Help my understanding please: GA-870A-UD3

[jolphil]

Hi Keith,
I too use this board and have commented on the board temps in various posts.. I also have commented on the North Bridge temperatures and have also directly asked Gigabyte what is the maximum temperature for the chip..They ignored my question..I hope you have better luck..
When I first discovered this condition I wrote this post..
http://forum.giga-byte.co.uk/index.php
Since then I fashioned a dedicated fan to direct airflow on the heat sink..That lowered the heatsink temperatures to something manageable..
Good luck to your investigation,
jolphil

Edit: I also don't have much faith in software temperature sensors readouts when they use SMD thermistors as the sensor..
I much prefer the good ole heat sink temperature as an indication...

[Dark Mantis]

Yes I would agree with jolphil regarding the dedicated fan as the best option. Some things are just best sorted out after manufacture by us users. Bit of customisation.

[KeithMyers]

Well GGTS only answered my question about which temp variable controls the CPU fan speed.  They offered no other information.  I've read everything I could find on the Northbridge temp problem and I never could decide whether it was real or not.  I took a dedicated fan and blew it directly on the heat sink and achieved exactly 1/2 degree change in the reported 82 degrees Centigrade reading reported.  The display would toggle between 81 and 82 degrees.  Meanwhile, I measured the actual die and heat sink temp with the temperature gun and came up with 55 degrees Centigrade.  Thus, I concluded that since I couldn't effect any real change in the reported temp and the measured temp was 55 degrees, the motherboard Northbridge temp readout was bogus. The only thing left to do would be to experiment by removing the heat sink and replacing the TIM with AS5 and fashioning a proper method of clamping the heat sink down to the chip so it would have proper force and couldn't move around or get knocked off.

[absic]

Well GGTS only answered my question about which temp variable controls the CPU fan speed. 

What was their reply to this question?

There have been several instances of strange NB temps being reported and this is down to faulty sensors. Re seating the cooler with new Thermal Grease can help although I have never had to do so, on any of the Gigabyte boards I have used.

[jolphil]

The only thing left to do would be to experiment by removing the heat sink and replacing the TIM with AS5 and fashioning a proper method of clamping the heat sink down to the chip so it would have proper force and couldn't move around or get knocked off.

My reported NB temp in both Linux (lmsensors) and Windows software  was 79C..I removed the Heat sink and TIM and replaced it with
a Enzotech CNB-S1R  solid copper heatsink with fins..My Reported temperature only went down to 78C..Not really worth the effort..
In addition the push pins would not fit the board holes and I had to use 632 nylon bolts and nuts to secure it flat to the chip..
One benefit of the the heatsink bolted,  it does not rock back and forth any more..
I did not have the nerve to measure the chip temperature without a heat sink attached for fear of damaging the chip..
Goodluck,
jolphil

[Beekeeper]

jolphil - 79 must be b...s..., some fake reading

KeithMyers - push ggts to give answers for all your questions
and if your cpu's "idle temps are only about a degree or so above ambient room temps" then it must be sth wrong with the software you use for reading. as i don't believe that cpu which consumes at least a few 10s of watts raises it's temp only by 1 degree higher than air in your room

[Dark Mantis]

No Beekeeper I am sure that you are right there, some of these figures that are reported are just not feasible. Either sticking sensors or bad software.

[bytheway_r]

Regarding the high NB temps I can only say that I replaced the heatsink on a 770T-UD3P with a Thermalright HR-05 ( Arctic Silver 5 ) and my temps reading went down from 80-82 to 78-79C. I can't say it ran cool on stock heatsink, though. It was bordering on too hot to the touch.

and if your cpu's "idle temps are only about a degree or so above ambient room temps" then it must be sth wrong with the software you use for reading. as i don't believe that cpu which consumes at least a few 10s of watts raises it's temp only by 1 degree higher than air in your room

I beg to differ. I'm pretty sure he has Cool'n'Quiet enabled which would downclock and undervolt the CPU when in idle. Now, whether the sensor is giving correct temperature is another thing but running within few degrees of ambient isn't all that hard.

Myself I have a bigger cooler and customized Cool'n'Quiet ( undervolted more ) and I'm running at 22C ( once again, this is just a sensor reading ) which can't be much higher than ambient. That's with CPU fan on auto running at 800 RPM ( max 1850 RPM ).

[jolphil]

I beg to differ. I'm pretty sure he has Cool'n'Quiet enabled which would downclock and undervolt the CPU when in idle. Now, whether the sensor is giving correct temperature is another thing but running within few degrees of ambient isn't all that hard.

I have to agree here..I forget what software I used but the power at down clock at idle verses full power full clock  FWIW stated
idle 12watts/22C to 120Watt/38C under full power.. Now these were software readings and with a Scythe Mugen 2 (an  enormous heat sink)
This was several months ago(winter) when it was around 20-21C ambient..At idle the CPU fan barely runs at all and did spin up
with either prime 95 or Intel burn test..I forget which was used..
jolphil

[Dark Mantis]

Fair enough if that is what you get, who am I to argue. I am more used to the Intel platform and they tend to run somewhat hotter. Even with my water cooling unless I use below ambient means my temps are higher than what you are getting.

[bytheway_r]

Ah, yeah, forgot for a moment there that Intel runs hotter. Or rather ran with previous gen. i7. Sandy Bridge is much like current Phenoms in terms of idle temps.

