Dell OptiPlex SX270 - thermal monitoring?
Torfinn Ingolfsen
my workstation (yes, my current workstation have lower specs than this
machine. Don't laugh!).
I have installed FreeBSD[2] 7.0-stable on it (i386). Normal[3] and
verbose[4] dmesgs are available. Everything seems to work, but there is
one thing tha doesn't: it doesn't report thermal parameters via acpi:
root@kg-work2# sysctl hw.acpi.thermal
sysctl: unknown oid 'hw.acpi.thermal'
In fact the whole acpi section from sysctl seems a bit incomplete:
root@kg-work2# sysctl hw.acpi
hw.acpi.supported_sleep_state: S1 S3 S4 S5
hw.acpi.power_button_state: S5
hw.acpi.sleep_button_state: S1
hw.acpi.lid_switch_state: NONE
hw.acpi.standby_state: S1
hw.acpi.suspend_state: S3
hw.acpi.sleep_delay: 1
hw.acpi.s4bios: 0
hw.acpi.verbose: 0
hw.acpi.disable_on_reboot: 0
hw.acpi.handle_reboot: 0
hw.acpi.reset_video: 0
hw.acpi.cpu.cx_lowest: C1
There is an acpidump[5] available too (from 'acpidump -dt').
The bios on this machine is from 2004, and that might be the cause. I
couldn't find a newer bios than version A06, release date 09/29/2004.
Does anybody know about a newer bios for thios machine anywhere?
Or does anybody know how to fix the DSDT in the bios to enable thernal
reporting?
Or perhaps there is a way to read the temperature off the cpu, like the
coretemp(4) driver?
For those who wonders: yes, I have tried mbmon, lmmon from ports, but
they didn't report anything useful as far as temperature goes.
root@kg-work2# mbmon -d -A
ioctl(smb0:open): No such file or directory
SMBus[Intel8XX(ICH/ICH2/ICH3/ICH4/ICH5/ICH6)] found, but No HWM
available on it!!
Summary of Detection:
* No monitors found.
InitMBInfo: Bad file descriptor
This program needs "setuid root"!!
Output from 'lmmon -si':
Motherboard Temp Voltages
255C / 491F / 528K Vcore1: +3.984V
Vcore2: +3.984V
Fan Speeds + 3.3V: +3.984V
+ 5.0V: +6.654V
1: 0 rpm +12.0V: +15.938V
2: 0 rpm -12.0V: -15.938V
3: 0 rpm - 5.0V: -6.654V
I think 255 degrees Celsius is a bit unrealistic, the machine can't fry
eggs yet. :-)
anyway, any good hints will be gratefully accepted.
References:
1) SX270 - http://tingox.googlepages.com/sx270
2) http://tingox.googlepages.com/sx270_freebsd
3) http://tingox.googlepages.com/w2-dmesg-7.0-stable-20080721.txt
4) http://tingox.googlepages.com/w2-dmesg-7.0-stable-20080721_verbose.txt
5) http://tingox.googlepages.com/sx270-acpidump-dt-20080721.txt.gz
--
Torfinn Ingolfsen,
Norway
Rob Warnock
+---------------
| I got this old machine, a SX270[1] that I thought I would put to use as
| my workstation (yes, my current workstation have lower specs than this
| machine. Don't laugh!).
| I have installed FreeBSD[2] 7.0-stable on it (i386). Normal[3] and
| verbose[4] dmesgs are available. Everything seems to work, but there is
| one thing tha doesn't: it doesn't report thermal parameters via acpi:
| root@kg-work2# sysctl hw.acpi.thermal
| sysctl: unknown oid 'hw.acpi.thermal'
|
+---------------
Well, one thing that's certain is that the support for thermal parameters
varies *wildly* across platforms/versions, e.g., three of my machines:
MB Manuf. FreeBSD Output of "sysctl hw.acpi.thermal.tz0.temperature"
========= ======= ==================================================
eMachines 4.6 sysctl: unknown oid 'hw.acpi.thermal.tz0.temperature'
sysctl: unknown oid 'hw.acpi.thermal'
H-P(laptop) 4.10 hw.acpi.thermal.tz0.temperature: 3552
Asus(64-bit) 6.2 hw.acpi.thermal.tz0.temperature: 40.0C
Only FreeBSD-4.10 on the H-P laptop is correct [reading in units of
0.1K, i.e., 3552 == 82.0 C], and varies as expected with room temp
and CPU load.[1] On the other hand, while FreeBSD-6.2 on the Asus
Athlon-64 *looks* reasonable, it's actually a constant that never varies!!
[But if one escapes to the BIOS on bootup, the *BIOS* can read the
CPU temp just fine.] And the 4.6 one, well, let's just let grampa
enjoy his dotage, shall we? ;-}
In short, IME getting reasonable temps from "sysctl" is a crapshoot.
Sometimes you're lucky, and sometimes not so much... :-{
-Rob
[1] Well, actually, the numbers seem to have a constant offset of
~20 C warmer than "reasonable". THat is, when I first turn the
machine on after an overnight at ~20-25 C, I've never seen an
initial reading cooler than 42-55 C. From working with other
Athlons/Opterons, I know that sometimes you need to calibrate
the CPU's temp sensor to get correct absolute temps, so I don't
worry about it, since I've *never* seen a reading higher than
90 C (which is ~70 C, really?).
-----
Rob Warnock <rp...@rpw3.org>
627 26th Avenue <URL:http://rpw3.org/>
San Mateo, CA 94403 (650)572-2607
Torfinn Ingolfsen
> MB Manuf. FreeBSD Output of "sysctl hw.acpi.thermal.tz0.temperature"
> ========= ======= ==================================================
> eMachines 4.6 sysctl: unknown oid 'hw.acpi.thermal.tz0.temperature'
> sysctl: unknown oid 'hw.acpi.thermal'
> H-P(laptop) 4.10 hw.acpi.thermal.tz0.temperature: 3552
> Asus(64-bit) 6.2 hw.acpi.thermal.tz0.temperature: 40.0C
>
> Only FreeBSD-4.10 on the H-P laptop is correct [reading in units of
> 0.1K, i.e., 3552 == 82.0 C], and varies as expected with room temp
> and CPU load.[1] On the other hand, while FreeBSD-6.2 on the Asus
> Athlon-64 *looks* reasonable, it's actually a constant that never varies!!
Ah, I know that one. Asus machine? AMD K8 cpu? Yes. Try to use k8temp[1]
instead of that sysctl.
I don't understand why Asus made that parameter a constant. Bad
programming maybe?
HTH
References:
1) k8temp - http://www.freshports.org/sysutils/k8temp/
--
Torfinn Ingolfsen,
Norway
Rob Warnock
| > MB Manuf. FreeBSD Output of "sysctl hw.acpi.thermal.tz0.temperature"
| > ========= ======= ==================================================
...
| > On the other hand, while FreeBSD-6.2 on the Asus Athlon-64
| > *looks* reasonable, it's actually a constant that never varies!!
|
| Ah, I know that one. Asus machine? AMD K8 cpu? Yes.
| Try to use k8temp[1] instead of that sysctl.
Thanks!!
Seems to sorta work, though a bit noisy [and suspiciously
close to the "constant" 40.0C that "sysctl" reports]:
# while sleep 1; do k8temp ; done
CPU 0 Core 1 Sensor 0: 43c
CPU 0 Core 1 Sensor 0: 41c
CPU 0 Core 1 Sensor 0: 42c
CPU 0 Core 1 Sensor 0: 42c
CPU 0 Core 1 Sensor 0: 41c
CPU 0 Core 1 Sensor 0: 41c
CPU 0 Core 1 Sensor 0: 40c
CPU 0 Core 1 Sensor 0: 41c
CPU 0 Core 1 Sensor 0: 41c
CPU 0 Core 1 Sensor 0: 42c
CPU 0 Core 1 Sensor 0: 41c
CPU 0 Core 1 Sensor 0: 41c
...
-Rob
Torfinn Ingolfsen
> Thanks!!
No worries. :-)
> Seems to sorta work, though a bit noisy [and suspiciously
> close to the "constant" 40.0C that "sysctl" reports]:
Yes. If only there was a way to get at the temperature that BIOS shows
from FreeBSD we couldcompare those and see if ther are differences.
--
Torfinn Ingolfsen,
Norway
Tore Lund
I wonder why you expect the BIOS to have any expertise in this matter.
k8temp reads from the chip according to instructions from AMD. Although
there is some dispute about the need to calibrate the temperatures, I
suppose k8temp still knows this better than ASUS.
