Re: Fwd: XPS 17 9730 runs significantly warmer on Kernel 6.5-rc3 compared to Kernel 6.4.3

From: Thorsten Leemhuis
Date: Fri Jul 28 2023 - 15:43:30 EST


On 28.07.23 20:13, Goran Brkuljan wrote:
> Hi Everyone,

Please don't sent HTML mail to Linux kernel lists, it often will be
rejected.

> Goran: On Tuesday, I asked you to perform bisection. Any update on it?
>
>  
> Sorry for the delayed response. I wasn't able to do git bisection as I
> was quite busy this week. 
>
> However, I am no longer convinced that there is a kernel related issue.
> I reinstalled both kernels and now I get consistent results between two,
> and they both return higher values.
> Now on both kernels, core20 spikes to +100.0°C and it stays there for
> the duration of the test. Other cores are going from 80-95°C.
>
> Note that I used the mainline kernel installer for handling installation. 

FWIW, I (and likely many others that read this) have no idea what
(distro specifc?) tool you are talking about. But I guess that doesn't
matter. :-D

> I would close the issue as tests and behaviour don't suggest significant
> differences between kernel versions. 

Okydo, thx for the update.

#regzbot resolve: notabug: seems everything is working fine

Ciao, Thorsten

> On Fri, Jul 28, 2023 at 3:42 PM Thorsten Leemhuis
> <regressions@xxxxxxxxxxxxx <mailto:regressions@xxxxxxxxxxxxx>> wrote:
>
> [CCing Thermal maintainer Rafael, maybe he can help guide us]
>
> Hi Bagas and everyone!
>
> On 27.07.23 16:24, Bagas Sanjaya wrote:
> >
> > I notice a regression report on Bugzilla [1].
>
> Bagas, it looks like you forwarded this to the maintainers of
> "THERMAL/CPU_COOLING"; I think "THERMAL" would have been the way better
> entry in MAINTAINERS, as many (all?) of the former are Linaro employees
> and thus primarily care about ARM based devices. That's why I CCed
> Rafael.
>
> > Quoting from it:
> >
> >> On my new XPS 17 9730 laptop, I recently installed the Kernel
> 6.5-rc3. I noticed, however, that during its operation, the laptop
> was getting significantly warmer than before. I did some analysis
> using the 'stress' tool to measure and compare the difference in CPU
> temperatures between Kernel 6.4.3 and Kernel 6.5-rc3. Here are my
> results:
>
> One thought: I wonder if there something wrong now or was there
> something wrong earlier that was fixed to unleash the full performance
> potential of the device. That 100°C there sounds like it's the latter
> case, as that's the TJUNCTION temp for that processor. But would be good
> to confirm from someone that knows better how hot such processors are
> allowed to get these days.
>
> Ciao, Thorsten
>
> >> OS: Ubuntu 23.10
> >> CPU: i7-13700H
> >>
> >> kernel 6.4.3:
> >>                                                                 
>                                                                    
>                                                    
> >> Initial CPU temperature:
> >> coretemp-isa-0000
> >> Core 0:        +47.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 4:        +45.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 8:        +46.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 12:       +46.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 16:       +45.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 20:       +44.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 24:       +44.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 25:       +44.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 26:       +44.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 27:       +44.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 28:       +45.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 29:       +45.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 30:       +45.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 31:       +45.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Stressing CPU for 30 seconds...
> >> stress: info: [4858] dispatching hogs: 14 cpu, 0 io, 0 vm, 0 hdd
> >> stress: info: [4858] successful run completed in 30s
> >> CPU temperature immediately after stress test:
> >> coretemp-isa-0000
> >> Core 0:        +67.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 4:        +76.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 8:        +69.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 12:       +77.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 16:       +69.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 20:       +76.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 24:       +68.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 25:       +68.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 26:       +68.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 27:       +68.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 28:       +71.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 29:       +71.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 30:       +71.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 31:       +71.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Test ended at: Tue Jul 25 01:10:16 PM CEST 2023
> >>
> >>
> >> kernel 6.5-rc-3:
> >>
> >> Initial CPU temperature:
> >> coretemp-isa-0000
> >> Core 0:        +46.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 4:        +46.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 8:        +46.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 12:       +46.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 16:       +44.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 20:       +44.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 24:       +45.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 25:       +45.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 26:       +45.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 27:       +45.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 28:       +45.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 29:       +45.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 30:       +45.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 31:       +45.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Stressing CPU for 30 seconds...
> >> stress: info: [11146] dispatching hogs: 14 cpu, 0 io, 0 vm, 0 hdd
> >> stress: info: [11146] successful run completed in 30s
> >> CPU temperature immediately after stress test:
> >> coretemp-isa-0000
> >> Core 0:        +84.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 4:        +97.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 8:        +87.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 12:      +100.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 16:       +86.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 20:       +99.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 24:       +83.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 25:       +83.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 26:       +83.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 27:       +83.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 28:       +89.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 29:       +89.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 30:       +89.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Core 31:       +89.0°C  (high = +100.0°C, crit = +100.0°C)
> >> Test ended at: Tue Jul 25 01:05:42 PM CEST 2023
> >>
> >> I tested few times with different setup and CPU temp on 6.5-rc3
> is always higher especially when stressing with fewer cores.
> >
> > See Bugzilla for the full thread.
> >
> > Goran: On Tuesday, I asked you to perform bisection. Any update on it?
> >
> > Anyway, I'm adding this regression to regzbot:
> >
> > #regzbot introduced: v6.4..v6.5-rc3
> https://bugzilla.kernel.org/show_bug.cgi?id=217703
> <https://bugzilla.kernel.org/show_bug.cgi?id=217703>
> > #regzbot title: significant temperature increase on XPS 17 9730
> >
> > Thanks.
> >
> > [1]: https://bugzilla.kernel.org/show_bug.cgi?id=217703
> <https://bugzilla.kernel.org/show_bug.cgi?id=217703>
> >
>