E4300, DS3 OC temps question

[billyc]

New to this forum, completed my first build and first oc, specs in my sig.

I'm running in an ambient of around 29c, I have 2 Seagate 320gig sata2 drives, and a 7900gs gpu oc'd. I'm also using the antec freezer 7 pro cpu cooler. I have DDR2 voltage at +.1 (1.9v) and vcore is 1 notch belos stock (1.31875), all other volts are norm. Now for my temps.

Idle

Easytune 30c
CoreTemp 30c
SpeedFan 30c
TAT 48c

Orthos Load

Easytune 55c
CoreTemp 51c
SpeedFan 55c
TAT 63c

TAT Load

Easytune 59c
CoreTemp 58c
SpeedFan 59c
TAT 70

NOTE: speedfan does not handle E4300 so good at this time. I have to use 4.31 to control fan rpm, 4.32 has bugs for this chip.

All seem fine but the TAT Tjunction temp on the TAT load test. I know that this is probably one of the hottest test for the CPU, so I'm not worried, it just seems high compared to the rest of the temps. I think it probably indicates I could have a HSF seating problem, but is it really worth the trouble to address it considering the rest of the load temps?

I relish any comments or suggestions anyone might have. I'm oc'd as far as I plan, at least in the near future.

 

[RonAKA]

You have a mild OC and I suspect nothing to worry about. I only have a stock HSF and am not worried at all. There are frequent posts here with a common theme - TAT reports a much higher temperature. It begs the question as to why? Does Intel have access to some other measurement on the CPU than anyone else? Or, do they choose to use the same input and report a different temperature? We are not talking instrument calibration here, as this is a digital exercise. Could it be they are just covering their butt, and showing a higher temp than real so they can avoid warranty claims? No idea, but from what I have seen TAT is the only measure that reports these high temps.

What are you getting with the individual core temps with SpeedFan 4.32? I find them much lower than the CPU temp, and thought they were in error. However if you look at a graph from loaded to unloaded they seem to converge on CPU temp and then read much lower unloaded. I suspect they are coming from a different part of the CPU which is located a different distance from the heat sink (with CPU pickup further away).

 

[billyc]

My last 10 min TAT test w/speedfan 4.32 showed cpu 55c, core 0 53c and core 1 54c. The thing is, the core temps are supposed to be Tjunction temps ~+15c from cpu, Tcase, temp. SF 4.32 has as number of problems w/ E4300. The core temps are wrong, my +12v is reported as 1.34v, vcore is wrong and the fan rpm is either 0 or 135000. This doesn't bode well for relying on this monitor, although the CPU temps and system temp does match Easytune temps (as reported in bios). The above problems have been accepted by speedfan and seem to be related to chip IT8712F.

Speedfan 4.31 does not support c2d cpu's per se, but I can control my fan and the cpu and system temp seem to be ok. Speedfan iven has a modifier for the temps if they are "off". I also use coretemp and that monitor reports the 2 cores ~2C cooler than easyfan and speedfan.

TAT is measuring the Tjunction temp which has the sensor on each core and is the hottest temp measured, as I said above, ~+15c from Tcase which has the sensor between the cores. TAT's torture test also generates the highest temps you will ever hit.

Having said all that, I am new at all this, and the above information comes from forums, guides, product sites and any other hits I could find on the subject, with the exception of the speedfan problems which I verified myself on the problem tracking sytem of speedfan site. Different folks swear by different temp monitoring tools.

What are you getting with the individual core temps with SpeedFan 4.32? I find them much lower than the CPU temp, and thought they were in error.

I found one reported problem where the cpu and core temps were reversed in speedfan. Maybe that's what your getting. What is the temp difference between the two?

I'll probably go with what I've got. I won't have any problems unless I decide to push the oc a little more and I'll address this issue further at that point.

Thanks for your response. Sorry about venting all my frustrations to you.

Bill

 

[RonAKA]

I ran the tests in the free version of 3DMark06 up to the end of the second CPU test. Following are the temps I got with SpeedFan 4.32, with first number the steady state idle, and the second the peak recorded during the 2 CPU tests:

System 32/34
CPU 33/55
Core 0 22/51
Core 1 21/51

So don't know whether they are right or not, but the two core temps don't always track together which would suggest one is being used more than the other at times. Don't see how SpeedFan could get them reversed. All the temps respond very fast, except for the system which pretty much flat lines. These temps suggest to me that the core pickups are closer to the heat sink than the CPU sensor.

It also reports Vcore at 1.41 which pretty much agrees with other measures, and the 12 volts at 12.51.

