On Fri, 14 Dec 2007 08:28:57 -0500, Crackles McFarly
<IrelandSux@ireland.sux> wrote:

>On Fri, 14 Dec 2007 03:28:17 -0500, kony <spam@spam.com> sayd the
>following:
>
>>On Thu, 13 Dec 2007 23:03:40 -0500, Paul <nospam@needed.com>
>>wrote:
>>
>>
>>>For measurement programs, you cannot run programs like Speedfan and
>>>Asus Probe, at the same time. They can both be talking to the
>>>hardware at the same time, and doing things to it. There are no
>>>agreed protocols, for shared control of such devices. (At least
>>>Asus won't change the way they do things.)
>>
>>
>>While it is true you can't assume two hardware monitoring
>>apps can run at the same time, they may... for example when
>>I ran the latest version of speedfan to see what it was
>>doing on a board here, I already had MBM5 running and they
>>peacefully co-existed.
>
>I thought MBM was discontinued?
>

Maybe no new development? Certainly still findable and
downloadable on the 'net.


>Will the MBM work with this A8S-X board still?


I don't know... try it and see.

On Fri, 14 Dec 2007 11:45:16 -0500, Paul <nospam@needed.com>
wrote:

>> LM99 (ID=$31) found on SMBus at $4C

>I wonder if the LM99 is actually on the graphics card ?

Yes, it is the video card. Running Speedfan here I also get
a $4C address which in the configuration menu shows up as
"I2CNVIDIA SMBUS", which refers to two temp values for the
video card core temp and video card ambient temp... these
also seen in the Display Properties page when a new enough
nVidia video driver is installed.

However on mine $4C is listed as a max6648 instead of LM99,
but it's an FX5700 Personal Cinema card instead of 6800.
The physical chip is on the card back, but it is a max6649
instead of 6648 and stamped as "6649 MUA". Speedfan is
calling the GPU temp the "remote" temp and the adjacent
sense transistor the "local" temp.



>An LM99 is apparently a diode type, and expects a 2N3904
>equiv just like the 8712F does. (It might not even
>physically be an LM99, but an emulation of one by
>another chip. Hardware can be tricky that way.)
>
>http://cache.national.com/ds/LM/LM99.pdf
>http://www.national.com/images/pkg/mua08a.jpg

On the following picture of a 6800GS, the temp sensor chip
"appears" to be the SOIC-8 on the left, bottom edge of the
card. Might be labeled "SU9" and it is an LM chip but I
can't make out the rest of the marking on it, nor the
marking on the remote sense transistor about 7mm below it.

http://www.ixbt.com/video2/images/palit ... n-back.jpg

On Fri, 14 Dec 2007 17:38:13 -0500, kony <spam@spam.com>
wrote:

>On the following picture of a 6800GS, the temp sensor chip
>"appears" to be the SOIC-8 on the left, bottom edge of the
>card. Might be labeled "SU9" and it is an LM chip but I
>can't make out the rest of the marking on it, nor the
>marking on the remote sense transistor about 7mm below it.
>
>http://www.ixbt.com/video2/images/palit-3/palit-6800gs-scan-back.jpg

I may be wrong about this, another picture I saw makes it
look like the chip I indicated is an LM358 opamp.
http://img114.imageshack.us/img114/9808/5hn4.jpg

On Fri, 14 Dec 2007 02:25:49 -0800 'Jon Danniken'
wrote this on alt.comp.hardware.homebuilt:

>"homebuilder" wrote:
>>
>> On Wed, 12 Dec 2007 07:53:40 -0500 'Paul'
>> wrote this on alt.comp.hardware.homebuilt:
>>
>>>The author of Speedfan may know more about what his program is doing,
>>>than we do.
>>
>> IIUC there is no way for SpeedFan to identify which reading applies
>> to which sensor. The subject's been debated on the SpeedFan forum
>> before now. Alfredo has provided some guidance on his website which
>> is usually ok for folks with 2-3 sensors but where you have multiple
>> sensors, it gets more difficult.
>> What's really needed is more info from mobo manufacturers to ID
>> the sensor names and what they monitor...
>
>IIRC, that was the reason the fellow who made MBM stopped doing it, because
>of the runarounds and hassles he received trying to get the information from
>the manufacturers and vendors.

Interesting thanks. It's odd because the mobo manufacturers could
provide the info quite easily, one wonders why they don't as there's
obviously a demand for it. Maybe they want users only to use their
s/w to regulate fan speeds etc.

