Oops, the difference here is which percentile we're looking at.

There's an improvement in the 99th percentile, but a regression in the 50'th percentile.

Here's the split by device, looking at the median.

There is a slight regression on some devices:
https://uma.googleplex.com/p/chrome/timeline_v2?q=%7B%22day_count%22%3A%22All%22%2C%22end_date%22%3A%22latest%22%2C%22window_size%22%3A%221%22%2C%22filters%22%3A%5B%7B%22fieldId%22%3A%22channel%22%2C%22operator%22%3A%22EQ%22%2C%22study%22%3A%22%22%2C%22selected%22%3A%5B%223%22%5D%7D%2C%7B%22fieldId%22%3A%22platform%22%2C%22operator%22%3A%22EQ%22%2C%22study%22%3A%22%22%2C%22selected%22%3A%5B%22A%22%5D%7D%2C%7B%22fieldId%22%3A%22version_tags%22%2C%22operator%22%3A%22CONTAINS%22%2C%22study%22%3A%22%22%2C%22selected%22%3A%5B%22D%22%5D%7D%2C%7B%22fieldId%22%3A%22short_hw_class%22%2C%22operator%22%3A%22COMPARE%22%2C%22study%22%3A%22%22%2C%22selected%22%3A%5B%22Nexus%205%22%2C%22Nexus%206%22%2C%22Nexus%205X%22%2C%22Nexus%206P%22%2C%22SM-G900F%22%2C%22SM-G900H%22%2C%22SM-G920F%22%2C%22SM-G920V%22%5D%7D%5D%2C%22histograms%22%3A%5B%5B%22Event.Latency.TouchToFirstScrollUpdateSwapBegin%22%5D%5D%2C%22default_entry_values%22%3A%7B%22measureModel%22%3A%7B%22measure%22%3A%22%22%2C%22buckets%22%3A%5B%5D%2C%22percentiles%22%3A%5B%2250%22%5D%2C%22selectedFormulas%22%3A%5B%5D%2C%22allFormulas%22%3A%5B%5D%7D%2C%22zeroBased%22%3Atrue%2C%22logScale%22%3Afalse%2C%22showLowVolumeData%22%3Afalse%2C%22showVersionAnnotations%22%3Atrue%7D%2C%22entries%22%3A%5B%7B%22measureModel%22%3A%7B%22measure%22%3A%22percentile%22%7D%2C%22zeroBased%22%3Afalse%7D%5D%7D

Sami, any thoughts on what might have improved this in the 99'th, but regressed it in the 50'th?

Is there a revision range for this?

Based on the dev graph:
https://uma.googleplex.com/p/chrome/timeline_v2/?q=%7B%22day_count%22%3A%22All%22%2C%22end_date%22%3A%22latest%22%2C%22window_size%22%3A%221%22%2C%22filters%22%3A%5B%7B%22fieldId%22%3A%22channel%22%2C%22operator%22%3A%22EQ%22%2C%22study%22%3A%22%22%2C%22selected%22%3A%5B%222%22%5D%7D%2C%7B%22fieldId%22%3A%22platform%22%2C%22operator%22%3A%22EQ%22%2C%22study%22%3A%22%22%2C%22selected%22%3A%5B%22A%22%5D%7D%2C%7B%22fieldId%22%3A%22version_tags%22%2C%22operator%22%3A%22CONTAINS%22%2C%22study%22%3A%22%22%2C%22selected%22%3A%5B%22D%22%5D%7D%5D%2C%22histograms%22%3A%5B%5B%22Event.Latency.TouchToFirstScrollUpdateSwapBegin%22%5D%5D%2C%22default_entry_values%22%3A%7B%22measureModel%22%3A%7B%22measure%22%3A%22%22%2C%22buckets%22%3A%5B%5D%2C%22percentiles%22%3A%5B%2250%22%5D%2C%22selectedFormulas%22%3A%5B%5D%2C%22allFormulas%22%3A%5B%5D%7D%2C%22zeroBased%22%3Atrue%2C%22logScale%22%3Afalse%2C%22showLowVolumeData%22%3Afalse%2C%22showVersionAnnotations%22%3Atrue%7D%2C%22entries%22%3A%5B%7B%22measureModel%22%3A%7B%22measure%22%3A%22percentile%22%7D%2C%22zeroBased%22%3Afalse%7D%5D%7D

It looks a bit like there may be two independent regressions, and neither of the happened at the same time as the improvement in 99th.

It's annoying we don't have useable data from Windows at this time - 51 reporting was broken for these metrics.

The improvement in 99th happened right at the start of 51:
https://chromium.googlesource.com/chromium/src/+log/50.0.2661.0..51.0.2681.0?n=10000

Range 1:
https://chromium.googlesource.com/chromium/src/+log/51.0.2700.0..51.0.2704.0?n=10000
This is a big range... This might just look like a separate regression due to noise.

Range 2:
https://chromium.googlesource.com/chromium/src/+log/51.0.2704.10..51.0.2704.22?n=10000
A few painting changes that might be related.
Only scheduling change of note was:
crrev.com//17a6f8c87ac9c10273d75f5c716b92a59567d57a

For range 2, there's also a regression in Event.Latency.Browser.TouchAcked, which indicates we're just seeing the renderer respond more slowly.

Sami, do you see anything suspicious in this range?

https://chromium.googlesource.com/chromium/src/+/0dd7b3fcb5d194651877f466ecf4101ae91bb953 may have made a difference, although hard to say if it would have accounted for the entire jump.

Nothing obvious jumps out from either of the two ranges :\ It's pretty suspicious how well the two regressions match the improvement.

Sometimes I'd like to see behavior of the same Chrome version over time to get an idea of how much inherent noise there is here.

What is the status of this issue?

This has recovered.