 

[billyc]

Those idle core temps look suspicious. What is your ambient? You can't have temps lower than your ambient which those core temps show around 70F. Have you used any other tool to compare? CoreTemp or Eaytune5?

My system temp is flat also, but around 40c. My idle is around 28c which is around ambient +1c.

At this point, I still don't trust speedfan as a reliable tool for the E4300. I'll keep watching for the fixes to come out.

Curious, do you control your fan (or fans) with speedfan?

 

[RonAKA]

Ambient would be about 20C (outside the case). Have not used those other tools for temp. Asus probe agrees with SpeedFan for CPU temp, but does not give core temps. I tried TAT and it does report much higher core temps, but no CPU temp that I can figure out how to display. Comparing SpeedFan to TAT, TAT reports core temps about 18C higher than SpeedFan core temps. TAT core temps are about 8C higher than than the CPU temp reported by SpeedFan and Asus. Looks like SpeedFan is not correct for core temps. I see that you are supposed to be able to calibrate it, as described in this link:

http://forumz.tomshardware.com/hardware...o-Temperature-Guide-ftopict221745.html

I don't use SpeedFan or the Asus utility to control CPU fan speed. It just runs at max - 2170 RPM, and I don't hear it, so I'm happy with that. Tried the Asus utility but found it runs too slow and lets the CPU run hotter than I like.

Using a Sonta II case with the stock 120 mm exhaust fan running on low, so I don't hear it.

 

[Binky]

You guys really need to ask yourself if it's possible to idle at ambient +1C or +2C. The answer is no.

 

[RonAKA]

Agree, I'm now convinced the SpeedFan core temps are incorrect, and they are the only temps close to ambient on my setup.

 

[billyc]

Yea, can't argue that point. It seems we only have so many thermometers to use. Maybe the best monitor would be the mb software, it's at least as right as the bios.

 

[billyc]

Here's some info that should be telling me something, but I'm not sure what. Using SpeedFan and TAT, the temp difference is 18c between cpu and tat. Running Tat torture test, the difference is 13c and running Orthos, the difference is 9c. Is there some revelation here? or is this further doc that speedfan is wacky w/E4300?

 

[CMC79]

I have a 4300 and I have had the same disparities in measurement--in TAT, I idled at around 49-51 degrees, but measured 34 degrees or so in Core Temp and Speedfan while using stock cooling at 2.4ghz. Load temps went way up when using TAT's torture test or Orthos. I put a Zalman 9500 on it and run now at 3.0ghz, but my temps are about 10 degrees cooler, but I still have the same discrepancy between programs. I figured that if TAT was to believed, running up way into the 70's under load wasn't too good for it if true, and 2.4ghz, while nothing to sneeze at, is far less than what I had hoped for overclocking.

 

[billyc]

Just found this reference in another forum. Perhaps there is more to the TAT question yet to be revealed.

Check the following forum

Text

 

[idiotekniQues]

yes intel does have access to some special sensors that TAT can see.only the latest version of core temp and for awhile only the beta release of speedfan had the ability to read these far more precise digital sensors on each core.

TAT should be the most accurate - however, i do know there are some random issues with TAT with certain chips of the e4300 family. for most core 2 duos the latest coretemp, speedfan & TAT should all be pretty much the same reading. you have to read up more on the ones that give errors.

also, TAT provides a true 100% load on each core vs. orthos small fft's at priority 9.

 

[AstroGuardian]

I will do measurements today and will reply tomorrow..

 

[billyc]

The newest Core Temp 0.95 now shows Tjunction (I think this is really called Tjmax) as 100c instead of 85c and now reports Tcore temps and not Tcase, also displays temps in tray which is nice. These temps now match Speedfan, which in my case, the core temps are modified by +15c. My rig shows TAT temps at idle +2 or 3c higher than these other 2.

There is some heated discussions on these temps in the intel forums:

Re: IA32_TEMPERATURE_TARGET MSR not documented

and although Intel tried to close the thread, it was brought up again here

Heated reply

The Tjunction (or TjMAX, I'm not sure which is the right teminology anymore) has been documented as 85c by Intel. CoreTemp now reports it to be 100c and coretemp and speedfan agree on the core temps reported. TAT is generally accepted as the most correct temp and is 2-3c higher.

I don't know if my logic is correct, but if the true Tjunction is 85c, I would expect TAT to be lower than CoreTemp, not higher. Am I making sense? Not sure whee I'm going with this, but could it mean the new spec for the C2D is Tmax is 100c and not 85c?