Quadcore temps delta?

[Rocky90]

I'm using Q6700.

However, in this temperature guide the delta for Q6xxx is only about 5°C!
My delta between TJ and TC is about 10-13!

Using stock cooler, my whole thing q67@3,2 1.3 volt.

Only prime pushes the baby further than 70. Usually in heavy renders hitting max 69.

And an interpretation problem,

the hotscale says "75" does that mean UNTIL 75 it is hot? Or does it mean with 75 it begins to fall into category HOT?

The second thing is when I was using SMARTFAN ( automatic fancontrol ) on automatic, pwm+volt for regulation, funnily my CPU was at 66°C TJ and Smartfan didn't think it was time to increase the CPU fan speed..

So I think either 66°C TJ is a total normal temp for Smartfan, or the delta or my TJ temps are a bit too high.

 

[Lupiron]

Alrighty then, lets start at the top! Can you run Core Temp and list the chips VID?

The delta between TCase and TJunction is a factor to watch! When loaded it shows how well your cooling solution is operating. The higher the delta while loaded, the better the cooling.

For instance, my TRUE has a delta no lower than 16, and it hits 17 at times. That means that the TRUE is doing a good job keeping the TCase nice and cold. (Since the cores are fully internal, they will always be hotter!)

A way to tell if your cooler is on right too, watch the delta. When my delta is 15 or 14, I know I havent seated the TRUE properly. And it appears that each Delta point is about 4-5 c on core temps!

A note about prime testing: Its totally fake! As in, not real world conditions for your Chip. You'll never push that processor like primes small ffts. So the max Temp for cores is a sliding scale.

The community seems to think that if you OC till your cores hit 65/66c, thats the best. Because you can expect the 3 years still, or more. Personally, I'd hit 75 c for a max OC, because you'll prolly never hit 70c in daily activities.

I Need yer VID!

--lupi!

 

[Lupiron]

Oh, I have hit 90c + and 1.8250 VCore on a quad, with no ill effects at all. (was 1.7xx in windows.) So the chip can actually take quite a bit more than what they tell you. BUT they have to tell you lies, or you'll be turning 200$ chips into the speed of over the 1000 dollar ones.

And then they have to MAKE SURE it runs stable for years. 3 Year warranty! The easiest way to do that is to run it under what it can do, of course!

So they make all q6600s as 3.0 chips and down clock them further to make SURE they remain fine for years. (Prolly not exactly like that, but you get the idea. Make things better than you sell them as, and they prolly will work for the many years you need to warranty a new product for!)

--Lupi!

 

[Rocky90]

Thank you for your reply Lupiron.

My VID is 1.3 and I'm currently running 1.33750 in order to sustain maximum 1.3 (Vcore in Gigabyte boards is always lower than what you set it to)

So I'm running 1.3 idle and 1.26 min in full power. Stable.


Back to the temperatures.

For testing I can use Carrara 6, which is a 3d rendering program. It takes excellent usage of 4 cores and keeps them at 100%.

Therefore, rendering longer than 20mins, my 2 hotspots cores will exceed 70 only slightly with the Tcase being 59 that I consider my personal uplimit.

What I don't understand is wether the writer of the temperature guide meant that at until 75 the HOT area goes, or at 75 the HOT area begins.
Waiting for a clarification on that one.

You are absolutely right I don't believe that one's CPU has to be PRIME STABLE in order to be called "stable". You'll never get prime95 conditions, but I can push the heat really high by using 3d renderingprograms that effectively pump with 100% cpu usage.

For example, I have another 3D rendering program (Vue6) that not always goes 100% but 95% - 100% and my temperatures there will only go to 65-67 which is without a doubt safe temp.


I'm not entirely sure but I experienced a border at about 70°C TJ.

Strangley, I had my Tcase temps once over 60 (mainboard beeping at over 60) while my TJs were still just hitting 70.
What I want to say by that is that this (70°TJ) is a breaking point where you'd end up anyways and which would only exceed if you really pushed the FSB way further.

But for now I will rely on my q67@3,2 1.3V -> 72°CTJ and I think it would be a miracle if that burns up in 1 year.

