Abit's OFFICIAL resolution on Intel Temp mis-reading

[Ruantic]

Wow, this seems like kind of a "were sick of your whining and thats, that" statement..... This really is a poor response IMHO from Abit, and speaks poorly of there Support....

Read it HERE (http://forum.abit-usa.com/showthread.php?s=&threadid=11496)

[OPPAINTER]

They closed that thread.
The thing is, I'm reading -15c on the IC7, if I were to put it on a IT7 I'd be reading -25 at least. So when they say they use a certain calculation, why don't they have the same calculation for their other Intel mobo's.

Dummies :D

OPP

[RichBa5tard]

I dissagree about the semantics, it's explained in a clear and professional way. A pitty they locked the thread to avoid further discussion, but I don't think it's necessary for Abit to change their temp reading.

Who cares about the exact temperature anyway? As long as the temp difference is correct, i'm a happy camper. The only ppl who complain about "to high temps" are noobs (sorry if this offends anyone) imho. I don't see why you would favour motherboard "A" over "B" just because the temp sensor on "A" gives a lower temp. In the end, it's only stability that matters. :)


-edit-
OPP, i think it's most likely they've used a different I/O chips from different manufacturers, hence the same calculation can lead to different results.

[Ruantic]

Who cares about the exact temperature anyway? As long as the temp difference is correct

Well I really don't have a problem with the high reading, Annoying at most, but there are others out there with REAL ISSUES, my IC7-G is sitting here at 42 C its been folding for a week, so no big deal.... However the people out there that are getting temp readings high enough to sound alarms and shut down the system due to massive inaccuracy do indeed have a valid problem.

As Opp pointed out they closed the thread so further discussion is not an option, its even mentioned in the post to reply there, then it is locked? Maybe I expect to much from Abit, But I think they handled this one quite poorly....

[JeffPH]

cant they calibrate it like what MSI did in the NEO?

[OPPAINTER]

Originally posted by RichBa5tard



-edit-
OPP, i think it's most likely they've used a different I/O chips from different manufacturers, hence the same calculation can lead to different results. \

Thats true but they are stating that they believe they have the right temps on the IC7, if thats the case then they have the wrong temps on the IT7, what do they say about that?

OPP

[RichBa5tard]

Originally posted by OPPAINTER
\

Thats true but they are stating that they believe they have the right temps on the IC7, if thats the case then they have the wrong temps on the IT7, what do they say about that?

OPP


So the temperature you saw might be higher comparing to other motherboard with the same chipset, it is because we use the different parameter to calculate the temperature. We cannot say that we are more accurate, but we sure follow the standard provided by the I/O chip vender, and we believe it should be the proper one.

He doesn't mention any specific abit P4 motherboard, so i don't know why you think he's talking about the IC7.

Basicaly, they're blaming the I/O chip vender. If the temp doesn't seem to be correct, then the vender must have given them the wrong standard voltage...

[MrIcee]

Originally posted by JeffPH
cant they calibrate it like what MSI did in the NEO?

The problem with the NEO is the same. If you use the best performance bios posted here, the 1.32, it is reading 10c higher than actual, approximately the same difference Abit users are seeing. Going to the 1.4 bios corrects it but with a drop in total system performance....I went the "deduct the 10c from the temp route with my eyes" and kept the performance with the 1.32 bios.:)

Randi:D

[BrainStorm]

The most interesting thing about this as far as I'm concerned is the fact that the "thermal" diode isn't measuring temperature at all...measuring a voltage to be used to calculate a temperature.

That's whacked! Or at least it seems like it is. Anyone have any better info on exactly how and why this is supposed to work?

[mdzcpa]

Originally posted by BrainStorm
The most interesting thing about this as far as I'm concerned is the fact that the "thermal" diode isn't measuring temperature at all...measuring a voltage to be used to calculate a temperature.

That's whacked! Or at least it seems like it is. Anyone have any better info on exactly how and why this is supposed to work?

That's how all diodes measure temps:)

As far as Abit's explanation, I think it is reasonable. I agree 100% with RichBa5tard, static temp is irrelevant anyway. Only temp changes matter...be it load versus idle, or one cooling solution versus another. There is far too much variation from board to board, brand to brand to put any real value in a single temp report.