Somewhere on teh interweb Paul typed:
> Crackles McFarly wrote:
>> On Wed, 12 Dec 2007 16:54:23 +0930, jmc
>> <NOnewsgroupsSPAM@NOjodiBODY.HOMEus> sayd the following:
>>
>>> Suddenly, without warning, Crackles McFarly exclaimed (12/12/2007
>>> 10:51 AM):
>>>> On Tue, 11 Dec 2007 17:39:29 -0600, Ed <nospam@hotmail.com> sayd
>>>> the following:
>>>>
>>>>> On Tue, 11 Dec 2007 11:50:04 -0500, Crackles McFarly
>>>>> <IrelandSux@ireland.sux> wrote:
>>>>>
>>>>> IT8712F-1 for CPU temps, fan speeds and some other stuff.
>>>>>
>>>>> LM99,LM75 would be for other devices like video card temps.
>>>>>
>>>>> No doubt Paul will provide a page full of info as soon as I post
>>>>> this LOL!
>>>>
>>>> What does the sensor labeled "AMD K8 Sensor" measure?
>>>>
>>>> thanks
>>>>
>>> I wonder about this one too. I use Speedfan, and am pretty sure the
>>> IT8712F-1 is CPU temp, because the "AMD K8 (Core)" seems too low. Right
>>> now the first reads 60c (hmmm. bit warm today; usually it's
>>> closer to 50c) and Core reads about 48, on my AMD Dual-core 5600+.
>>>
>>> jmc
>>
>>
>> On mine the "AMD K8 Sensor" rises FAST to any actions I perform. But
>> the IT8712F-1 rises much Slower to things I do.
>>
>> I guess the slower rising sensor is the accurate one?
>> Or is it just as plausible the fast rising sensor is accurate?
>>
>> As an example, IT8712F-1 "creeps" to 55-59C using SpyBot.
>> But the "AMD K8 Sensor" IMMEDIATELY JUMPS to 55-59C when SpyBot is
>> running.
>>
>> kony?
>> anyone?
>
> The author of Speedfan may know more about what his program is doing,
> than we do.
>
> In principle, the author of Speedfan doesn't know what is there. What
> he does, is probes the various low speed busses in the computer,
> looking for responses from known sensing devices. For example, he was
> probably
> able to discover the presence of the ITE 8712F without too much
> trouble.
> The device has a fixed number of channels. I don't know if there are
> any standards for passing a definition of what is connected, to a user
> program. The ITE8712F probably has three thermal channels, like
> many of the other Super I/O chips that have hardware sensor support.
>
> (PDF page 117 section 9.6.3.3 has a figure)
> http://web.archive.org/web/200610181411 ... V0.9.1.pdf
>
> Of the three channels, they can operate in "diode" or "thermistor"
> mode. In thermistor mode, a NTC thermistor is placed in series with a 10K
> resistor. The result is a voltage that varies with temperature. The
> hardware block either reads out a number for the raw voltage, or in
> some devices, they have a lookup table for instantly giving a (potentially
> incorrect)
> temperature.
>
> In the diode case, the assumption is that the device responds
> similarly to a 2N3904 transistor junction. A fixed current is put through
> the
> diode, and the developed voltage is measured. (The Athlon64 has a
> diode inside it, as part of the silicon die, and the anode and cathode
> pins on the
> processor, are connected as if they were a 2N3904.)
>
> You could be correct, in that a rapidly rising reading means the
> thermal diode in the processor is being used. But if that is the
> case, what device is making the measurement, and what channel is
> being used ? I like to be able to account for all devices and
> channels, before concluding I've identified them all correctly.
> The 8712F has three channels, and if you are getting more than
> three temp readings, then it means another device could be involved.
> Note that Speedfan uses SMART (supported inside the hard drive)
> to read out hard drive temperatures, so that is where those readings
> come from.
>
> So, I'd say, "yes, the fast rising one is probably the processor".
> But I cannot be certain, unless all devices and channels are
> accounted for in some way. It doesn't make sense to provide
> both a CPU diode readout, and then also put a thermistor
> in the CPU socket. In the case of S939, where would you
> put such a sensor ? It would have to be soldered to the
> back of the motherboard, and that would not function well
> at all. There just isn't room on S939 for a thermistor.
> Sockets with an open center, like my S478, are better
> candidates for using a thermistor soldered to the
> motherboard.
>
> Paul

There could be a thermistor under the IHS but external to the die maybe?
That would fit with the observed behaviour of rising more slowly but
reaching similar temps.
--
TTFN,

Shaun.