Sign in with
Sign up | Sign in
Your question

C2D E4300 GA965P-DS3 v3.3 F10 temps

Last response: in Overclocking
Share
April 6, 2007 3:27:52 PM

Following the guide, I have recorded the following:

Tcase = Idle = 14C & Load = 48C
Tjunction = 23C & 55C, Hottest Core

Ambient = 21C
Chipset = P965
C2D = E4300 Allendale
CPU Cooler = Noctua NH-U12F
Frequency = 3.00gHz
Load = TAT
Motherboard = Gigabyte GA965P-DS3 v3.3 F10 BIOS
Vcore = 1.3500

All temps according to Speedfan 4.32 without offsets.
Core readings are 40C idle and 70C load in TAT.

It seems that TAT is too high, Speedfan may be a bit low and using the temp2 reading in Speedfan for Tcase idle temp of 14C is lower than ambient of 21C. What am I doing wrong?
April 6, 2007 5:36:15 PM

TAT does in fact report about 13-15C too high on my DS3 rev 3.3 and E4300 also.

I have a 70% OC and S.F. reports my cores at 27-28C idle, and 48C load. Load being Orthos CPU stress.

I dont like TAT for load temp tests, Computronics swears by it, but even he will totally agree that no matter how many apps you run at once your temps will never get as high as TAT 100% will get your temps. So if TAT is going to give unrealistically high temps, then I dont see a point in using it for a benchmark.

I dont want to step on Computronix toes as he surely understands the heat issue far better then I do. But here is my point. Using TAT I have a delta of over 30. Which people will say is WAY TOO MUCH! But using Orthos, which is probably even hotter then I will ever get this CPU,but is more realistic even while gaming, My delta is maybe 20. Since my temps max out under 50C, I dont see a problem. The one thing I will say about TAT is that it may show if you have a iffy seating job on your HS. If your HS mount is straight up bad you should see it straight off.

But to keep things in perspective, originally Computronix thought TAT was correct and I had high temps. So I lapped my Ninja AND my CPU. Took a good 8C off my load temps(I am very pleased with it). I later found out that TAT was too high and that my temps are now very good. I dont want to dis Comp in any way though. the DS3 rev 3.3 board is very new and TAT does in fact report temps correctly on the majority of systems. But its something to do with this board and Allendale cores. Or maybe its just Allendale cores, or maybe just L2 stepping, but TAT sometimes uses 100C for Tj max instead of 85C. Which is the case for people with rev 3.3 and E4300's
April 6, 2007 6:04:55 PM

BTW, I have a question about your rig. How far can you push your FSB with that E4300? I can only get 370 and dont even know if its stable. Anything higher and it wont post. I have started a thread about this issue. Seems other people are saying that the DS3 rev 3.3 doesnt like anything over 370 when used with an E4300. Can you confirm or debunk this thoery?
Related resources
Can't find your answer ? Ask !
April 6, 2007 6:15:52 PM

I tried the 375 and booted into windows ok. At the time, I was only using coretemp 95 and TAT and the temps freaked me out >70C load, so I backed down to the 333. It was working and I'm trying to figure out what the real temps are so I can get there again. If the Speedfan temps are correct without offsets, then I'd be happy to crank it up again. The difference between 65C and 75C seems to be the difference between normal and too hot.
April 6, 2007 7:11:06 PM

scrapper, you are missing something. There is nothing wrong with the e4300 or the DS3. Temps are being read just fine regardless of your CPU or board.

The temperature that is read for the cores from the CPUs is the DTS. It is the difference in degrees C between the current core temp and the point where TCC (throttling) kicks in. All these programs are reagind this DTS value. Problem is that to calculate an absolute temperature, you have to know the temperature where TCC kicks in. So all these programs *read* DTS, *assume* a Tj_max, and *calculate* an *inferred* absolute temperature. TAT assume Tj_max is 100 for the e4300 (and L2 e6xxx's also?). Core Temp 0.94 assumed Tj_max was 85. Core Temp 0.95 assumes Tj_max is 100. Not sure what Speedfan is doing.