When it comes to people with shutdown problems due to high temps, I cannot help but wonder if these people have even gone into the BIOS to select a higher shutdown temp. The mandatory Intel overtemp circuit is a seperate reading and non adjustable, but does not engage until 135c...so that is NOT the cause of these shutdowns. That means that the shutdowns are coming from the Abit BIOS built in shutdown feature which works off the other sensor (the one that reports CPU temps) and is fully adjustable. So why don't these guys just select a higher setting for shutdown? duh.

[shogan191]

mdzcpa not everyone is that confident or competent. Maybe they are worried about burning up their cpu or board. Fortunately for everyone there are noobs about.:)

[twiztdskatr]

...And thats the reason DigiDocs and CompuNurses were made :D

[mdzcpa]

I know...I don't mean to be too harsh. But I hate to see cutting edge products being "dumbed down" so the less sophisticated mainstream feels better about their CPU temps.

BTW, the IC7 BIOS 1.3 diid in fact change (raise) the default temp for the shutdown feature. Looks like Abit is trying to help those who do not know any better to adjust it themselves:)

[shogan191]

naw, not too harsh. I just happen to be one of the noobs. :D :D I just got my first intel chip and can't get over how small it is. Maybe by next week I'll know enough to worry about my temps. I hope so.

[BrainStorm]

Originally posted by mdzcpa
That's how all diodes measure temps:)

That's exactly what I mean. I just didn't know that's how they worked until now, and as far as I'm concerned, it throws into doubt the belief I've seen espoused all over the net (and believed myself) that diode temps are more accurate than in socket thermistor temps. At least it's in doubt until someone explains to me the principle behind it and why it is really accurate so I can make up my own mind.


As far as Abit's explanation, I think it is reasonable. [/B]

I think it's reasonable too. However, again, the accuracy has to be called into question when the same setup (except for different mobo by same manufacturer that also "reads" from the diode) gives two very different cpu temps.

[mdzcpa]

Originally posted by BrainStorm
However, again, the accuracy has to be called into question when the same setup (except for different mobo by same manufacturer that also "reads" from the diode) gives two very different cpu temps.

I see your point here. But, that does not take into account the possiblity that the two boards, although made by the same manufacturer, are using two different I/O chips (the chip that interprets the diode signals). Nor does it address the possibility that if the I/O chip is the same, the manufacturers of the I/O chip have altered the chip, or have instructed the mobo maker to use different algorithms to interpret temps.

Heck, I'm not sure I've even ever had two boards of the same make and model report the same temp using the same CPU and cooling. That's how much variation in temp reading implementation that can exist.

[Mr Gadget]

This thermal design guide from Intel may help explain the operation of the temperature sensing and protection circuitry that they have built into the P4.

Thermal Design Guidelines (http://developer.intel.com/design/pentium4/guides/25216101.pdf)

The section covering the thermal diode is "3.4.7.1 Thermal Diode"

"The Pentium 4 processor with 512-KB L2 cache on 0.13 micron process incorporates an on-die
thermal diode, which can be used with an external device (thermal diode sensor) to monitor longterm
temperature trends. By averaging this data over long time periods (hours/days vs. min/sec), it
may be possible to derive a trend of the processor temperature. Analysis of this information could
be useful in detecting changes in the system environment that may require attention. Design
characteristics and usage models of the thermal diode sensors are described in datasheets available
from the thermal diode sensor manufacturers."

It appears that Intel intended this function to be used to measure relative long term temp changes vs absolute temp measurement.

Both the TH7II and the IC7 use the Winbond W83627HF I/O controller chip, the IT7 strangely enough uses something called "Super I/O" Perhaps this would explain the temp differences between the IT7 and IC7 that OPP noted.

The Winbond I/O chip takes the analogue output from the P4 thermal diode and converts it to a digital measurement that is loaded into an output register on the I/O chip. This register is read by the BIOs, Hardware Doctor MBM etc. It would appear that the MOBO manufacturer has much control over the accuracy of the conversion... other than chosing the supplier of the chip.

