HW Monitor values for GA-970A-UD3

I'm hoping there are some other Gigabyte GA-970A-UD3 owners on here that can help me figure this out. I've been scratching my head over the HW Monitor values for the last few weeks trying to figure out what is what. HW Monitor gives me three temperature values: TMPIN0, TMPIN1, and TMPIN2. It's pretty obvious that TMPIN0 matches the "System temperature" reading in the bios so this must correspond to some sensor on the motherboard. The TMPIN0 reading usually stays at 33C which is what I always see in the bios. The other two values have me stumped though. For the last couple of weeks I've been under the impression that TMPIN2 is the CPU temp because it seems to fluctuate the most based on CPU load. It will go anywhere from 23C to 54C based on CPU load. However, the "CPU temperature" reading in the bios stays right around 27C idle, which is a few degrees higher than the TMPIN2 value in HWM. Furthermore, the TMPIN1 value in HWM idles around 27C, which makes me think that TMPIN1 is actually the CPU temp and NOT TMPIN2 like I had previously thought. The only thing keeping me from believing that TMPIN1 is the CPU temp is that TMPIN1 maxes out at only 42C under full load at a 4.5GHz overclock. This seems very low to me considering I only have a Cooler Master Hyper 212 Plus. Is it possible that my 212 Plus is keeping my 8 core CPU at only 42C?

To sum all of that up, my questions for other GA-970A-UD3 owners are:
1) Which HWM value is actually the CPU temp? TMPIN1 or TMPIN2?
2) If TMPIN1 is the CPU temp then what is TMPIN2? Northbridge?

It makes sense to me that the Northbridge would be running hotter than the CPU, but it does NOT make sense to me that my CPU temp would max out at 42C with a measly CM Hyper 212+.

Double post time.

I found this thread that claims TMPIN1 is the VRM temperature and TMPIN2 is actually the CPU temperature. So now I'm wondering... why does the temperature reading in the bios for CPU match the VRM reading in HWM. And why is the CPU reading in HWM 4C lower than the CPU reading in the bios?

Triple post. After reading a number of threads on this in our forums and other forums it seems that there is no consensus on this. A lot of people are saying that TMPIN2 is CPU and an equal number of people are saying that TMPIN1 is CPU. It would really be nice to get this figured out!

Is there anyone who can chime in on this?

As far as I can tell, TMPIN1 is the CPU temp (reading a bit less than my core temp readings), TMPIN2 is the NB temp.

Only thing I'm working from is that my TMPIN2 temp went up by 2 degrees when I upped my CPU VID NB to 1.20v.

However, the "CPU temperature" reading in the bios stays right around 27C idle = Yes it would in bios since it is not reading a loaded temperature.

Have you got a capture of HWMonitor under load? Run P95 Blend for about 30 mins to be sure the temps are fully up as high as likely to go and post while P95 Blend is running caputure and then post the capture so we can look at the temps shown.

I have seen temp 1 and 2 close together in HWM and that is a pain to know which is which on that dang Giga board, but neither temps was near an issue so it was not a big issue.

I have seen users speak of blowing freon on the VRMs to see temps drop and know for sure which of those temps and a particular board was the VRM temp and that made the other the CPU/Socket temp.

I just thought of one idea that might work. See below.

HWiNFO/32/64 Tools Website has freeware download. You would have to post pic of it since it has so dang much info that is really useless when trying to help a newbie that is just confusing and hard to scan all the numbers. But it might let us correlate your HWM temps and know which is which.

With gigabyte boards it has been established that TMPIN2 is the CPU socket temp.

Here are my temps after 30 minutes of P95. As you can see, TMPIN1 is much closer to the AMD FX-8320 package temperature (and I have no idea what that is). TMPIN2 is the highest value maxing out at 52c. On past desktops my northbridge temps were always higher than my CPU temps, which makes me think TMPIN2 is northbridge and TMPIN1 is CPU, but this seems to go against the majority of opinions on the internet for Gigabyte temps.



And to further demonstrate the confusion... Here is hwinfo showing that the "CPU" temp corresponds with HWMonitor's TMPIN2 and AMD Overdrive showing that the CPU temp corresponds with HWMonitor's TMPIN3.

Good deal on the THREE captured Images...

...since we had three images to Compare across the board and knowing that AMD AOD shows Socket Temp for it's temperature, it was possible to sort the rest since I have seen some HWINFO64 for the motherboards before.

I think the graphic below will help shake-out the temps as we know them in HWMonitor which we use all the time. At least for the often thought strange Giga UD3 temp read-outs.

IF you use Firefox as I do just right click the image and choose View Image for full size viewing. Because it truly is a big image at 2400 by 1100 pixels. RGone...ster.

Thanks RGone. That makes sense. I was thinking about unplugging one of my heatsink fans to see which temperature reading goes up, but if you're correct and TMPIN1 is the CPU-NB then I should see both temperatures rise, which wouldn't help me much in determining which is the CPU socket.

