is much faster. So it seems like a large part of the slowdown from 0.16 to 0.18 is due to visualization of the poses.

[jleibs]

Any chance the recursive clear is impacting things now that we do query-time clears?

[teh-cmc]

Any chance the recursive clear is impacting things now that we do query-time clears?

The clear in that example snippet doesn't seem to apply to the transforms at all, but maybe that's just a typo? 🤔

In any case i'll try both.

[teh-cmc]

(This will be my ever-involving live-investigation comment)

---

import rerun as rr

rr.init("rerun_example_transform3d")
rr.save("/tmp/buncha_transforms_without_clear.rrd")

rr.set_time_seconds("timestamp", 42)
for i in range(0, 250):
    rr.log(f"transforms/{i}", rr.Transform3D(translation=[i, 0, 0]))

0.17:

main:

---

import rerun as rr

rr.init("rerun_example_transform3d")
rr.save("/tmp/buncha_transforms_with_clear.rrd")

rr.set_time_seconds("timestamp", 42)
rr.log("transforms", rr.Clear(recursive=True))
for i in range(0, 250):
    rr.log(f"transforms/{i}", rr.Transform3D(translation=[i, 0, 0]))

0.17:

main: