Re: lm_sensors trustworthiness
- From: floyd@xxxxxxxxxx (Floyd L. Davidson)
- Date: Wed, 28 Dec 2005 22:40:25 -0900
ebenONE@xxxxxxxxxxxxxxxxxxxxxxxxxxxx (Hactar) wrote:
>Hi, I recently compiled lm_sensors for my Asus A7V333 (2 GHz Athlon CPU).
>The output, once tweaked to remove nonexistent hardware, is this:
The configuration of lm_sensors allows you to adjust readings
in just about any way you want. Which also means that if whatever
is in the config file is not an exact match for the hardware,
the readings will not be "correct">
>asb100-i2c-0-2d
>Adapter: SMBus Via Pro adapter at e800
>VCore 1: +1.70 V (min = +1.57 V, max = +1.73 V)
>+3.3V: +3.10 V (min = +3.14 V, max = +3.46 V)
This one is almost certainly not right.
>+5V: +4.84 V (min = +4.74 V, max = +5.24 V)
>+12V: +12.35 V (min = +10.83 V, max = +13.19 V)
>-12V: -12.90 V (min = -0.00 V, max = -0.00 V)
>-5V: -5.41 V (min = -0.00 V, max = -0.00 V)
Those are all within range., though the min/mx for the two
negative voltages are clearly not set up correctly.
>CPU Fan: 5037 RPM (min = 1997 RPM, div = 4)
>Chassis Fan:
> 2678 RPM (min = 3994 RPM, div = 2)
It's hard to say with the fans, but those are reasonable numbers.
>M/B Temp: +30°C (high = +45°C, hyst = +40°C)
>Power Temp:
> +14°C (high = +45°C, hyst = +40°C)
>CPU Temp: +28°C (high = +60°C, hyst = +50°C)
None of these temperatures are likely to be right.
>vid: +1.650 V (VRM Version 9.0)
This suggests your VCore 1 voltage is not right.
>alarms:
>
>The thermostat says 74F (about 23C), although it's likely a bit warmer
>where the computer is. In particular, the "CPU Temp" seems a bit
>unbelievable, considering the dinky heat sink on the CPU. Should I believe
>the numbers shown here, and if not, how do I find the correct values? Thanks.
Motherboard models may not have the exact same hardware in all
versions, hence someone gets the specifications for one
particular version, sets up a configuration and distributes the
config file as something useful... and on a different
motherboard the results are odd.
And manufacturers don't necessarily provide (any/accurate) details
about the hardware.
Another problem is noise on the inputs to the system monitor
chip. Apparently it isn't common to waste money on bypass
capacitors; the result being a noise spike gets measured now and
then, and it trips the chip alarm circuit. That means chip
alarm indications cannot be trusted unless the measured voltage
indicates an out of bounds voltage too.
That's the bad news. The good news is that in fact you don't
necessarily need "accurate" measurements! What you really
really want... is an indication of a *change*. The trick is
figuring out how big a change should be indicated! That may not
be easy either.
You'll probably want to take the easy route, which can be to
just leave it as it is and remember what it reads so that you
can compare that to whatever it says tomorrow. (At my age,
remembering something tomorrow appears to be impossible, so that
doesn't work.) What I've done in the past was adjust the config
file for each voltage to make it read right on the target. That
way I can take one look, and I may not know what the voltage is,
but it sure gives an instant indication if it changes.
But the problem with that is trying to figure out how much it
changed, and knowing what is "normal". If you actually do have
the hardware specs it is possible to get fairly good voltage
measurements. But if they change the resistor values on the
motherboard, you might well have a reading of 5.00 volts on the
5 volt line when it is in fact 5 volts, but if it reads 5.25
volts that might actually be a voltage of 5.01 or 5.99! That
far off makes it useless...
What I did was download the "tellerstats" package, and then
vastly modify it. I run a cron job that generates a rolling
graph of voltages and temperatures. That provided a useful
baseline to see what kinds of data it produced over time. Things
like "normal" voltage variations.
http://www.apaflo.com/floyd_davidson/sensors/
--
Floyd L. Davidson <http://www.apaflo.com/floyd_davidson>
Ukpeagvik (Barrow, Alaska) floyd@xxxxxxxxxx
.
- Follow-Ups:
- Re: lm_sensors trustworthiness
- From: Hactar
- Re: lm_sensors trustworthiness
- References:
- lm_sensors trustworthiness
- From: Hactar
- lm_sensors trustworthiness
- Prev by Date: Re: lm_sensors trustworthiness
- Next by Date: Re: boot hangs in POST after connecting SATA HDD
- Previous by thread: Re: lm_sensors trustworthiness
- Next by thread: Re: lm_sensors trustworthiness
- Index(es):
Relevant Pages
|
