Big perf drop on Win 7 x64 between 60.0b9 and 60.0b10

I was running 59.0.2 x64 and did some speed/performance testing using 
speed-battle.com. On my old clunker (Latitude E6510), I was hitting 
around 1200-1300. I updated to 60.0b11 and score went down to 850-1000 
range. I tried 60.0b10, same score range. Tried 60.0b9 and score went 
back up to 1200-1300 range.

I was not successful in using Mozgression in trying to find the problem 
as it wouldn't let me choose to see what changed between 60.0b9 and 
60.0b10. I've used Mozgression successfully before but not this time.

If anyone could throw me a bone on finding the right config parameters 
based on the below info I would like to help find the regression range.

Last Good: 20180402175344
Built from: d2564a0c25e5084dbd064cf546d98a9f72eed1f3

First Bad: 20180404171943
Built from: 8bf7d4cef4c3fd2048a56aae828cad0fbafc6ff3
0
TCW
4/10/2018 4:23:02 PM
mozilla.support.firefox 24285 articles. 8 followers. Post Follow

6 Replies
38 Views

Similar Articles

[PageSpeed] 56

On 4/10/18 12:23 PM, TCW wrote:
> I was running 59.0.2 x64 and did some speed/performance testing using 
> speed-battle.com. On my old clunker (Latitude E6510), I was hitting 
> around 1200-1300. I updated to 60.0b11 and score went down to 850-1000 
> range. I tried 60.0b10, same score range. Tried 60.0b9 and score went 
> back up to 1200-1300 range.
>
> I was not successful in using Mozgression in trying to find the 
> problem as it wouldn't let me choose to see what changed between 
> 60.0b9 and 60.0b10. I've used Mozgression successfully before but not 
> this time.
>
> If anyone could throw me a bone on finding the right config parameters 
> based on the below info I would like to help find the regression range.
>
> Last Good: 20180402175344
> Built from: d2564a0c25e5084dbd064cf546d98a9f72eed1f3
>
> First Bad: 20180404171943
> Built from: 8bf7d4cef4c3fd2048a56aae828cad0fbafc6ff3


I'm not quite understanding this.

Firefox 60.0b11 performs better than Fx 60.0b9, and you want to find out 
why and fix 60.0b9? Just find out why so you understand the process?

I've never used https://mozilla.github.io/mozregression/ but should 
probably spend some time learning how to use it.

-- 
Best name ever! Cadet Bone Spurs
NRA = Newly Recruited Activists
National Popular Vote <http://www.nationalpopularvote.com/>
Ubuntu 16.04LTS - Unity Desktop

0
WaltS48
4/11/2018 2:22:44 PM
On 4/11/2018 9:22 AM, WaltS48 wrote:
> On 4/10/18 12:23 PM, TCW wrote:
>> I was running 59.0.2 x64 and did some speed/performance testing using 
>> speed-battle.com. On my old clunker (Latitude E6510), I was hitting 
>> around 1200-1300. I updated to 60.0b11 and score went down to 850-1000 
>> range. I tried 60.0b10, same score range. Tried 60.0b9 and score went 
>> back up to 1200-1300 range.
>>
>> I was not successful in using Mozgression in trying to find the 
>> problem as it wouldn't let me choose to see what changed between 
>> 60.0b9 and 60.0b10. I've used Mozgression successfully before but not 
>> this time.
>>
>> If anyone could throw me a bone on finding the right config parameters 
>> based on the below info I would like to help find the regression range.
>>
>> Last Good: 20180402175344
>> Built from: d2564a0c25e5084dbd064cf546d98a9f72eed1f3
>>
>> First Bad: 20180404171943
>> Built from: 8bf7d4cef4c3fd2048a56aae828cad0fbafc6ff3
> 
> 
> I'm not quite understanding this.
> 
> Firefox 60.0b11 performs better than Fx 60.0b9, and you want to find out 
> why and fix 60.0b9? Just find out why so you understand the process?
> 
> I've never used https://mozilla.github.io/mozregression/ but should 
> probably spend some time learning how to use it.
> 

