need a histogram report filtered on gfx(today)

RESOLVED WONTFIX

Status

Mozilla Metrics
Data/Backend Reports
RESOLVED WONTFIX
6 years ago
6 years ago

People

(Reporter: (dormant account), Unassigned)

Tracking

unspecified
Unreviewed
x86_64
Windows 7

Details

(Reporter)

Description

6 years ago
HTTP_PAGE_COMPLETE_LOAD_DISK where info.adapterDescription='AMD Radeon HD 6290 Graphics' or 'AMD Radeon HD 6250 Graphics'
(Reporter)

Comment 1

6 years ago
objective:
We have a significant performance issue.
we need to confirm that it's critical to prioritize against other work.
(Reporter)

Comment 2

6 years ago
for the technical issue behind this see bug 816642.

Comment 3

6 years ago
room.
[11:43am] taras: joy: can you or one of the analysts get me a histogram
[11:43am] taras: a filtered one
[11:43am] taras: within the next few hours?
[11:43am] joy: meaning?
[11:43am] taras: i need HTTP_PAGE_COMPLETE_LOAD_NET filtered on cpu/gfx
[11:43am] taras: err
[11:43am] taras: HTTP_PAGE_COMPLETE_LOAD_DISK
[11:45am] bcolloran left the chat room. (Quit: Leaving.)
[11:45am] bcolloran joined the chat room.
[11:45am] joy: taras: cannot see cpu or gfx in json$info ...
[11:45am] taras: joy: hold on a sec
[11:45am] taras: gonna give you exact fields and values
[11:48am] taras: joy: info.adapterDescription='AMD Radeon HD 6290 Graphics'
[11:48am] taras: should be enough
[11:48am] taras: let me just see if i can include multiple gfx types for this
[11:50am] taras: ok
[11:50am] taras: so either 6290 or 6250
[11:50am] taras: joy: so HTTP_PAGE_COMPLETE_LOAD_DISK where info.adapterDescription='AMD Radeon HD 6290 Graphics' or 'AMD Radeon HD 6250 Graphics'
[11:52am] joy: objective? dates?
[11:52am] taras: joy: https://bugzilla.mozilla.org/show_bug.cgi?id=816651
[11:52am] taras: joy: last 6months ok?
[11:53am] joy: not gonna happen today then and HTTP_PAGE_COMPLETE_LOAD_DISK  is not even present in 18? so why bother?
[11:53am] taras: it's not?
[11:53am] taras: that's weird
[11:53am] taras: i thought it was ancient
[11:54am] joy: HTTP_PAGE_COMPLETE_LOAD
[11:54am] joy:  is present
[11:55am] taras: hrm
[11:55am] taras: yeah that wont do
[11:55am] taras: let me see if i can find another proxy for this
[11:55am] joy: its not there in http://mxr.mozilla.org/mozilla-central/source/toolkit/components/telemetry/Histograms.json
[11:56am] joy: also, why do we want this?
[11:56am] joy: what si objective?
[11:56am] joy: not in the bug
[11:56am] taras: joy: cos we have significant performance issue
[11:56am] taras: we need to confirm that it's critical
[11:56am] taras: to prioritize over other work
[11:59am] joy: yes, but what is the performance issue? is there  abug?
[11:59am] joy: is 6290 worse than 6250
[11:59am] joy: or other way around?
[11:59am] joy: did it happen recently?
[12:00pm] christina joined the chat room.
[12:00pm] taras: joy: is HTTP_PAGE_CACHE_READ_TIME
[12:00pm] taras: available in older data?
[12:00pm] taras: joy: nah they should be similar
[12:01pm] taras: i'm basically using this to identify netbook-class hw
[12:01pm] joy: do elaborate the concern, issue, why this is an issue in the bug
[12:01pm] taras: where is suspect disk cache of being slower than no disk cache
[12:01pm] taras: ie superslow pageload times
[12:05pm] joy: so HTTP_PAGE_CACHE_READ_TIME is comparable for both 6290 an 6250
[12:05pm] joy: what does that tell you?
[12:05pm] taras: can you tell me what it is?
[12:06pm] taras: joy: i'm not comparing those two..i'm comparing those two against the world
[12:06pm] joy: what is the world?
[12:06pm] joy: i dont have any numbers nw
[12:06pm] taras: i wanna see if HTTP_PAGE_CACHE_READ_TIME on those machines