To Rob,
You could try to run powerd. If that makes the cpu frequency change,
temperatures should change too. My Athlon 64 X2 reports 30 C/38 C right
now, with powerd running. Temperatures are a few degrees lower on
cooler days.
Strange that you only get one temperature, BTW. the k8temp port warns
that "Core numbers on single core systems may be incorrect".
--
Tore (home period no, user tl18)
jpd
On Sat, 26 Jul 2008 20:36:11 +0200, Tore Lund <ch...@my.sig.invalid> wrote:
>
> I wonder why you expect the BIOS to have any expertise in this matter.
Because it has to know intimate details about chipset and cpu and so
forth already, perhaps?
> k8temp reads from the chip according to instructions from AMD. Although
> there is some dispute about the need to calibrate the temperatures, I
> suppose k8temp still knows this better than ASUS.
ASUS probably didn't write the BIOS, as it probably sourced it (from
ami? award?), but I'd expect the BIOS writer to have full documentation
on the board from ASUS and on the CPU from AMD, which may or may not be
the case for the writers of k8temp. Apropos that, is this an on-chip
temperature sensor or one of those things close to it on the pcb?
--
j p d (at) d s b (dot) t u d e l f t (dot) n l .
This message was originally posted on Usenet in plain text.
Any other representation, additions, or changes do not have my
consent and may be a violation of international copyright law.
Tore Lund
> Begin <rqydnQuYMs0...@telenor.com>
> On Sat, 26 Jul 2008 20:36:11 +0200, Tore Lund <ch...@my.sig.invalid> wrote:
>> I wonder why you expect the BIOS to have any expertise in this matter.
>
> Because it has to know intimate details about chipset and cpu and so
> forth already, perhaps?
Sounds plausible, but sysctl does not report anything that resembles a
reasonable chip temperature.
>> k8temp reads from the chip according to instructions from AMD. Although
>> there is some dispute about the need to calibrate the temperatures, I
>> suppose k8temp still knows this better than ASUS.
>
> ASUS probably didn't write the BIOS, as it probably sourced it (from
> ami? award?), but I'd expect the BIOS writer to have full documentation
> on the board from ASUS and on the CPU from AMD, which may or may not be
> the case for the writers of k8temp. Apropos that, is this an on-chip
> temperature sensor or one of those things close to it on the pcb?
On-chip. Readings from k8temp look very convincing when cpu frequeny
and ambient temperature are taken into account.
Torfinn Ingolfsen
> I wonder why you expect the BIOS to have any expertise in this matter.
I don't expect the BIOS to have an 'expertise', I just would like
another (extra) data point.
Not specifically for the machines using a K( cpu, as I am quite happy
with k8temp already.
More to do with the original post in this thread - I am looking as hard
as I can to find a way toget a temperature reading from this old Dell
OptiPlex SX270 I have. So far with no success.
--
Torfinn Ingolfsen,
Norway
Rob Warnock
+---------------
| jpd wrote:
| >> I wonder why you expect the BIOS to have any expertise in this matter.
| >
| > Because it has to know intimate details about chipset and cpu and so
| > forth already, perhaps?
|
| Sounds plausible, but sysctl does not report anything that resembles a
| reasonable chip temperature.
Indeed, since "sysctl" reports [on my Asus Ath64] an absolutely
*constant* 40.0 C, regardless of CPU load.
+---------------
| >> k8temp reads from the chip according to instructions from AMD.
...
| > Apropos that, is this an on-chip
| > temperature sensor or one of those things close to it on the pcb?
|
| On-chip. Readings from k8temp look very convincing when cpu frequeny
| and ambient temperature are taken into account.
+---------------
Well, while "k8temp" readings *do* vary with CPU load, I wouldn't
call them "very convincing" at all. For example, when the system
is sitting essentially idle [modulo standard background daemons
waiting with nothing happening], the output from "k8temp" jitters
up and down between 40-42 C. True, when I impose a ~10 second CPU load
[by running a tight 4-instruction loop] "k8temp" does step up to
a jittery 47-49 C, so it's reading *something* from the CPU.
But it's a much smaller step than I recall from running the same
thing on my laptop [where "sysctl hw.acpi.thermal.tz0.temperature"
*does* work quite well (ignoring the apparent +20 C offset)], and
without all the jitter.