CPU Temp 'Issues'...HELP!

[[Dr. V]]

New thread with 800x600 pics (sorry Capp).

IDLE:


LOAD:


3 different programs, 3 different sets of temperatures.

I understand that programs sense the temps differently that's fine, but this is quite the difference. Especially considering the motherboard reads the temp between 35 and 37 upon startup.

Which one is right?

I used another program - SISoftware SANDRA - and it's giving me the same temps as the motherboard for idle, and between 50 and 53 on full load. There is also a marking, "td", beside the temp when it's displayed in SANDRA (no clue what that means).

I've checked the mounting, I checked the pattern of the thermal paste, everything fastened correctly and snug.

Help?

P.S. This is an Intel Pentium E2140 @ 350x8 = 2800MHz.

Thanks.

[Yellowhello]

Ever since multi-core CPU's there have been temperature differences between the cores. See HERE.

Originally from Rich's review of the QX9650
However, what really stood out during testing was the temperatures. At default clock speeds, cores #0 and #1 idled at 48C, while cores #2 and #3 were in the 30-34C range. Under load, I saw a 13C difference between core #0 and core #3, with Core Temp .96 giving me 62C for cores #0 and #1, and cores #2 and #3 at 51 to 56C. Using a variety of cooling solutions, I saw no real change in the temperature discrepancy, with a 11-14C difference between the first and last core. I'd like to think the difference was caused by the cooling solution used, but after switching out several coolers, the results were fairly constant, which leads me to believe that either the IHS was making poor contact with the cores, or that the temperature monitoring hardware was inaccurate. I also found it interesting that the QX9650 ran significantly warmer than the 65nm QX6850.

[[Dr. V]]

Thanks YH, but that's not really my concern. I understand the difference in cores in terms of temps.

I'm worried as to why I'm getting many different results, and which program to follow. If the motherboard is the most accurate, then I'll follow SANDRA (which gives me temp readouts I'm VERY happy with).

I'm just not sure...

[Yellowhello]

Hmm, it it were me, I'd take the highest temp just to be safe... I personally use CoreTemps for whatever reason...

[[Dr. V]]

The only thing that's perplexing me is that the BIOS temps seem to me like they would be the most accurate all the time...

And, I've had CoreTemp and the BIOS temps resemble each other extremely closely before. But other times not.

[Capper]

I'd really put little stock into the temperatures....there is no 100% accurate program. with the temps you are shopwing (39/45C) you have nothing to worry about.

Having said that, try Core Temp .96 and Everest 4.0, those two applications should be very close.

[[Dr. V]]

Well, both CoreTemp and Everest 4.20 Ultimate Edition is giving me pretty much the same readout (check original CoreTemp readout in first post). BUT, there is a section aside from "Core #1" and "Core #2" in Everest called just "CPU" and THAT'S giving me the BIOS's readout. It looks to me like, though, I need to pay attention the high 30s to low 40s on IDLE and mid low 60s to low 70s on LOAD...right?

In light of that, another question comes to mind. Why the hell are my temps so high?

[Yellowhello]

Originally Posted by ST!X

In light of that, another question comes to mind. Why the hell are my temps so high?

What are your ambient temps?

[Capper]

Honestly, it could be any number of things.....a bad IHS (not making contact with the cores, concave surface), it could be bad sensors (I've seen some really wonky temps with my QX9650, and others are having all sorts of temp issues).

[[Dr. V]]

Ambient temps where? In the room or in the case?

Motherboard temp is giving me 38*C if that helps any...

@Capper

Yea those are all possible, but what do I do now? What's my reference point for overclocking, or feeling safe to do anything that may bring my CPU to 50% load?