No, you misunderstood. I updated to 60.0b11 from 59.0.2 and saw a perf 
drop. I then tried 60.0b10, still saw a perf drop. I then tried 60.0b9 
and perf went back to the same levels as 59.0.2. Something regressed 
between 60.0b9 and 60.0b10.
0
TCW
4/11/2018 2:44:38 PM
On 4/11/18 10:44 AM, TCW wrote:
> On 4/11/2018 9:22 AM, WaltS48 wrote:
>> On 4/10/18 12:23 PM, TCW wrote:
>>> I was running 59.0.2 x64 and did some speed/performance testing using 
>>> speed-battle.com. On my old clunker (Latitude E6510), I was hitting 
>>> around 1200-1300. I updated to 60.0b11 and score went down to 
>>> 850-1000 range. I tried 60.0b10, same score range. Tried 60.0b9 and 
>>> score went back up to 1200-1300 range.
>>>
>>> I was not successful in using Mozgression in trying to find the 
>>> problem as it wouldn't let me choose to see what changed between 
>>> 60.0b9 and 60.0b10. I've used Mozgression successfully before but not 
>>> this time.
>>>
>>> If anyone could throw me a bone on finding the right config 
>>> parameters based on the below info I would like to help find the 
>>> regression range.
>>>
>>> Last Good: 20180402175344
>>> Built from: d2564a0c25e5084dbd064cf546d98a9f72eed1f3
>>>
>>> First Bad: 20180404171943
>>> Built from: 8bf7d4cef4c3fd2048a56aae828cad0fbafc6ff3
>>
>>
>> I'm not quite understanding this.
>>
>> Firefox 60.0b11 performs better than Fx 60.0b9, and you want to find 
>> out why and fix 60.0b9? Just find out why so you understand the process?
>>
>> I've never used https://mozilla.github.io/mozregression/ but should 
>> probably spend some time learning how to use it.
>>
> 
> No, you misunderstood. I updated to 60.0b11 from 59.0.2 and saw a perf 
> drop. I then tried 60.0b10, still saw a perf drop. I then tried 60.0b9 
> and perf went back to the same levels as 59.0.2. Something regressed 
> between 60.0b9 and 60.0b10.

Okay. So higher is faster and all it is measuring is the speed of 
JavaScript in the browser.

My Firefox Nightly tested at 710-729 range, but the browser is listed 
3rd on the ranking list for Linux with an overall score of 1095.44.

My Fx Beta 60.0b11 tested at 752-791 range. The browser is ranked 5th 
with an overall score of 1054.57.

My 59.0.2 is the best performer testing in a range of 884-910. Ranked 
2nd with an overall score of 1103.77.

Maybe I'll test on my Win10 laptop later, but there does appear to be a 
perf drop.

I'm not noticing anything with the naked eye.

Thanks for trying to track down the reason for the drop.

-- 
Best name ever! Cadet Bone Spurs
NRA = Newly Recruited Activists
National Popular Vote <http://www.nationalpopularvote.com/>
Ubuntu 16.04LTS - Unity Desktop
0
WaltS48
4/11/2018 3:41:19 PM
On 4/11/2018 10:41 AM, WaltS48 wrote:
> On 4/11/18 10:44 AM, TCW wrote:
>> On 4/11/2018 9:22 AM, WaltS48 wrote:
>>> On 4/10/18 12:23 PM, TCW wrote:
>>>> I was running 59.0.2 x64 and did some speed/performance testing 
>>>> using speed-battle.com. On my old clunker (Latitude E6510), I was 
>>>> hitting around 1200-1300. I updated to 60.0b11 and score went down 
>>>> to 850-1000 range. I tried 60.0b10, same score range. Tried 60.0b9 
>>>> and score went back up to 1200-1300 range.
>>>>
>>>> I was not successful in using Mozgression in trying to find the 
>>>> problem as it wouldn't let me choose to see what changed between 
>>>> 60.0b9 and 60.0b10. I've used Mozgression successfully before but 
>>>> not this time.
>>>>
>>>> If anyone could throw me a bone on finding the right config 
>>>> parameters based on the below info I would like to help find the 
>>>> regression range.
>>>>
>>>> Last Good: 20180402175344
>>>> Built from: d2564a0c25e5084dbd064cf546d98a9f72eed1f3
>>>>
>>>> First Bad: 20180404171943
>>>> Built from: 8bf7d4cef4c3fd2048a56aae828cad0fbafc6ff3
>>>
>>>
>>> I'm not quite understanding this.
>>>
>>> Firefox 60.0b11 performs better than Fx 60.0b9, and you want to find 
>>> out why and fix 60.0b9? Just find out why so you understand the process?
>>>
>>> I've never used https://mozilla.github.io/mozregression/ but should 
>>> probably spend some time learning how to use it.
>>>
>>
>> No, you misunderstood. I updated to 60.0b11 from 59.0.2 and saw a perf 
>> drop. I then tried 60.0b10, still saw a perf drop. I then tried 60.0b9 
>> and perf went back to the same levels as 59.0.2. Something regressed 
>> between 60.0b9 and 60.0b10.
> 
> Okay. So higher is faster and all it is measuring is the speed of 
> JavaScript in the browser.
> 
> My Firefox Nightly tested at 710-729 range, but the browser is listed 
> 3rd on the ranking list for Linux with an overall score of 1095.44.
> 
> My Fx Beta 60.0b11 tested at 752-791 range. The browser is ranked 5th 
> with an overall score of 1054.57.
> 
> My 59.0.2 is the best performer testing in a range of 884-910. Ranked 
> 2nd with an overall score of 1103.77.
> 
> Maybe I'll test on my Win10 laptop later, but there does appear to be a 
> perf drop.
> 
> I'm not noticing anything with the naked eye.
> 
> Thanks for trying to track down the reason for the drop.
> 

