[lm-sensors] Fintek F71805F on Jetway 627FWE1G

[lm-sensors] Fintek F71805F on Jetway 627FWE1G

[Forest Bond]

Hello,

I am getting what appear to be incorrect temperature readings for the CPU on a
627FWE1G mainboard, which has an F71805F.  The other two temperatures appear to
be reasonable, but the CPU temperature appears to be low (although it does seem
to scale appropriately in response to CPU burn-in, momentarily disconnecting the
fan, etc.).

From the documentation:

--------------------------------------------------------------------------------
All temperature channels are external, there is no embedded temperature
sensor. Each channel can be used for connecting either a thermal diode
or a thermistor. The driver reports the currently selected mode, but
doesn't allow changing it. In theory, the BIOS should have configured
everything properly.
--------------------------------------------------------------------------------

I'm wondering if the BIOS did not configure the sensor properly, although
the sensors program reports that the sensor is configured as a diode (which, I
believe, is correct).

Is it possible that the diode is not characterized correctly?

I have a technical contact at Jetway that may be able to assist.  Should I
pursue this as a BIOS issue, or ought I seek information with regard to how the
circuit is physically wired?

Any input you can provide is appreciated.  I am actually using this information
as a means of characterizing the vt1211 wiring on VIA EPIA boards with the same
CPU.  If I can get the information I need, I will be happy to pass it all along
to the lm-sensors project.

thanks,
Forest
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
Url : http://lists.lm-sensors.org/pipermail/lm-sensors/attachments/20070321/ab7f5a11/attachment.bin 

[lm-sensors] Fintek F71805F on Jetway 627FWE1G

[Jean Delvare]

Hi Forest,

I am the author of the f71805f driver.

On Wed, 21 Mar 2007 11:51:03 -0400, Forest Bond wrote:
> I am getting what appear to be incorrect temperature readings for the CPU on a
> 627FWE1G mainboard, which has an F71805F.  The other two temperatures appear to
> be reasonable, but the CPU temperature appears to be low (although it does seem
> to scale appropriately in response to CPU burn-in, momentarily disconnecting the
> fan, etc.).
> 
> From the documentation:
> 
> --------------------------------------------------------------------------------
> All temperature channels are external, there is no embedded temperature
> sensor. Each channel can be used for connecting either a thermal diode
> or a thermistor. The driver reports the currently selected mode, but
> doesn't allow changing it. In theory, the BIOS should have configured
> everything properly.
> --------------------------------------------------------------------------------
> 
> I'm wondering if the BIOS did not configure the sensor properly, although
> the sensors program reports that the sensor is configured as a diode (which, I
> believe, is correct).

If the type was improperly selected, you wouldn't see the reading
scaling properly as you report. So chances are good that this is indeed
a thermal diode that is connected.

> Is it possible that the diode is not characterized correctly?

Yes, this is possible. Hardware monitoring chips are typically meant to
be used with one given type of thermal diode, and where a different
type is used, the measurements have to be corrected. If the correction
is a simple offset, it can be done by the F71805F itself, otherwise it
must be done in software.

Another possibility is that this is a labelling issue, and one of the
other two temperatures is actually the CPU temperature. You said that
"the other two temperatures appear to be reasonable" but you didn't
give us the actual numbers. Please copy the output of "sensors" so that
we get an idea. You also did not tell us what CPU model it was.

> I have a technical contact at Jetway that may be able to assist.  Should I
> pursue this as a BIOS issue, or ought I seek information with regard to how the
> circuit is physically wired?

If you could know for sure which temperature channel is connected to
what, it would help. You might get a first idea by checking in the
BIOS, and comparing the hardware monitoring information there with what
"sensors" displays.

You can also dump the contents of the F71805F chip to find out if
temperature offsets have been set by the BIOS. Assuming that the
F71805F lives at the standard address 0x290, you would run:

isadump 0x295 0x296

And look at the values of registers 0x90, 0x91 and 0x92.

-- 
Jean Delvare

[lm-sensors] Fintek F71805F on Jetway 627FWE1G

[Forest Bond]

Hi Jean,

On Wed, Mar 21, 2007 at 05:21:04PM +0100, Jean Delvare wrote:
> I am the author of the f71805f driver.