[BrainStorm]

Originally posted by mdzcpa
I see your point here. But, that does not take into account the possiblity that the two boards, although made by the same manufacturer, are using two different I/O chips (the chip that interprets the diode signals). Nor does it address the possibility that if the I/O chip is the same, the manufacturers of the I/O chip have altered the chip, or have instructed the mobo maker to use different algorithms to interpret temps.

Heck, I'm not sure I've even ever had two boards of the same make and model report the same temp using the same CPU and cooling. That's how much variation in temp reading implementation that can exist.

All of what you've mentioned is true and just leads inevitably to the conclusion that nothing can be completely trusted for absolute temps.

[ctgilles]

This IC7 is driving me mad with those temps :mad:
But then again... My Vapochill isn't mounted that good, JC is coming over to fix it :)

[JeffPH]

Originally posted by MrIcee
The problem with the NEO is the same. If you use the best performance bios posted here, the 1.32, it is reading 10c higher than actual, approximately the same difference Abit users are seeing. Going to the 1.4 bios corrects it but with a drop in total system performance....I went the "deduct the 10c from the temp route with my eyes" and kept the performance with the 1.32 bios.:)

Randi:D

Nice to see you have an MSI i should ask you some tips when i get my P4 C chip, am currently using the P4B chip. hope you dont mind :)

Jeff

[Nohto]

Hey Randy,
are you having any voltage problems with the v-core? On my first board I would set my voltage to 2.3, but it would show up in the bios as 1.87 with a fluctuation of anywhere from 1.7 to 1.87.

[MrIcee]

Originally posted by JeffPH
Nice to see you have an MSI i should ask you some tips when i get my P4 C chip, am currently using the P4B chip. hope you dont mind :)

Jeff

I WAS using the MSI 875P NEO LSR...but as of last night I'm working with an IC7. The MSI was a holdover until I could get an IC7 in my hands. Its a great board indeed with very good performance...but as all things go..I'll be selling it to re-invest in hardware upgrades for the test bench:)

Randi:D

[MrIcee]

Originally posted by Nohto
Hey Randy,
are you having any voltage problems with the v-core? On my first board I would set my voltage to 2.3, but it would show up in the bios as 1.87 with a fluctuation of anywhere from 1.7 to 1.87.

Nate:)

Good to see you my friend:)

I never took my Vcore higher than 1.8v on the MSI..so I couldn't tell you if my board was doing what yours does. Actually..my overclock on that board topped and 1.8v did not help...so I backed it down to 1.7v.

The voltages flucuate on the MSI quite a bit..like some early Abit AMD boards. The IC7 is rock steady in holding its voltages...and with a 5 sec non solder Vmem mod and 3.2v available...I've been benching 3Dmark at 286Mhz w/ Mem at 5:4 and timings maxed:)

With the MSI 271 was tops overall on the FSB...making a 200 Mhz overclock differential, 3794Mhz MSI vs. 4036Mhz Abit.

The MSI is a sweet,sweet board tho..I really do like it and its bundle of features along with great performance.:)

Randi:D

[Nohto]

Randy, Great to see you back at it!!!!!!!!!!!!!
My results were similar to yours as far as the Abit seemed to have the voltage regulation down a little (in my case a lot) better than the MSI. I could obtain a higher OC with the Abit also approximately the same as yours without any mods. Do you have a link to the Vmem mod for the IC7. I just picked up another 3.0c to test and hopefully it will be better than the one I have.
Nate

[OPPAINTER]

Originally posted by Nohto
Randy, Great to see you back at it!!!!!!!!!!!!!
My results were similar to yours as far as the Abit seemed to have the voltage regulation down a little (in my case a lot) better than the MSI. I could obtain a higher OC with the Abit also approximately the same as yours without any mods. Do you have a link to the Vmem mod for the IC7. I just picked up another 3.0c to test and hopefully it will be better than the one I have.
Nate

http://www.xtremesystems.org/forums/showthread.php?s=&threadid=12914

Mem mod.

OPP

[MrIcee]

LOL...mines easier than that;) But you need adjustable 3.3v line on your PSU:)

Randi:D

[Ruantic]

Originally posted by MrIcee
LOL...mines easier than that;) But you need adjustable 3.3v line on your PSU:)

Randi:D


any chance you could post it, even though mine is already done I would be interested to see it....

