First off, I have read and followed the temperature guide here, but still have some confusion on my temperatures.
I have an E2160 (L2) cooled by an Arctic Cooling Freezer 7 on an Asus P5k-e/WIFI board. All being housed in an Antec Sonata 3 with 2 120mm fans (one intake, one exhaust).
I followed the temp guide here. Ambient was 22C. I had everything on stock/auto, let the computer run for 10 minutes or so with no programs on ( under 1% CPU load during this time), and calibrated speed fan 4.33 for 2C above ambient. This 2C was from an E2xxx series chip (1 point) and a AF7 cooler (3 points) added together and divided by 2. This added to the ambient should be the Tcase temp at stock idle.
I then, still at stock/auto everything ran prime 95 for 10 minutes. I confirmed both cores running at 100%. I then offset speedfan so that both cores were Tcase + 5C. At idle this was roughly 40C for Tcase, and thus 45C for Tjunction for each core.
I then did my overclocking taking the chip to 3gHz (333 FSB x 9x) and had to up the vCore to get it there (1.35). Under prime 95 small sff test, it showed right at 52-57/57, which is right between "safe" and "warm" for a "scale 5" chip like the E2160 with L2 stepping.
Since doing all of that calibration, I have tried "CPUID Hardware monitor", "Coretemp 0.97", and most recently "speedfan 4.34 beta 51." Each has the same results, and show roughly 10C higher than the "calibrated" speedfan 4.33. If these are indeed accurate, I need to lower my overclock immediately. Which is accurate, and which should I trust? The "calibrated software" or the other 3 programs which all say the same thing?
Trust the guide. That's why it was made. Dont worry so much about temps. IF you Know that you have a solid connection to the mobo and are only using 1.35v I wouldnt worry about temps at all. Anything below 70c load is excelent on air.
I am trusting the guide, to a T. That's why I'm worried that temp readings on 3 different programs show that my chip is actually above the "hot scale", and reading 10C above the speedfan program that I spent 30 minutes calibrating.....
Does anyone else have these kinds of disparities in temperature readings from different programs?
I don't bother with speedfan since it's known to be wrong and needs callibration. If you are worried about temps then lower the overclock. I don't consider $70 to be a great deal of money though so burning up a 2160 wouldn't really bother me.
However, cooling does seem lacking as my far more extreme air cooling lets me run a Q6600 at 3.6ghz with 51c core temps under full load.
My thoughts is that it's your case causing the high temps. The antec sonata is a "quiet" case meaning slow and low noise case fans. There could also be poor contact between your CPU/heatsink and a less than ideal TIM.
Contact between CPU and heatsink seems fine (no play at all). I haven't lapped either surface, but with AS5, I really didn't see a need to until seeing these high temps. I only got this cooler because it was so inexpensive (same idea with the chip itself). I was out less than $100 for the pair, and I can take it to 3.0 gHz and have it perform on par with chips 2-3x as expensive.
As for the case, ehh, it was cheap and came with a very nice power supply which I needed anyway. I thought it would cool just as well as my previous Antec LanBoy case (both have 2 120mm fans). Would going back to the LanBoy help at all?
p3matty... as far as I can understand you are assuming that 57c is your load temps. That's fine! Everyone on air get's about that on average. I get close to 59c with my E8400 overclocked to 4.05ghz with a TRUE (which is the best IMO) and Antec 900 with 4 120mm and 1 200mm.
Dont worry so much! Your fine. And besides... Intel says these chips are fine below 70c (which you are below) and even if you did damage it... IT'S 70 bucks!
If the actual temps are 57C, then I'm fine and not worried at all. This is only true if I'm to believe the "calibrated" speed fan 4.33.
If, though, I should rely on any of the other 3 programs I mentioned, which all show the same thing, my temps are 10C more, and I should really lower something if I don't want to replace my chip..... Are Coreetemp, Speedfan 4.34 and HW Monitor all showing inaccurate temps?
Hey, P3matty, I know this has little to do with your thread, but next time you are on stock speeds and voltages could you disable speedstep in the bios and then DL and run core temp version 0.96.1 and post your VID, please?
When you run core temp, watch the VID area and run prime or your testing software and see if the VID changes. If it doesnt, will you please post that here?
Thanks, just for some research I am doing! Hope they all are not 1.3250!
Does it have to be that version? .97 tells me 1.325 (still currently OC'ed though). I will do it again will everything stock, but I wanted to know if I really needed to get the older version or not (as I can't find the older version).
I'm actually leaning toward relying on the other three programs. I wonder what your temps are with everything in auto or stock. Set to stock if you want and post temps. Idle temps should be in the 30's and load in the 50's. If you are experiencing anything higher than that maybe you have bad sensor on the chip itself.
Idle temps in the 30's at stock/auto?!?! Did I read the guide wrong? Doesn't it say that my temps at idle (with an E2xxx chip with ACF7) should be ambient + 2C? (X + Y / 2 = Z so 1+3 / 2 = 2) Ambient was 22C, so idle should be 24C, right?
and I quote....
(D) Tcase should indicate Ambient + Z, which is the Delta based on Idle power dissipation and CPU cooler efficiency. Check the links in Section 14 to determine CPU cooler efficiency. Since there are multiple processor and cooler combinations, use the following table to choose the Delta which most closely matches the hardware configuration:
NO way! I dont buy that at everything auto CPU temps will be +2c ambient.
I think that guide might be screwing you up. You are using very reliable programs that have been tried and true. I would go by the programs (not so much speedfan) and try to stay under 60C at full load.
Also I really dont pay to much attention to my temps. I'm not overclocking to where I really have to worry about it IMO. If you are running it at 3ghz with nice voltages (like below 1.4v) and KNOW that you have a good mount (feel warmth on the HSF knowing you have a solid mount) than you really shouldnt be worrying about temps. Especially if you arent experiencing system crashes or istability.
By the way. Mine idles in the upper 20's like 28-30C at everything stock. I have superior air cooling too.
I won't be able to stop worrying and enjoy my PC if I think it's overheating. I'd like to keep the temps in the "safe to warm" scale, but I'm just not sure what to trust. Everyone says to go by the guide, and you're telling me the guide is wrong and to listen to the other programs. I'm just not sure what to do.....
Other way around. Following the guide I'm at 57C, but Coretemp, a newer version of speedfan and hw hardware all show about 10C higher.
If I'm to believe the guide, which many people say, then I'm fine and I'll leave it as is. If I should follow the other programs, which many say are accurate and don't need calibration, I need to lower things ASAP.
You are probably right. Maybe you are running it kind of hot. I would if I were you LEAVE it but if you want this sucker to last awhile maybe lower voltage, check HSF connection (make sure good heat transfer), or just lower overclock.