I just don't bother getting a better cooling system until it is really neccessary which in this case it is NOT.

 

[Lupiron]

See the post below this one for an explanation of why your voltages are lower.

If you need any specific help, fell free to ask! You appear to be doing fine.

--Lupi!

 

[Rocky90]

Hey man got another question.

Trying out to set the vcore to default now! 1.3 and I let EIST and CE1 do the work for me.

In idle i get nice 1.12 - 1.14 when load comes about 1.25. Q6700@3,2ghz

Is this kind of drooping safe to do?

Undervoltage don't hurt does it lol.

 

[Rocky90]

Ok I really need this one question to be answered.

Because it's really strange!



I just used fft stress test in prime

my Tcase went over 60 (confirmed by mainboard beeping) while the TJ was still 69.

And usually I have Tcase 54 with TJ also at 69-70.

How can this be? I am reading TJ with HardwareMonitor and CoreTemp!

How can Tcase rise when Tjunction does not rise at the same time???

 

[Lupiron]

Yeah, that's interesting. What cooler is it?

As the cooling fails at a temp, the TCase will reflect it, because you'll lose the ability to keep the delta at a decent number. When the Tcase starts to near the core temps, that may mean you've reached your coolers dissipation limit.

Or.. try running prime on all the cores, when it reaches a stable hot zone where its not rising or lowering, press firmly downward on the top of your cooler, hold for 30 seconds and watch the temps. If it goes down by more than 1 c, tighten it!

You can also press on the sides of the cooler, to see if more contact with a certain corner makes the temps go down at all. Just in case your cooler isnt attached properly. Did you apply the paste and cooler??

--Lupi!

 

[Rocky90]

Stock cooler.


So do you think it is it not possible that when the CPU is Tcase 61°C the TJ are 70°C?

I used to go over 72 before and I actually trust my TJ readings. They're mostly correct.

I applied the cooler very tightly, I had to take out the motherboard to apply the last clicker with force and they're all 100% in now. The thermal paste provided should be alright.

 

[Lupiron]

Now keep in mind, the stock cooler is designed to keep it cool at stock only. So your OC will be minimal.

Sure its possible! That means the cooler blows! As you reach the coolers maximum dissipation, the delta between the CPU temp and the cores will lower. That means the heat is to much for the cooler, because it cant even keep the top of the CPU cool.

But for it to all of a sudden lower, thats odd. Especially if you have had all the cores at 100% and seen what the delta was before. A quick decrease in delta with no more temp increase could mean that your cooler has come lose.

Did you do the test I suggested over your reply?

--lupi!

 

[Rocky90]

Now keep in mind, the stock cooler is designed to keep it cool at stock only. So your OC will be minimal.

Sure its possible! That means the cooler blows! As you reach the coolers maximum dissipation, the delta between the CPU temp and the cores will lower. That means the heat is to much for the cooler, because it cant even keep the top of the CPU cool.

But for it to all of a sudden lower, thats odd. Especially if you have had all the cores at 100% and seen what the delta was before. A quick decrease in delta with no more temp increase could mean that your cooler has come lose.

Did you do the test I suggested over your reply?

--lupi!

Well first of all I must tell you that,

my Delta stands in extreme contradiction to the Temperature guide.

It says that the Delta is 5 +-2.

Mine is always 10+13

My TJs should be correct because my TJMax is correctly read at 100.


Can assure you that the CPU cooler sits *tight* and close to the CPU, that's why it does so well @3,2 . Prime is the only thing that really pushes the Tcase to 61. And my mainboard beeps when Tcase is 60... Strangely, when my Tcase reached 61 i pressed the reset button, and BIOS showed me that CPU was only 44...To me atleast, it seems IMPOSSIBLE that the heatsink can drop temperature so fast ( the TJ can do tho)


Oh and by the way, you're underestimating stock cooler.

At time Tcase was at about 60 the cooler was only running at 2/3 or 3/4 max speed. For some reason Smartfan doesn't want to let it go faster than necessary to keep TJ beneath 70. When I turn off smartfan there's still a mechanism keeping the fan from going as fast as it can.

There will be a point, when it hurts my CPU more than it would hurt my ears otherwise

Had it at 60°C heatsink for about 10 seconds, Hope it didn't damage..

 

[Lupiron]

::Chuckles.:: The temps you have are minimal, just so you know. When you see 100c on a core, push the reset.

Once its re set, the CPU cools in like 1 second, so it's misleading to go by shutting down and then jumping to the bios.

And Prime 95 IS the only thing that will push your system! So I am guessing that its normal for an intel stock cooler to start to max out around there.

My delta is 16 or 17, Always, or my cooler isnt on right. Like now, for instance.

So I'd say its fine. 3.2 is a great OC for a stock cooler, BTW.

What did you expect? And while the metal on the heat sink may retain temps for a small amount of time, there is a fan blowing on it for the 10 seconds to restart, and 10 more to get in bios and select power management. So yeah, it cools fast!

--Lupi!

 

[Rocky90]

::Chuckles.:: The temps you have are minimal, just so you know. When you see 100c on a core, push the reset.

Once its re set, the CPU cools in like 1 second, so it's misleading to go by shutting down and then jumping to the bios.

And Prime 95 IS the only thing that will push your system! So I am guessing that its normal for an intel stock cooler to start to max out around there.

My delta is 16 or 17, Always, or my cooler isnt on right. Like now, for instance.

So I'd say its fine. 3.2 is a great OC for a stock cooler, BTW.

What did you expect? And while the metal on the heat sink may retain temps for a small amount of time, there is a fan blowing on it for the 10 seconds to restart, and 10 more to get in bios and select power management. So yeah, it cools fast!

--Lupi!

No mate you got me wrong there lol.

See,

when I ran prime95

a) the cooler was only at 2400rpm while it's max is 2900
b) it was sbabilizing at 60-62 heatsink (tcase)
c) I am not able to use any "normal" program to push my tcase temp beyond 55.

--70--/--75--75--75--75-- Hot
--65--/--70--70--70--70-- Warm
--60--/--65--65--65--65-- Safe
--25--/--30--30--30--30-- Cool

If left is Tcase and right Tj, why is the Delta only 5 in this scale?


Also I'd like to add (because it's interesting) that Smartfan uses Tjunctions not Tcase to regulate the rpm to keep it just slightly beyond 70tj.

 

[Lupiron]

Okay, you just keep enjoying your temps! While I enjoy my 3.8 @ less than your 3.2 temps, lol!

--Lupi!

 

[Rocky90]

Dude if your temps are too low, clock higher!

And now I need some answers for those above

 

[Lupiron]

hehe, at 3.6-3.8 depending on your VID, you reach a barrier where it simply needs waaaayyy to much VCore to justify the speed gains.

Now as far as the deltas go, ask computronix, or someone who likes to work with temps and their formula of calibrating. If you are going by the guide he wrote, I believe it has you calibrate the temps so that they are more accurate then the general or wrong data that is normally given, but that means his formula isnt gonna work with those of us who dont wanna go through that process and are happy with general temps.

Anyways, they are just Guides. I have had my chip over 100c TCase in the bios before. 10 times in a row, really, because I didnt have my cooler in contact with the IHS. If 10 re boots at 100+ c cant hurt the chip, think your 75 c is gonna harm it? Sure, it will shorten its life span, but from what? Thats the question!

I'll let you all know when I manage to burn a chip out. But so far, they dont wanna die!

--lupi!

 

[Rocky90]

hehe, at 3.6-3.8 depending on your VID, you reach a barrier where it simply needs waaaayyy to much VCore to justify the speed gains.

Now as far as the deltas go, ask computronix, or someone who likes to work with temps and their formula of calibrating. If you are going by the guide he wrote, I believe it has you calibrate the temps so that they are more accurate then the general or wrong data that is normally given, but that means his formula isnt gonna work with those of us who dont wanna go through that process and are happy with general temps.

Anyways, they are just Guides. I have had my chip over 100c TCase in the bios before. 10 times in a row, really, because I didnt have my cooler in contact with the IHS. If 10 re boots at 100+ c cant hurt the chip, think your 75 c is gonna harm it? Sure, it will shorten its life span, but from what? Thats the question!

I'll let you all know when I manage to burn a chip out. But so far, they dont wanna die!