[MrIcee]

I have successfully used this method of increasing Vdimm after killing a few boards on difficult pin type mods. It is easy but you must use caution as always. I have utilized this for 6 months now for Vdimm mods without any repercussion or failure, but that is not to say it can't happen....

What I basically do is find the output leg of the Vdimm MOSFET and attach an SMD grabber to it. The other end has a metal pin attached to it that is inserted into the 3.3v line at the ATX connector. You must have a PSU capable of 3.3v line adjustment. I find that there is about .1v difference between the 3.3v line and what you end up with for actual mem voltage..at least with my PC Power and Cooling PSU. So in essence 3.42 gives me a steady 3.3v Vdimm. In the following pic I just took tonight while benching..the arrow points to the SMD grabber where it attaches to the output leg of the VDimm output MOSFET. Good luck and by all means be careful:)

Randi:D

[Ruantic]

thanks for taking the time to post that, actually thats a really good idea. The interesting thing is we could even take the idea further, build a nice little high accuracy regulation circuit and utilize it on more then just this... Could even be helpfull for those that have fried regulation circuits on NF7-S etc, just replace the VDD line with your own regulated source. Only thing I would think is adding a current limiting circuit for safety... HEHE, very nice, you've gotten me thinking!

Appreciate it greatly...

[MrIcee]

You're very welcome my friend:)

Indeed, I have been thu a couple different approaches to building regulation circuits, but this method has worked the best and flawlessly...that's why I hadn't furthered the regulation circuit buildup.

Randi:D

[macci]

Hehhe always nice to see MrIcee using an INTEL system :D
Nice work on 3DMark with your Ti4200 too :)

[MrIcee]

macci my friend:)

Thank you indeed:)

I am not finished with the Ti4200 work yet, this IC7 I have set up Friday evening has much better bandwidth due to the abilty to run higher FSB and tight timings compared to my retail MSI NEO.

I have a gem of a Ti4200 and really enjoy benching that class for sure...I will see if I can reach you in the coming weeks, but your excellent high FSB/bandwidth scores in Car and Lobby I cannot overcome...I'll have to really count on higher clocks to make any gain..tho with the Abit I've been achieving geat bandwidth at 286Mhz and 5:4 ratio.

As with your main rig being Intel...my allegiance and main rig remain AMD. I will be putting together an NF7-S Rev2 soon in the Prommy with a very special 1700+ and possibly a Barton for the benching adventures.

It is rather funny seeing my name and Intel side by side, but I had been considering building a second rig w/ Intel for some time for comparitive benching purposes, and for benching review hardware.

BTW...last night I ran a 35sec SuperPi at 1M, and 50.85sec PiFast run....not too shabby. I'll be posting up the screenshots soon, and thought it was not too bad.

Again..thanks my friend:)

Randi:D

[Manoj]

Guys do you want to test a new beta bios for IC7 (beta bios 1.4).

It is supposed to improve memory performance.

Regards,
Manoj Mahtani

[MrIcee]

Manoj:)

I'm up for something improved...post up a link my friend:)

Thank you for the offer indeed !

Randi:D

[JDLT1]

Yeah were is the beta 1.4 bios? Please do let us know.


JDLT1

[LikwidKool]

now this is the second time I have heard of this 1.4 bios. Me wants to try real bad!;)

[Player0]

MrIcee, good to see you are keeping as insane as ever. What did you do to that poor video card.

Love the clamp ;)

[MrIcee]

Player0:)

Just noticed the region you're from...not far from me. I'm 40 miles NE of Albany on the Vermont border:)

Vid card..just some volt mods and Swiftech cooling torture vise-gripped to it:D

Randi:D

[Player0]

Hey Icee,

Yeah I went to college up in Plattsburgh, and I lived in Glenns Falls for a while. Used to go to Berlington lots. I do like it up there...its a craphole down here ;)

Vise-grips...apart from duct tape, its the most widely used tool I think ;)

[jmke]

anyone buying an Abit motherboard can hardly be called a complete n00b, as he knew that Abit is supposed to be good

if you buy high tech gear, be sure you are ready for high tech action :)