Don't get too cocky, though, DM. There's still the issue of sub zero temperatures and Intel CPUs .

Man, I really liked that saying about Intel+Nvidia being great space heaters, too .

[Dark Mantis]

There is alot of truth in that to be honest especially the GPUs. Nice on a cold winters day to sit around it and warm your hands It's more the noise I couldn't stand!

The below zero cooling is only an issue until you get to a decent overclock.

[KeithMyers]

What was their reply to this question?

Question:
"Which CPU sensor controls the CPU Fan speed? Which CPU sensor controls the automatic CPU throttling? Motherboard CPU temp sensor or CPU DTS sensor?"

Answer:
"ITE IT8720 chip controls all sensor on the motherboard"

All the monitoring programs I have used are reading the same temps as each other.  OpenHardware Monitor, CPUID HWMonitor and CoreTemp all agree when run simultaneously. The reported temps are pretty darn close to what I measure with the IR temp gun too.

Correct, I have both C1E and C'NQ enabled so when I dump BOINC work from the system and the CPU's and GPU's go Idle, they are just a degree or so above ambient for the room.  The house thermostat is set for 68 degree F. and the computer sits on the desk right next to an cracked open outside window and it has been rainy and cold here forever it seems. So the computer right now is getting very cool. Going to be a different situation come summer.

I just replaced the stock cooler on the Phenom II X4 965BE with a CoolerMaster Hyper 212+ and when I turned it on for the first time with the side cover off to make sure the fan spun up, I was diving for the power switch because the fan did not in fact spin up.  I was checking out the fan and and the motherboard for an hour trying to figure out what I had done wrong.  Spent some time researching the problem on the web and came to discover that the CPU cooler is such a good heat sink that the fan won't spin up under motherboard PWM control until the system sees that CPU temp start to rise.  With the knowledge, I then turned on the system and watched the computer go through POST, through my Boot Manager and finally load Windows7 and start up BOINC and then the fan finally kicked on once the CPU finally got occupied with all 4 cores crunching.  Proved to me that the new CPU cooler does a damn fine job even without a fan blowing through it.  Probably also helps that the top fan in the case draws a lot of air on its own through the CPU  cooler even without the CPU fan being on.  One of the reasons I chose to orient the cooler North-South was to take advantage of the natural air flow through the case.

Depending on the mix of work going through the system at any time, I have seen the motherboard temp sensor match the core DTS sensor equally or within a degree or so, plus or minus.  The core is reporting about 1 1/2 degree above the motherboard sensor for example right now.  From what I have read, a lot more people see greater differences between the two sensors.  I don't know whether I have a good motherboard or whether my temps just are a reflection of the type of work the computer does being atypical to most users.

I am pretty damn happy with this system all said.  If  I could just get some direct comments about the unusual behavior with a certain type of MW work, I would be ecstatic and would just shut up. I haven't had any of the really hard MW work now for almost a week, maybe it was just a specific set of data that really made the system grunt. Who knows.  It is all academic anyway as I verified I wasn't getting into thermal trouble with a IR temp gun verification.

Keith

[bytheway_r]

Probably also helps that the top fan in the case draws a lot of air on its own through the CPU  cooler even without the CPU fan being on.  One of the reasons I chose to orient the cooler North-South was to take advantage of the natural air flow through the case.

Unofrtunately, these 200mm fans aren't all that capable of moving air through heatsinks. I'd say it's more due to physics of warm air going up. My graphic card's 80mm fans are blowing air with so much force under load that it exits through the side 200mm fan .

If  I could just get some direct comments about the unusual behavior with a certain type of MW work, I would be ecstatic and would just shut up.

Computers, huh? For some reason I'm experiencing odd behaviour when reducing noise in photos with my CPU locked to overclocked values. Have yet to check if it's only in Canon's software or all programs. Not to mention that my keyboard seems to have stopped working again during boot .

[KeithMyers]

Probably also helps that the top fan in the case draws a lot of air on its own through the CPU  cooler even without the CPU fan being on.  One of the reasons I chose to orient the cooler North-South was to take advantage of the natural air flow through the case.

Unofrtunately, these 200mm fans aren't all that capable of moving air through heatsinks. I'd say it's more due to physics of warm air going up. My graphic card's 80mm fans are blowing air with so much force under load that it exits through the side 200mm fan .

If  I could just get some direct comments about the unusual behavior with a certain type of MW work, I would be ecstatic and would just shut up.

Computers, huh? For some reason I'm experiencing odd behaviour when reducing noise in photos with my CPU locked to overclocked values. Have yet to check if it's only in Canon's software or all programs. Not to mention that my keyboard seems to have stopped working again during boot .

Well, the convection of naturally rising hot air certainly plays a major role also.  Assisting the natural heat flow out the top with fans directing it that way couldn't hurt.

My two GPU's ostensibly have exit paths out the rear of the card at the back plane, but mostly all the hot air coming off their heat sinks dumps back into the case.  The rear case fan nearest the cards seems to do the bulk of the work removing that heat load from the  case.  It seems to intercept it before it impinges on the CPU fan.

I process my astro photos on my field laptop. Core 2 Duo at 1.6Ghz.  When I run a Photoshop CS filter or Action on one of my images,  I just walk away for about 15 minutes because for all intents, the computer is frozen. No need for overclocking to freeze that computer up.