Who do you believe for an absolute temperature? Who know. But I say, who cares?! The only reading that matters, imho, is the DTS value, which can be displayed directly in Core Temp 0.95 or calculated by:

DTS = Tj_max - T_core
April 6, 2007 9:08:58 PM

But, should I be using some type of offset values for speedfan or just "go with it"? From what I've seen, some people say to add 15C to each core in speedfan and some say to shave 15C off of TAT. My biggest question is where does the 14C Tcase that ramps up to 48C fit in? Is this indicative of a bad reading or some type of concave convex problem where I should consider lapping?
April 7, 2007 12:34:16 AM

I know exactly where throttleing kicks in. I have tested it twice. And I also know that its not the board or the CPU. It is totally TAT and possibly coretemp that produces a +15C error. The Tj max value of 100C is for CPU's WITHOUT IHS or something like that, that is not normal C2D. And therefore they are reporting too high. for this particular settup. Maybe for ALL L2 stepping I dont know for sure.

Throttling for my DS3 rev 3.3 and E4300 kicks in at exactly 97C in TAT.

You expect me to believe this is correct temp. Not likely. Though my Ninja WAS hot, it wasnt that hot. Besides, if im not mistaken, 85C is like shutdown temp. So how in any way shape or form is TAT correctly reading 97C before it throttles.

Speedfan, throttling kicks in at exactly 83C. right on que.

Dont mean to sound ornery, but these programs (TAT, and from readings, coretemp latest version)seem to be giving +15C temps on this settup. From what I can decypher from my readings into this problem is that it is the allendale cores, L2 stepping in general that TAT doesnt work properly on.
April 7, 2007 1:46:14 AM

SWEEEET! Thanks, man. You've put my mind at ease. 3.4, here I come!
April 7, 2007 1:49:19 AM

mrknowitall,
Am I jumping the gun here?[/quote]
April 7, 2007 5:43:46 PM

I dont think you are jumping the gun. Computronix, the guy who wrote the temp guide, has agreed with me that in my scenario TAT was in fact reporting approx 13-15C too high. He also has agreed in another thread with someone else who has the same problem.

lwsandl, if your ambient room temp is in fact 21C and TAT is reporting 40C for idle, then -15C = 25C idle. 4 degrees above ambient. Sounds reasonable to me. Also your load temp of 70C in TAT. -15C = 55C.

So correct me if im wrong here. Is SpeedFan reporting approximately 25-28C idle and 55-60C load for core 0/1? Cause those temps sound reasonable to me and your nocture is a good cooler y/n?

You can use TAT 100% for a good loading effect. I personally use orthos for a more realistic loading factor. Orthos gives me a max load of 48C and when gaming I barely break the 40C mark.

As for what mrknowitall is saying about absolute temps, yea its all kind of arbitrary because the programs age just using a static value in thier calculations that may or may NOT be correct. We just use the tools given us and have to try to figure out which one is closest to correct. Of these 3 afore mentioned tools, I believe that SpeedFan is "more" correct for my settup. And now that I know what are normal readouts from S.F. I simply use it as a place holder to judge wether things go astray in the future. ie. suddenly S.F. says my core 0/1 temps are 68C. In that case I better check my HS.
April 7, 2007 6:29:41 PM

Not necessarily.

I just worry about you listeneing to scrapper. He obviously doesn't have as much knowledge/experience as he thinks. And there is no way Computronix would agree with what he has written here. Scrapper thinks 25-C is a reasonable Tcore when Tambiant is 21-C. That is obviously not true.

It certainly seems Tj_max is >85-C for the e4300. Is is 100? I dunno. But just because Tj_max was raised for the L2 CPUs, does that necessarily mean it is *wise* to run your CPU that hot? I dunno.
April 7, 2007 6:54:33 PM

It was sooo much easier with the 939s.
April 7, 2007 7:44:02 PM

I dont think I have more knowledge or experience than I do. I AM fairly inexperienced. And you notice that I am not dolling out broad range coverage of advice. I am and always have been talking about a very specific case. But I do not think that my E4300 should be at 97-98C before it starts to throttle.