[12:06pm] joy: your bug is emptuy
[12:06pm] joy: fill int he bug
[12:07pm] taras: compares poorly to https://metrics.mozilla.com/data/content/pentaho-cdf-dd/Render?solution=metrics2&path=%2Ftelemetry&file=telemetryHistogram.wcdf&bookmarkState={%22impl%22%3A%22client%22%2C%22params%22%3A{%22startDate%22%3A%222012-10-30%22%2C%22endDate%22%3A%222012-11-28%22%2C%22appVersion%22%3A%22%22%2C%22appName%22%3A%22Firefox%22%2C%22arch%22%3A%22%22%2C%22OS%22%3A%22Darwin%22%2C%22version%22%3A%22%22%2C%22channel%22%3A%22nightly%22%2C%2
[12:07pm] taras: err
[12:07pm] taras: to the NET equiv
[12:07pm] taras: joy: i did fill in the bug
[12:07pm] joy: okay looking at linked bug
[12:08pm] taras: also HTTP_PAGE_CACHE_READ_TIME is not good
[12:08pm] taras: well
[12:08pm] taras: actually  it's good
[12:09pm] joy: so you want to compare  HTTP_PAGE_CACHE_READ_TIME for 6250/6290 (which is proxy for netbooks) vs. non 6250/6290
[12:09pm] joy: are you sure 6250/6290 is a proxy for netbook?
[12:09pm] taras: yes
[12:09pm] taras: so ideally i want to compare HTTP_PAGE_COMPLETE_LOAD_DISK vs HTTP_PAGE_COMPLETE_LOAD_NET
[12:10pm] taras: so HTTP_PAGE_COMPLETE_LOAD_DISK or HTTP_PAGE_CACHE_READ_TIME on netbooks
[12:10pm] taras: vs HTTP_PAGE_COMPLETE_LOAD_NET in general population(this part i dont need you to do)
[12:11pm] taras: i just need to see the distribution, our dashboard doesn't let us filter by gfx
[12:11pm] taras: dont need any fancy analysis, just a simple data pull
[12:12pm] joy: taras: are you sure HTTP_PAGE_COMPLETE_LOAD_DISK has same distribution as HTTP_PAGE_COMPLETE_LOAD_NET on gen pop?
[12:12pm] joy: if not, why compare two different measures?
[12:12pm] joy: i saw the bug
[12:12pm] taras: because that's better than nothing
[12:12pm] joy: its' wrong approach
[12:12pm] taras: it's not
[12:12pm] joy: you first want to confirm
[12:12pm] joy: HTTP_PAGE_COMPLETE_LOAD_DISK has same dist  as HTTP_PAGE_COMPLETE_LOAD_NET on genpop
[12:12pm] joy: if so, then compring those two for netbook vs world make sense
[12:13pm] joy: and you want to compare to non netbook
[12:13pm] joy: not world
[12:13pm] joy: because it might well be HTTP_PAGE_COMPLETE_LOAD_DISK has different dist than  HTTP_PAGE_COMPLETE_LOAD_NET  in non netbook world too
[12:13pm] taras: i dont care if the distributions match
[12:14pm] taras: i just need to see disk behavior
[12:14pm] joy: please explain how your analysis is valid?
[12:14pm] joy: you're comparing X for netbook with Y for world
[12:14pm] joy: and want to see if they are different
[12:14pm] joy: what happens if X for world and Y for world are also different
[12:14pm] joy: then what is your conclusion?
[12:15pm] joy: and you should compare netbook with non netbook, why world?
[12:16pm] hamilton left the chat room. (Quit: Leaving.)
[12:16pm] taras: um
[12:16pm] taras: i'm not sure where to begin
[12:16pm] aphadke1 joined the chat room.
[12:16pm] aphadke left the chat room. (Ping timeout)
[12:16pm] taras: why can't we do a data pull without arguing about details
[12:16pm] aphadke1 left the chat room. (Connection reset by peer)
[12:16pm] taras: i can explain while the job runs
[12:16pm] aphadke joined the chat room.
[12:17pm] taras: also i gotta run, got lunch then meetings
[12:17pm] taras: will be back at 2
(Reporter)

Updated

6 years ago
Status: NEW → RESOLVED
Last Resolved: 6 years ago
Resolution: --- → WONTFIX

Comment 4

6 years ago
Hey Tara's before you go can you give a more detailed description of what you need? I cut and paste the discussion, but am not sure there's enough info there still.

Comment 5

6 years ago
Ah never mind . You resolved it.
You need to log in before you can comment on or make changes to this bug.