Thanks for verifying as well.
0
TCW
4/11/2018 3:52:35 PM
On 4/11/2018 11:52 AM, TCW wrote:
> On 4/11/2018 10:41 AM, WaltS48 wrote:
>> On 4/11/18 10:44 AM, TCW wrote:
>>> On 4/11/2018 9:22 AM, WaltS48 wrote:
>>>> On 4/10/18 12:23 PM, TCW wrote:
>>>>> I was running 59.0.2 x64 and did some speed/performance testing 
>>>>> using speed-battle.com. On my old clunker (Latitude E6510), I was 
>>>>> hitting around 1200-1300. I updated to 60.0b11 and score went down 
>>>>> to 850-1000 range. I tried 60.0b10, same score range. Tried 60.0b9 
>>>>> and score went back up to 1200-1300 range.
>>>>>
>>>>> I was not successful in using Mozgression in trying to find the 
>>>>> problem as it wouldn't let me choose to see what changed between 
>>>>> 60.0b9 and 60.0b10. I've used Mozgression successfully before but 
>>>>> not this time.
>>>>>
>>>>> If anyone could throw me a bone on finding the right config 
>>>>> parameters based on the below info I would like to help find the 
>>>>> regression range.
>>>>>
>>>>> Last Good: 20180402175344
>>>>> Built from: d2564a0c25e5084dbd064cf546d98a9f72eed1f3
>>>>>
>>>>> First Bad: 20180404171943
>>>>> Built from: 8bf7d4cef4c3fd2048a56aae828cad0fbafc6ff3
>>>>
>>>>
>>>> I'm not quite understanding this.
>>>>
>>>> Firefox 60.0b11 performs better than Fx 60.0b9, and you want to 
>>>> find out why and fix 60.0b9? Just find out why so you understand 
>>>> the process?
>>>>
>>>> I've never used https://mozilla.github.io/mozregression/ but should 
>>>> probably spend some time learning how to use it.
>>>>
>>>
>>> No, you misunderstood. I updated to 60.0b11 from 59.0.2 and saw a 
>>> perf drop. I then tried 60.0b10, still saw a perf drop. I then tried 
>>> 60.0b9 and perf went back to the same levels as 59.0.2. Something 
>>> regressed between 60.0b9 and 60.0b10.
>>
>> Okay. So higher is faster and all it is measuring is the speed of 
>> JavaScript in the browser.
>>
>> My Firefox Nightly tested at 710-729 range, but the browser is listed 
>> 3rd on the ranking list for Linux with an overall score of 1095.44.
>>
>> My Fx Beta 60.0b11 tested at 752-791 range. The browser is ranked 5th 
>> with an overall score of 1054.57.
>>
>> My 59.0.2 is the best performer testing in a range of 884-910. Ranked 
>> 2nd with an overall score of 1103.77.
>>
>> Maybe I'll test on my Win10 laptop later, but there does appear to be 
>> a perf drop.
>>
>> I'm not noticing anything with the naked eye.
>>
>> Thanks for trying to track down the reason for the drop.
>>
>
> Thanks for verifying as well.


My Win10 results are slower, probably because of a 1 Ghz difference in 
CPU speed and less RAM than my Linux box. It appears that Windows test 
rankings are lower overall.

You might want to try the tests at http://browserbench.org/

-- 
Mozilla/5.0 (Windows NT 10.0; WOW64; rv:60.0) Gecko/20100101 Thunderbird/60.0
AMD A8-7410 APU with AMD Radeon R5 Graphics 2.20 GHz 6.00 GB RAM
Windows 10 Home 1709

0
WaltS48
4/11/2018 4:02:34 PM
On 4/10/2018 11:23 AM, TCW wrote:
> I was running 59.0.2 x64 and did some speed/performance testing using 
> speed-battle.com. On my old clunker (Latitude E6510), I was hitting 
> around 1200-1300. I updated to 60.0b11 and score went down to 850-1000 
> range. I tried 60.0b10, same score range. Tried 60.0b9 and score went 
> back up to 1200-1300 range.
> 
> I was not successful in using Mozgression in trying to find the problem 
> as it wouldn't let me choose to see what changed between 60.0b9 and 
> 60.0b10. I've used Mozgression successfully before but not this time.
> 
> If anyone could throw me a bone on finding the right config parameters 
> based on the below info I would like to help find the regression range.
> 
> Last Good: 20180402175344
> Built from: d2564a0c25e5084dbd064cf546d98a9f72eed1f3
> 
> First Bad: 20180404171943
> Built from: 8bf7d4cef4c3fd2048a56aae828cad0fbafc6ff3

Filed bug: https://bugzilla.mozilla.org/show_bug.cgi?id=1454654
0
TCW
4/17/2018 2:10:17 PM
Reply: