So, I came back from vacation and updated my rawhide machine to the latest. Firefox 3 is all broke for me (haven't diagnosed that yet) and not having a Firefox 2 package handy that doesn't conflict with xulrunner, I decided to build one. Core temp quickly went to 125C°+ and the computer powered itself off to avoid starting any fires.
This doesn't seem to happen with kernel-2.6.24-0.83.rc5.fc9, which I had installed before the break. Oh, it gets hot, but more like 95C°. Processor is an AMD Athlon(tm) 64 Processor 3200+ but I'm running in 32-bit mode.
Anyone else seeing this?
Matthew Miller writes:
So, I came back from vacation and updated my rawhide machine to the latest. Firefox 3 is all broke for me (haven't diagnosed that yet) and not having a Firefox 2 package handy that doesn't conflict with xulrunner, I decided to build one. Core temp quickly went to 125C°+ and the computer powered itself off to avoid starting any fires.
This doesn't seem to happen with kernel-2.6.24-0.83.rc5.fc9, which I had installed before the break. Oh, it gets hot, but more like 95C°. Processor is an AMD Athlon(tm) 64 Processor 3200+ but I'm running in 32-bit mode.
Anyone else seeing this?
This is a hardware problem! Even running all cores 100%, you should have sufficient cooling to prevent this from happening. Don't blame the software.
Andrew.
On Sat, Jan 05, 2008 at 01:50:56PM +0000, Andrew Haley wrote:
This is a hardware problem! Even running all cores 100%, you should have sufficient cooling to prevent this from happening. Don't blame the software.
Oh, I'm sure it is. However, it doesn't get as hot with the kernel build from a month ago, which seemed worth mentioning.
On Sat, 2008-01-05 at 13:50 +0000, Andrew Haley wrote:
This is a hardware problem! Even running all cores 100%, you should have sufficient cooling to prevent this from happening. Don't blame the software.
I wouldn't always assume that hardware is always protected from damage due to faulty software. There have been a lot of cases in the past where programming inadvertently damaged hardware: the recent Load_Cycle_Count kerfuffle with harddisk drives, burning-out CRT monitors with faulty Modelines, overheating laptops with covers closed and monitor and CPU not going low, etc. The fact that there was a difference indicates, at the least, seemingly more power-consumption for the same amount of work with just a different kernel. Certainly worth investigating. --
Richi Plana
On Sat, Jan 05, 2008 at 08:13:12AM -0700, Richi Plana wrote:
Modelines, overheating laptops with covers closed and monitor and CPU not going low, etc. The fact that there was a difference indicates, at the least, seemingly more power-consumption for the same amount of work with just a different kernel. Certainly worth investigating.
PC systems today are usually designed with system managed fans and a hardware cut out at the point things go 'critical'. So it is in fact possible that a kernel change caused a problem which then shut the machine down as it hit the hardware thermal limit. Unlikely but possible.
Running old and new kernels will verify that after checking the fans etc are in fact working properly. If this is the case it would be very nice to know ASAP and also to let Len Brown lenb@kernel.org know that you have a case where a box overheats with one kernel and not another
On Saturday 05 January 2008 08:46:27 Matthew Miller wrote:
This doesn't seem to happen with kernel-2.6.24-0.83.rc5.fc9, which I had installed before the break. Oh, it gets hot, but more like 95C°. Processor is an AMD Athlon(tm) 64 Processor 3200+ but I'm running in 32-bit mode.
Anyone else seeing this?
I have a Phenom processor and it reads 40C no matter what. When I had a Brisbane processor in the same mb, it was correct. Rebooting and going into the BIOS shows the correct temperature. But
# cat /proc/acpi/thermal_zone/THRM/temperature
is dead wrong. Not quite the same problem, but if its not able to get the temperature right, it may not be doing other power/thermal policy things right.
-Steve
2008/1/5, Steve Grubb sgrubb@redhat.com:
On Saturday 05 January 2008 08:46:27 Matthew Miller wrote:
This doesn't seem to happen with kernel-2.6.24-0.83.rc5.fc9, which I had installed before the break. Oh, it gets hot, but more like 95C°. Processor is an AMD Athlon(tm) 64 Processor 3200+ but I'm running in 32-bit mode.
Anyone else seeing this?
I have a Phenom processor and it reads 40C no matter what. When I had a Brisbane processor in the same mb, it was correct. Rebooting and going into the BIOS shows the correct temperature. But
# cat /proc/acpi/thermal_zone/THRM/temperature
is dead wrong. Not quite the same problem, but if its not able to get the temperature right, it may not be doing other power/thermal policy things right.
I see the same thing on my computer (which has an AMD Athlon(tm) 64 X2 Dual Core Processor 3800+), but lm_sensors reports the correct temperature.
Le samedi 05 janvier 2008 à 08:46 -0500, Matthew Miller a écrit :
So, I came back from vacation and updated my rawhide machine to the latest.
Core temp quickly went to 125C°+ and the computer powered itself off to avoid starting any fires.
This doesn't seem to happen with kernel-2.6.24-0.83.rc5.fc9, which I had installed before the break. Oh, it gets hot, but more like 95C°.
Either way that's too high (my X2 rarely goes over 50°C). Check your heatsinks, vacations have high hardware death probabilities, since the hardware is not stressed the same as the rest of the year.
Matthew Miller wrote:
So, I came back from vacation and updated my rawhide machine to the latest. Firefox 3 is all broke for me (haven't diagnosed that yet) and not having a Firefox 2 package handy that doesn't conflict with xulrunner, I decided to build one. Core temp quickly went to 125C°+ and the computer powered itself off to avoid starting any fires.
This doesn't seem to happen with kernel-2.6.24-0.83.rc5.fc9, which I had installed before the break. Oh, it gets hot, but more like 95C°. Processor is an AMD Athlon(tm) 64 Processor 3200+ but I'm running in 32-bit mode.
Anyone else seeing this?
Do you have lm_sensors configured? Perhaps a driver bug is causing your CPU fan to be stopped / to slow. Try configuring lm_sensors and take a look at your CPU fan speed with both the cool and the hot kernel.
Regards,
Hans
Matthew Miller escribió:
So, I came back from vacation and updated my rawhide machine to the latest. Firefox 3 is all broke for me (haven't diagnosed that yet) and not having a Firefox 2 package handy that doesn't conflict with xulrunner, I decided to build one. Core temp quickly went to 125C°+ and the computer powered itself off to avoid starting any fires.
This doesn't seem to happen with kernel-2.6.24-0.83.rc5.fc9, which I had installed before the break. Oh, it gets hot, but more like 95C°. Processor is an AMD Athlon(tm) 64 Processor 3200+ but I'm running in 32-bit mode.
Saame kernel, but with a turion 64 chip, running the x86_64 kernel and I get 50°C or 53°C.
95°C is the max temperture of this core.