--lupi!

Yeah I know I'm so paranoid of temperature spikes. They haunt me lol.

Of course Intel set their specifications so even at 71°Tcase it would still not hurt it running 4/24 i guess .

I had used a stupid SPD so my Motherboard raised voltages for 1.47 once :S. Fortunately I didn't go for load temperatures on that one lol.

Just want to make sure my CPU runs fine for 3 years and I hope my 8800GTX keeps head up, going straight 85°C load that is normal )

 

[CompuTronix]

Rocky90, your Delta is high because your BIOS (CPU) temperature and DTS (Core) temperatures are "canned" values, and are uncalibrated. As I just explained to someone, according to Intel engineers, the maximum Tcase to Tjunction Delta is 5.2c at load using stock configurations: http://eda-publishing.imag.fr/spip/IMG/pdf/TMI23.pdf

Also, according to Intel's Processor Spec Finder - http://processorfinder.intel.com/Default.aspx - the Maximum Case temperature for your Q6700 G0 is 71c. From the Temp Guide:


Section 5: Findings

(C) ...BIOS Calibrations from motherboard manufacturers, Factory Calibrations from Intel, and popular temperature utilities are frequently inaccurate.

(D) Intel shows Maximum Case Temperature (Tcase Max) in the Processor Spec Finder, which is the only temperature that Intel supports on Core 2 desktop processors...

(E) Intel does not provide documentation for Maximum Junction Temperature (Tjunction Max) on Core 2 desktop processors...


Section 6: Scale

Safe and sustainable temperatures vary according to Spec#. The temperature Scales shown below illustrate the Delta between Idle and Load, and the 5c Delta between Tcase and Tjunction. Although the 5c Delta is relatively consistent, Tcase and Tjunction tend to converge at Idle and diverge at Load due to Variables such as Vcore and CPU cooler efficiency. Low Vcore and clock may cause Tcase to Tjunction Delta to indicate as low as 3c at Idle, while high Vcore and overclock may cause the Delta to reach 7c at Load.

...as Tcase Max will be exceeded before Tjunction Max is reached, Tcase Max is always the limiting thermal specification.


Does that clear it up? Have you tried running the calibrations in the Guide? Have you tried Real Temp? Are you aware of what it does differently than the other popular temperature utilities, and the sensor linearity problems it solves? Have you read the thread? http://www.xtremesystems.org/forums/showthread.php?t=179044


Comp

 

[Rocky90]

Thanks I will read that PDF tomorrow.

will try realtemp now.

 

[CompuTronix]

CPU temperature AND Core temperatures both require calibration. On certain motherboards, I've noted several consecutive BIOS releases which each displayed different CPU temperatures; some that were well above Tjunction as well as some that were well below Ambient. Additionally, Intel's factory DTS calibrations aren't always accurate, and sometimes aren't even close. I've also noted Cores which were more than 30c above their neighbors and didn't throttle, as well as Cores which were nearly 10c below Ambient.

Since BIOS (CPU) and DTS (Cores) are "canned" values, calibrations are frequently the source of inaccurate temperatures, This is precisely why I feature SpeedFan 4.34 in the Guide, because it can provide calibrated CPU temperature AND Core temperatures in 1c increments. The solution for your rig is to follow the calibration procedure in the Guide. When the calibrations are completed, your CPU temperature will average 5c below your Core temperatures at load, and your Core temperatures will typically be a few degrees below Core Temp 0.98.1, and a few degree above Real Temp 2.5.

 

[Rocky90]

Ok realtemp looks nice.

I can already see differences.

Hardware Monitor / CoreTemp

66
60
62
66

Realtemp

61
55
57
61

That comes A LOT closer to my Tcase.

What I don't understand is why Realtemp says that my Tjmax is 95, not 100...

Knowing that my temps are even cooler would please me lol.

I will calibrate them correctly tomorrow.

 

[CompuTronix]

In order to understand what Real Temp does differently, you have to read the 1st post on the thread, which explains the basics. Also, check the "ReadMe.txt" file for links. I've read the entire 40 page thread, and believe me, this is the most revolutionary piece of temperature monitoring software currently available.