Consider this quote from the temp guide Computronix posted:

"The maximum junction temperature is defined by an activation of the processor Intel® Thermal Monitor. The Intel Thermal Monitor’s automatic mode is used to indicate that the maximum TJ has been reached.

Desktop C2D's = 85c"

I am of course intrepreting this statement as referring to "throttling" but everywhere I read says 85C is the shutdown temp.

Now consider that itat IS...i repeat TAT ..IS using 100C for Tj Max on my settup therefore all subsequent calculations ARE +15C. So since he has the same settup and getting the SAME TAT temps as I was. I can only conclude that TAT is reporting too high on both systems.

Skool me if Im wrong here. I will listen very open mindedly to everything you say. After all, the one thing that I am, without a doubt, isa "realist". I realize that no matter how much I read in the subject I could still be wrong. So teach me. But since S.F. shows 83C when throttling kicks in I can only assume S.F. is more correct than TAT for this settup.
April 7, 2007 8:01:39 PM

a) nobody knows for sure what the tj_max is on a desktop core 2. Especially on the L2 models.

b) If you think your Tcore is 25-C at idle. What do you think your Tcase is at idle?

I agree TAT is using Tj_max ~ 100 on your system. But when you think about b) above you realize that Tj_max ~ 85 doesn't really make sense for your system.
April 7, 2007 8:11:46 PM

OK I dont know if Computronix wrote this or just pasted it but I have found it on another site where it was obviously been edited, or maybe Comp edited his, either way it says:

"TAT will typically indicate ~ 2c lower than SpeedFan. Intel's Thermal Junction Maximum spec (Tj max) of 85c is the shutdown limit. 80c is overtemp, where TAT redlines and the CPU throttles"


Since S.F. Throttles at 83C and TAT throttles af 97C I think Ill use S.F.

I still want to know what I am missing so that get the whole picture. So I will wait for mrknowitall to skool me.

EDIT: whups I see you posted while I was typing. Sorry
April 8, 2007 1:21:02 AM

just so we are all on the same page:

"The processors with a processor signature (CPUID) of 00000F30h through 00000F64h implement the IA32_TEMPERATURE_TARGET (msr 1A2h). For those processors, IA32_TEMPERATURE_TARGET[15:8] is an offset value that must be added to a BASE value which is package dependent and is specified in the appropriate data sheet for the processor. For most but not all, the BASE value is 50C.



For those processors, it is assumed that temperature measurements are obtained from the thermal diode which is connected to base board management controller that also contains a fan speed controller. If the temperature read via the diode is above the BASE + IA32_TEMPERATURE_TARGET[15:8] (offset) then the processor fan is expected to be operating at it’s maximum RPM. Any temperature below that is below base + offset may scale the fan speed down.



Note Tj is not a fixed value and the IA32_TEMPERATURE_TARGET[15:8] value can vary from part to part. Tj is also not software readable.



For the processors with a processor signature of 000006Fxh and 0001066xh the IA32_TEMPERATURE_TARGET[15:8] specifies an offset (below) the PROCHOT# assertion point. For these processors the relative temperature is read via the PECI interface by the base board management controller that also contains the fan speed controller. In the case of the PECI interface the fan speed controller knows that it must spin the fans at their max RPM if the value read from the PECI interface is less than the value from IA32_TEMPERATURE_TARGET[15:8]. Any value greater than IA32_TEMPERATURE_TARGET[15:8] read on the PECI interface indicates the processor is not operating in a hot condition and the fans can be scalled back to a lower RPM.