First off, thanks for your time spent on this driver, and thanks for your quick
response.

> On Wed, 21 Mar 2007 11:51:03 -0400, Forest Bond wrote:
> > Is it possible that the diode is not characterized correctly?
> 
> Yes, this is possible. Hardware monitoring chips are typically meant to
> be used with one given type of thermal diode, and where a different
> type is used, the measurements have to be corrected. If the correction
> is a simple offset, it can be done by the F71805F itself, otherwise it
> must be done in software.

> Another possibility is that this is a labelling issue, and one of the
> other two temperatures is actually the CPU temperature. You said that
> "the other two temperatures appear to be reasonable" but you didn't
> give us the actual numbers. Please copy the output of "sensors" so that
> we get an idea. You also did not tell us what CPU model it was.

Well, the other two temperatures happen to be giving me room-temperature
readings (the mainboard is not in a case), and this is the one that rises
quickly when I unplug the CPU fan :)

sensors output and /proc/cpuinfo are attached.

> > I have a technical contact at Jetway that may be able to assist.  Should I
> > pursue this as a BIOS issue, or ought I seek information with regard to how
> > the circuit is physically wired?
> 
> If you could know for sure which temperature channel is connected to
> what, it would help. You might get a first idea by checking in the
> BIOS, and comparing the hardware monitoring information there with what
> "sensors" displays.

This BIOS does display temperature and voltage information at boot time.  Is
your intent simply for me to match up the readings with those read by sensors?
I am confident that the reading in question is, in fact, the CPU temperature.
Or do I misunderstand?

> You can also dump the contents of the F71805F chip to find out if
> temperature offsets have been set by the BIOS. Assuming that the
> F71805F lives at the standard address 0x290, you would run:
> 
> isadump 0x295 0x296
> 
> And look at the values of registers 0x90, 0x91 and 0x92.

Output of "isadump 0x295 0x296" is attached.  I am not quite sure how to
interpret that, but I belive it is indicating that an offset of 0x5d is being
applied to temp1, while temp2 and temp3 are being read as-is.  (Assuming the
chip is indeed at that address).

Thanks again,
Forest
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
Url : http://lists.lm-sensors.org/pipermail/lm-sensors/attachments/20070321/78f49085/attachment.bin 

[lm-sensors] Fintek F71805F on Jetway 627FWE1G

[Forest Bond]

On Wed, Mar 21, 2007 at 12:58:21PM -0400, Forest Bond wrote:
> sensors output and /proc/cpuinfo are attached.

> Output of "isadump 0x295 0x296" is attached.  I am not quite sure how to
> interpret that, but I belive it is indicating that an offset of 0x5d is being
> applied to temp1, while temp2 and temp3 are being read as-is.  (Assuming the
> chip is indeed at that address).

...and I forgot to attach all of those files.  Sorry about that.  They are
attached here.

-Forest
-------------- next part --------------
     0  1  2  3  4  5  6  7  8  9  a  b  c  d  e  f
00: 01 09 01 ff ff ff 00 00 ff 37 01 17 0f ff 0f ff 
10: cf af 87 8a 77 8c 87 86 77 00 00 00 23 23 ff ff 
20: 01 21 0f ff 0f ff ff ff 0f ff 0f ff 0f ff fc 00 
30: 00 00 00 80 00 08 80 00 08 00 80 00 ff ff ff ff 
40: fc 00 fc 00 fc 00 fc 00 fc 00 fc 00 fc 00 80 00 
50: fc 00 fc 00 ff 00 ff 00 ff 00 03 04 10 19 34 ff 
60: 24 30 10 80 00 03 93 03 19 00 00 ff ff ff ff ff 
70: 24 30 25 80 00 03 93 03 10 00 00 ff ff ff ff ff 
80: 24 30 25 80 00 03 93 03 10 00 00 ff ff ff ff ff 
90: 5d 00 00 ff ff ff ff ff ff ff ff ff ff ff ff ff 
a0: 64 64 64 ff 0f ff 0f ff 0f ff ff ff ff ff ff ff 
b0: 64 64 64 ff 0f ff 0f ff 0f ff ff ff ff ff ff ff 
c0: 64 64 64 ff 0f ff 0f ff 0f ff ff ff ff ff ff ff 
d0: 00 00 00 00 00 00 00 ff ff ff ff ff ff ff ff ff 
e0: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff 
f0: 80 00 00 00 00 ff ec ff ff 00 00 00 ff ff ff ff 
-------------- next part --------------
processor	: 0
vendor_id	: CentaurHauls
cpu family	: 6
model		: 9
model name	: VIA Nehemiah
stepping	: 8
cpu MHz		: 997.271
cache size	: 64 KB
fdiv_bug	: no
hlt_bug		: no
f00f_bug	: no
coma_bug	: no
fpu		: yes
fpu_exception	: yes
cpuid level	: 1
wp		: yes
flags		: fpu vme de pse tsc msr cx8 mtrr pge cmov pat mmx fxsr sse rng rng_en ace ace_en
bogomips	: 1995.86
clflush size	: 32