The only think that I still can't figure out is that the CPU temperature in the bios idles around 27C, which is withing 1 or 2 degrees of TMPIN1's idle temp. TMPIN2 (what we think is the socket) tends to idle around 3 to 4 degrees cooler. So, A) is the CPU temperature in bios actually giving me the CPU-NB temp, or B) is the CPU reading in the bios 4 degrees off from the CPU reading in HWM, or C) does my CPU idle 4 degrees cooler in Windows than it does in the bios?

Edit: Also, RGone, thank you for helping me to identify the "package" temps as the core temp. My next question is, then, which temperature should I give the most consideration to when overclocking the CPU? Should I pay more attention to the core temp or to the socket temp?

1. Today that being with newer motherboards, I do not pay or make that I seldom if ever pay attention to bios temps unless it is a brand new setup and I just fired it up and have not even been to windows for the first time. This so I can quickly see if I have Super High temps at all. After that I don't pay any attention to bios temps.

2. Now you want to know why I no longer pay attention to bios temps with newer mobo hardware. Here is my answer. At one time in history, the bios was setup to force the cpu to right at 100% load. If at that time the load was 100% I could use the bios to burn-in my Heat Sink compound. Plus IF bios forced Cpu to ~100% load then the temp I saw in bios could be expected to be roughly the highest I should ever see. That all worked BACK THEN. No more.

3. Today I get tired of g00gling for information that is just no longer paid attention to. I get numbers like the bios now loads Cpu from as low as 35% up to maybe about 50% or a shade more. But no more or less cut and dried WFO on the processor. So when I cannot nail it down closer than that...then screw a bios temp except at the very first fire-up of a mobo. So based on what I have found the Cpu Idle in winders could be cooler than Idling in bios. Cooler at Idle in winders by some amount. The amount of course would vary depending on that Bios Loading factor that is no longer verifiable.

4. I pay attention about equally to Both the Core Temp and the CPU Temp. Maybe a shade more to the Core temps. Make that for numbers folks; 40% attention to CPU Temp and 60% to the Core Temps. About it. RGone...

Sounds good to me. I feel a lot more comfortable with my Hyper 212+ knowing that my core temps max out at 39C. That's not bad for a heatsink that I paid $15 for

Those are pretty good temps for the Vcore to cpu to cpu Mhz ratio.

"ssjwizard" has a good thead where he compared his 8120 to his 8320. D*mn good read. What was pretty evident was that the earlier 8 core BD processors clocked up and gave performance increase and had reasonable HEAT output up to about 4.3Ghz and then leveled off. You could push more but the extra Vcore and resultant HEAT just did not make sense except for E-pene. IMO and YMMV.

Now we have the PD processor that clock for clock does more work than the BD. I have not tested to the same extent with my newer FX-8350 that I did with my FX-8120 and found the exact same thing "ssjwizard" found and saw. I even posted a graph in his 8120 vs 8320 thread that correlated on my FX-8120; exactly what he had seen on his. Now does the later 8 core, PD processor level off at 4.3/4.4 or is it 4.5Ghz? I do not know since I have not tested for such results.

I will say this. I run my FX-8350 at 4.5Ghz on CHV non-Z mobo at 1.38Vcore using a non-PD bios. I can stand that for sure. RGone...ster.

RGone said:

Those are pretty good temps for the Vcore to cpu to cpu Mhz ratio.

"ssjwizard" has a good thead where he compared his 8120 to his 8320. D*mn good read. What was pretty evident was that the earlier 8 core BD processors clocked up and gave performance increase and had reasonable HEAT output up to about 4.3Ghz and then leveled off. You could push more but the extra Vcore and resultant HEAT just did not make sense except for E-pene. IMO and YMMV.

Now we have the PD processor that clock for clock does more work than the BD. I have not tested to the same extent with my newer FX-8350 that I did with my FX-8120 and found the exact same thing "ssjwizard" found and saw. I even posted a graph in his 8120 vs 8320 thread that correlated on my FX-8120; exactly what he had seen on his. Now does the later 8 core, PD processor level off at 4.3/4.4 or is it 4.5Ghz? I do not know since I have not tested for such results.

I will say this. I run my FX-8350 at 4.5Ghz on CHV non-Z mobo at 1.38Vcore using a non-PD bios. I can stand that for sure. RGone...ster.

Click to expand...

That sounds pretty similar to my results. I can run up to 4.2GHz on stock 1.4 vcore and 4.4GHz with 1.425 vcore with low temperatures. However, I needed 1.475 to get 4.5GHz stable last time I tried (didn't try too hard though) and that caused a ~5 degree increase in socket temp (at the time I wasn't paying attention to the "package" core temps).

I will definitely hunt down ssjwizard's thread and give it a read, and I'll experiment with some overclocking and pay closer attention to my core temps to see where/if it levels off. I was able to boot Windows and run SuperPi for 20 minutes at 4.8GHz with 1.5 vcore, but I didn't test for stability with Prime95.

RGone said:

...I think the graphic below will help shake-out the temps as we know them in HWMonitor which we use all the time. At least for the often thought strange Giga UD3 temp read-outs.

Click to expand...

Very helpful. Thx.

FlailBoy said:

Very helpful. Thx.

Click to expand...

Thank you kindly sir. RGone...