To support transition mainstream desktop platforms where PECI may not be supported or implemented the Conroe processor (Intel® Core™2 processor) added the BASE + OFFSET method used by the Pentium® 4 processor and temperature reading via the thermal diode. However the OFFSET value is loacated in the THERM_DIODE_OFFSET register (msr 3Fh). THERM_DIODE_OFFSET contains a valid bit and an offset value. If THERM_DIODE_OFFSET[7] is set to 1, then THERM_DIODE_OFFSET[4:0] contains a valid offset value which can be used to program the fan speed controller using the BASE + OFFSET. If THERM_DIODE_OFFSET[7] is cleared to 0 then the thermal diode method of reading the processor temperature and controlling the fans is not available.



Some steppings of the mobile Intel® Core™2 processor do indicate Tj to be approximately 85 or 100 via a single bit in the EXT_CONFIG register (msr 0EEh) but desktop, workstation and server processors do not. Nor is there a register implemented in those processors that software can read to get the Tj value for either the Pentium® 4 processor, Intel® Xeon® processors or Intel® Core™2 processors.



In all implementations the IA21_THERM_STATUS[22:16] value is relative to PROCHOT assertion and not an absolute temperature."


I think the last sentance sums up what Mr know it all was trying to say. Unless of coures Im wrong.
April 8, 2007 10:18:12 AM

Well, after screwing around yesterday, I think lapping and using X23 instead of AS5 will be happening next weekend. even the temps that I like in Speedfan seem too high with Orthos @ 3.4.
a b à CPUs
a b K Overclocking
April 8, 2007 5:06:50 PM

lwsandl, did you test as shown in the Guide?

Quote:
Testing

Vcore = Manual
C1E / EIST = Disabled
CPU Fan = Manual, 100%
Computer Case Fans = Manual 100%
Primary Test = TAT @ 100% 10 Minutes
Alternate Test = Orthos @ P9 Small FFT’s 10 Minutes


Also, how did you acquire Ambient?

Thanks,

Comp 8)
April 9, 2007 12:58:52 AM

Quote:
lwsandl, did you test as shown in the Guide?

Testing

Vcore = Manual
C1E / EIST = Disabled
CPU Fan = Manual, 100%
Computer Case Fans = Manual 100%
Primary Test = TAT @ 100% 10 Minutes
Alternate Test = Orthos @ P9 Small FFT’s 10 Minutes


Also, how did you acquire Ambient?

Thanks,

Comp 8)

Vcore = Manual = Yes 1.3500
C1E / EIST = Disabled = Yes
CPU Fan = Manual, 100% = Not sure need to check BIOS settings again
Computer Case Fans = Manual 100% = Sort of - Manual @ medium speed
Primary Test = TAT @ 100% 10 Minutes = I did 5 minutes
Alternate Test = Orthos @ P9 Small FFT’s 10 Minutes = 5 minutes here too.

To get ambient temperature, I used a digital thermometer that reads Celsius.
April 9, 2007 11:27:00 AM

Ok,
I have tested again as recommended in the guide.

Vcore = Manual = Yes 1.3500
C1E / EIST = Disabled = Yes
CPU Fan = Manual, 100% = YES
Computer Case Fans = Manual 100% = YES (all fans on high)
Primary Test = TAT @ 100% 10 Minutes = YES
Alternate Test = Orthos @ P9 Small FFT’s 10 Minutes = 5 minutes here too.

Tcase = Idle = 18C & Load = 52C Speedfan
Tjunction = 27C & 58C, Hottest Core Speedfan (45C and 71C - TAT)

Ambient = 22C
Chipset = P965
C2D = E4300 Allendale
CPU Cooler = Noctua NH-U12F
Frequency = 2.0
Load = TAT
Motherboard = Gigabyte GA965P-DS3 v3.3 F10 BIOS
Vcore = 1.3500
April 9, 2007 12:04:29 PM

its fine, all looks good to me.. only time i would worry is when TAT exceeds 75+ this is what i personally limit myself to.
April 9, 2007 12:42:07 PM

34C Tcase delta that begins below ambient?
Ultimately I want to take my 333 FSB and jack it up to 375 FSB.
Do the temps look good enough for that?
April 13, 2007 10:24:51 AM

**Bump**
!