-------------- next part --------------
f71805f-isa-0290
Adapter: ISA adapter
+3.3V:     +3.31 V  (min =  +0.00 V, max =  +4.03 V)  
Vtt1.2V:   +1.40 V  (min =  +0.00 V, max =  +2.02 V)  
Vram:      +2.16 V  (min =  +0.00 V, max =  +4.03 V)  
Vchip:     +3.45 V  (min =  +0.00 V, max =  +6.31 V)  
+5V:       +4.96 V  (min =  +0.00 V, max = +10.59 V)  
+12V:     +12.23 V  (min =  +0.00 V, max = +22.18 V)  
Vcc1.5V:   +1.08 V  (min =  +0.00 V, max =  +2.02 V)  
Vcore:     +1.07 V  (min =  +0.00 V, max =  +1.02 V)  ALARM
5VSB:      +5.00 V  (min =  +0.00 V, max = +10.59 V)  
CPU Fan:  5190 RPM  (min =    0 RPM)                  
Sys Fan:     0 RPM  (min =    0 RPM)                  
Aux Fan:     0 RPM  (min =    0 RPM)                  
CPU Temp:     +0 C  (high =  +255 C, hyst =    +0 C)         [diode]
Sys Temp:    +35 C  (high =  +255 C, hyst =    +0 C)         [thermistor]
Aux Temp:    +35 C  (high =  +255 C, hyst =    +0 C)         [thermistor]

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
Url : http://lists.lm-sensors.org/pipermail/lm-sensors/attachments/20070321/f2012e60/attachment-0001.bin 

[lm-sensors] Fintek F71805F on Jetway 627FWE1G

[Jean Delvare]

Forest,

On Wed, 21 Mar 2007 12:58:21 -0400, Forest Bond wrote:
> On Wed, Mar 21, 2007 at 05:21:04PM +0100, Jean Delvare wrote:
> > Another possibility is that this is a labelling issue, and one of the
> > other two temperatures is actually the CPU temperature. You said that
> > "the other two temperatures appear to be reasonable" but you didn't
> > give us the actual numbers. Please copy the output of "sensors" so that
> > we get an idea. You also did not tell us what CPU model it was.
> 
> Well, the other two temperatures happen to be giving me room-temperature
> readings (the mainboard is not in a case), and this is the one that rises
> quickly when I unplug the CPU fan :)
> 
> sensors output and /proc/cpuinfo are attached.

> CPU Temp:     +0 C  (high =  +255 C, hyst =    +0 C)         [diode]
< Sys Temp:    +35 C  (high =  +255 C, hyst =    +0 C)         [thermistor]
> Aux Temp:    +35 C  (high =  +255 C, hyst =    +0 C)         [thermistor]

Hmm, 35 degrees C in your room? Where do you live? :) Do temp2 and
temp3 change at all?

So this "0 degree C" for CPU is raising on load / fan unplug? Up to
what value?

