I don't see this on Chrome OS. Sorry, I don't have a Windows machine to try this on. Nick, can you please take a look?

Here on Windows 10 x64 with Chrome 55 update, memory column is never populated (waited for 5 minutes).

afakhry: I can still repro on my samus on 56.0.2924.58. It's only a delay of 2-3 seconds, but it's very consistent.

This is believably a Windows-specific bug, and I believe we can probably understand what's going on by adding some TRACE_EVENTs.

There's no reason for us to not show memory asap. CPU usage may actually need need a baseline off of which to delta.

Re #2: That issue was fixed recently in  issue 673175 .

Re #3: I just tried on 57.0.2981.0 on a peach_pit and CPU and Memory usages take about 1 second to show up which is expected. I'll try to borrow a samus and see.

#5: CPU takes exactly 3 seconds. Until then everything is marked "0.0" for CPU usage, which is misleading. There is one update every second:

t+0: Task Manager loads:
     Memory and Network columns each display "-" for every row
     CPU usage column displays "0.0" for every row

t+1: Network column is populated

t+2: Memory column is populated

t+3: CPU column is corrected (some values greater than 0.0 are now shown)

If CPU is the sort column, tasks are sorted in no discernible order until the CPU column is correctly populated.

This is at low load.

২৩ মে, ২০১৭ ১১:২৫ pm এ, "michae… via monorail" <
monorail+v2.3419636982@chromium.org> লিখেছেন:

**UI Mass Triage**

This issue is still reproducible on windows 10 using latest canary #72.0.3618.0. Adding appropriate labels.

Is this a dupe of issue 657009?

No, this seems to be different. This seems to be some weirdness in the data collection model. I just invoked Chrome's task manager a few times and there was usually about a one second delay before the memory data populates (that seems too long) and sometimes the *Start Time* column took two seconds (two refresh passes) to populate.

Maybe it's as simple as the tasks to gather the data being tagged as best-effort or some-such, and therefore sometimes getting delayed for a while. I understand the reasoning behind having the data gathering tasks be tagged as best-effort (we don't want task manager slowing down Chrome) but having task manager lagging too much is arguably worse because then we can't reliably get timely data on what Chrome is doing.

TL;DR - my guess is that our data gathering tasks are running at too low a priority and my opinion is that this should probably be fixed. Adding davidbienvenu@ (who might be doing some task manager work), removing stanisc@.

it does look like the tasks are running at best effort:
https://cs.chromium.org/chromium/src/chrome/browser/task_manager/sampling/task_manager_impl.cc?rcl=b39cccfb917ee1fbab09d970bd41c4835e56c485&l=68

As an experiment, I increased the refresh interval in task_manager_table_model.cc to 20 seconds, which made it easier to see how the data collection works
1st paint - Pid and Priority, which is to be expected - the data collection first enumerates the PIDS and displays rows for each of them.
2nd paint - user handles, cpu of 0
3rd paint - mem footprint, actual cpu, network

Given that the same thing happens with a much larger interval, I doubt this is a task priority issue. In any case, I'll do a bit more debugging.