> > If you could know for sure which temperature channel is connected to
> > what, it would help. You might get a first idea by checking in the
> > BIOS, and comparing the hardware monitoring information there with what
> > "sensors" displays.
> 
> This BIOS does display temperature and voltage information at boot time.  Is
> your intent simply for me to match up the readings with those read by sensors?
> I am confident that the reading in question is, in fact, the CPU temperature.
> Or do I misunderstand?

If you think the labels are correct, alright. However, you still want
to compare temperature values themselves. How many temperatures does
the BIOS display? What are the values?

> > You can also dump the contents of the F71805F chip to find out if
> > temperature offsets have been set by the BIOS. Assuming that the
> > F71805F lives at the standard address 0x290, you would run:
> > 
> > isadump 0x295 0x296
> > 
> > And look at the values of registers 0x90, 0x91 and 0x92.
> 
> Output of "isadump 0x295 0x296" is attached.  I am not quite sure how to
> interpret that, but I belive it is indicating that an offset of 0x5d is being
> applied to temp1, while temp2 and temp3 are being read as-is.  (Assuming the
> chip is indeed at that address).

Yes, the chip is at that address, as can be seen in the first line of
sensors' output.

You are correct, an offset is applied to temp1. This is a 7 bit value
in 2's complement format, so 0x5d translates to... -33 degrees C. What
an offset, no surprise that you get really low values.

You can reset the offset using the following command if you want:

isaset 0x295 0x296 0x90 0x00

Now the question is, why is Jetway setting this offset? A question you
want to ask to your technical contact. This indeed looks like a BIOS
issue.

-- 
Jean Delvare

[lm-sensors] Fintek F71805F on Jetway 627FWE1G

[Forest Bond]

On Wed, Mar 21, 2007 at 06:32:19PM +0100, Jean Delvare wrote:
> Forest,
> 
> On Wed, 21 Mar 2007 12:58:21 -0400, Forest Bond wrote:
> > On Wed, Mar 21, 2007 at 05:21:04PM +0100, Jean Delvare wrote:
> > > Another possibility is that this is a labelling issue, and one of the
> > > other two temperatures is actually the CPU temperature. You said that
> > > "the other two temperatures appear to be reasonable" but you didn't
> > > give us the actual numbers. Please copy the output of "sensors" so that
> > > we get an idea. You also did not tell us what CPU model it was.
> > 
> > Well, the other two temperatures happen to be giving me room-temperature
> > readings (the mainboard is not in a case), and this is the one that rises
> > quickly when I unplug the CPU fan :)
> > 
> > sensors output and /proc/cpuinfo are attached.
> 
> > CPU Temp:     +0 C  (high =  +255 C, hyst =    +0 C)         [diode]
> < Sys Temp:    +35 C  (high =  +255 C, hyst =    +0 C)         [thermistor]
> > Aux Temp:    +35 C  (high =  +255 C, hyst =    +0 C)         [thermistor]
> 
> Hmm, 35 degrees C in your room? Where do you live? :) Do temp2 and
> temp3 change at all?

Sorry, I live in a non-Celsius country, and didn't realize how hot that is. :)

Those temperatures do fluctuate, however, and the readings from them are
approximately the same temperatures I was getting from a laser temperature gun
on the surface of the mainboard.

> So this "0 degree C" for CPU is raising on load / fan unplug? Up to
> what value?

Yes, it does climb up to the 30-40 C range, at least, when I run my CPU burn
utility and unplug the CPU fan momentarily.  I don't recall the exact value, but
it did seem to be a typical fluctuation (under the circumstances).

I imagine that value gets chopped at zero rather than going negative.  When I
grabbed the data, the CPU burn utility was not running, and the CPU fan was on.
I didn't realize until after that that the CPU temperature was being reported as
0C.

> > > If you could know for sure which temperature channel is connected to
> > > what, it would help. You might get a first idea by checking in the
> > > BIOS, and comparing the hardware monitoring information there with what
> > > "sensors" displays.
> > 
> > This BIOS does display temperature and voltage information at boot time.  Is
> > your intent simply for me to match up the readings with those read by
> > sensors?  I am confident that the reading in question is, in fact, the CPU
> > temperature.  Or do I misunderstand?
> 
> If you think the labels are correct, alright. However, you still want
> to compare temperature values themselves. How many temperatures does
> the BIOS display? What are the values?

The BIOS displays two temperatures:

CPU Temp: 12 C
System Temp: 37 C

What do you suppose is the third temperature I see from sensors?  Could it be a
duplicate reading of the system temperature?  Or is the BIOS just ignoring a
temperature sensors for some reason?

> > > You can also dump the contents of the F71805F chip to find out if
> > > temperature offsets have been set by the BIOS. Assuming that the
> > > F71805F lives at the standard address 0x290, you would run:
> > > 
> > > isadump 0x295 0x296
> > > 
> > > And look at the values of registers 0x90, 0x91 and 0x92.
> > 
> > Output of "isadump 0x295 0x296" is attached.  I am not quite sure how to
> > interpret that, but I belive it is indicating that an offset of 0x5d is being
> > applied to temp1, while temp2 and temp3 are being read as-is.  (Assuming the
> > chip is indeed at that address).
> 
> Yes, the chip is at that address, as can be seen in the first line of
> sensors' output.
> 
> You are correct, an offset is applied to temp1. This is a 7 bit value
> in 2's complement format, so 0x5d translates to... -33 degrees C. What
> an offset, no surprise that you get really low values.
> 
> You can reset the offset using the following command if you want:
> 
> isaset 0x295 0x296 0x90 0x00
> 
> Now the question is, why is Jetway setting this offset? A question you
> want to ask to your technical contact. This indeed looks like a BIOS
> issue.

Agreed.  I will pose this question to my contact at Jetway, and perhaps we can
get to the bottom of this.

Thanks again for your time, Jean.

-Forest
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
Url : http://lists.lm-sensors.org/pipermail/lm-sensors/attachments/20070321/4e910bc2/attachment.bin 

[lm-sensors] Fintek F71805F on Jetway 627FWE1G

[Jean Delvare]

Forest,

On Wed, 21 Mar 2007 14:05:03 -0400, Forest Bond wrote:
> On Wed, Mar 21, 2007 at 06:32:19PM +0100, Jean Delvare wrote:
> > Hmm, 35 degrees C in your room? Where do you live? :) Do temp2 and
> > temp3 change at all?
> 
> Sorry, I live in a non-Celsius country, and didn't realize how hot that is. :)

Poor you ;)

You might enjoy "sensors -f" then.

> > So this "0 degree C" for CPU is raising on load / fan unplug? Up to
> > what value?
> 
> Yes, it does climb up to the 30-40 C range, at least, when I run my CPU burn
> utility and unplug the CPU fan momentarily.  I don't recall the exact value, but
> it did seem to be a typical fluctuation (under the circumstances).

I'm not sure you want to do that, actually 33 + 40 leads to 73 degrees
C which is quite high for a Nehemiah. Don't go frying your CPU just to
demonstrate that your BIOS has a bug ;)

> I imagine that value gets chopped at zero rather than going negative.  When I
> grabbed the data, the CPU burn utility was not running, and the CPU fan was on.
> I didn't realize until after that that the CPU temperature was being reported as
> 0C.

Yes, the value gets chopped at 0 by the F71805F.

> The BIOS displays two temperatures:
> 
> CPU Temp: 12 C
> System Temp: 37 C

OK, so the CPU temp is unreasonably low just as it is under Linux.

> What do you suppose is the third temperature I see from sensors?  Could it be a
> duplicate reading of the system temperature?  Or is the BIOS just ignoring a
> temperature sensors for some reason?

Cannot be a duplicate reading, physically it wouldn't make sense. It
really has to be a 3rd sensor. Either the BIOS is averaging temp2 and
temp3 into "System Temp", or they indeed just ignore temp3. Given that
temp2 appear to have the same value anyway, no big deal.

> Agreed.  I will pose this question to my contact at Jetway, and perhaps we can
> get to the bottom of this.
> 
> Thanks again for your time, Jean.

You're welcome. I'm happy to see my driver is widely used :)

-- 
Jean Delvare