Where do you draw the line for temps?

[imported_wired247]

Originally posted by: Idontcare

Originally posted by: wired247

Originally posted by: CatchPhrase

Originally posted by: wired247
I draw the line where the manufacturer tells me to do so.

It is usually available on the website or in the documentation.


People are overly paranoid about temps, with the mentality that "what is too hot for me is too hot for my processor!"


this is a silly notion but a good number of people will assume that 25C must be superior to 50C in terms of performance

I deeply apologize that I care about my components getting damaged.

The point is, the manufacturer's knowledge about appropriate temps > your knowledge.

People think "50C feels hot to the touch. It must be bad for a processor" "80C is really cooking."

When these very same temps are 110% tested and appropriate for various CPU/GPUs .


What is hot for you != what is hot for your processor

I seriously doubt Intel knows what is best for my Western Digital hardrives.

Are you going to grow up anytime soon or do we get to continue to enjoy your perception that your opinion > anyone else's? You continue to press this angle of attack that your opinion is superior to all others by way of assuming everyone else's ability to have an opinion is subject to your assessment and approval. Totally FTL dude.

Please stop trying to start a flame war with me.

If you don't like what I have to say, you are free to ignore me, but your posts are seriously bordering on harassment.

Also, the topic at hand was about CPU/GPU, I don't know why people put words in my mouth when I never mentioned "your" WD HDs. (Straw man argument...)

Now let me make something clear about temperature versus heat.

If you do not change any variables, like assume a non-OC'ed processor under 100% load measuring a temperature of 50C:

regardless of what temperature your processor core reads, it is putting out the same amount of heat (all other variables remaining constant).

All of the heat is coming into your case. The only difference between high temperatures and low temperatures (with all else remaining constant) is the efficiency of your heatsink/fan combo.

So, it matters very little to your hard drives whether your CPU reads "32C" or "50C" if your processor is in fact putting out the same amount of heat.


Now, I think lower temperatures are fantastic. It gives your CPU/GPU more overhead in case dust, ambient temperatures, or failing fan speeds becomes an issue. It also may extend the life of your processor.

However, when judging whether a CPU/GPU is "too hot" or "well within limits" what reference *should* one use, if not the manufacturer's?

 

[soccerballtux]

Originally posted by: Binky

Originally posted by: Idontcare

I seriously doubt Intel knows what is best for my Western Digital hardrives.

Are you going to grow up anytime soon or do we get to continue to enjoy your perception that your opinion > anyone else's? You continue to press this angle of attack that your opinion is superior to all others by way of assuming everyone else's ability to have an opinion is subject to your assessment and approval. Totally FTL dude.

Keep it civil. I tend to agree with wired247 and I don't see anything inflamatory in his/her post. Many people worry too much about temps. Your post seems a bit angry.

Nothing was said about HD temps. If you want info there, read the Google study on HD temps and HD lifespans.

I don't know, I kinda got the same feeling as Idontknow at the line "the manufacturer's knowledge is > your knowledge", and took it kinda personally (this could be my fault I guess) because this temps thing is not a rocket science, as being an EE myself I have about as much understanding as the manufacturer does about these things. (For instance if the gradient between the deepest part of the transistors in your processor, the channels lets say, is only 20C higher than the temperature of the top of the processor that you're paying attention to, then you could run it theoretically at 100C and be fine; or hotter (the temperature that silicon channels begin breaking evades my memory at the moment)).

To be honest, in a day to day conversation if someone told me "Well you don't know as much as the manufacturer" I would think they were being asinine and proceed to not spend any more time with them.

 

[imported_wired247]

I apologize for sounding abrasive, it was intended to be in a very lighthearted tone.

(Certain language is well accepted on certain forums and not on others. On the other hand, I have pretty thick skin and sometimes expect incorrectly that others do too.)

 

[tigersty1e]

My card fan is supposed to hit 31% at 70C...59% at 94C....68% at 97C and then finally hits 100% when the core reaches 100C. I recken the core is perfectly fine at 90C... I don't like seeing it that high, so I've changed the profiles, but 80C is perfectly fine for a GPU.


For the processor, I just see how high the temps hit on a stock heatsink with no OC and put that is a soft limit....

 

[Andvari]

Idontcare, just to clear up a small part of this, wired247 never "bolded" his statement about people being paranoid. Catchphrase bolded it when he quoted wired.

Aside from that, let's all play nice now. =)

 

[LOUISSSSS]

i draw the line @ 75C for my Q6600 and @ 110C for my 8800gt

 

[zeroburrito]

my 6800 started to throttle at 130c. most the time it stayed around 120c but cod4 kept getting it to 130c. was like that for a year or so. cleaned it out and it runs in the 70's now. finally upgraded my whole comp so i don't have to worry about that anymore. no idea on my c2d as it only goes to like 51-52 load.

 

[Gaurav Duggal]

my pentium D (not oc'd) gives above 80C in summers but my 7300GT qiuts at about 62-65c only.

 

[JAG87]

You dont draw any lines, the manufacturer draws lines. You didn't engineer the chip so you have no idea whats safe and whats not.

Different parts have different thermal specifications:

Q6600 B3 - 62.2C
Q6600 G0 - 71C
QX6850 G0 - 64.5C
QX9650 C0 - 64.5C

E6600 B2 - 60.1C
E6550 G0 - 72C
E8400 C0 - 72.4C

Now keep in mind these are the temperatures that is recommended NOT to exceed, doesn't mean thats where the CPU starts throttling. As far as GPUs are concerned, nvidia cards will throttle at 127C. They don't have a safe operating temperature specification like CPUs, but obviously the lower the better, not so much for the life of the GPU, but more for the life of the surrounding parts, like the memory ICs. When you see artifacts 9/10 chances its not GPU but the memory causing it.

 

[ArchAngel777]

Originally posted by: JAG87
You dont draw any lines, the manufacturer draws lines. You didn't engineer the chip so you have no idea whats safe and whats not.

Different parts have different thermal specifications:

Q6600 B3 - 62.2C
Q6600 G0 - 71C
QX6850 G0 - 64.5C
QX9650 C0 - 64.5C

E6600 B2 - 60.1C
E6550 G0 - 72C
E8400 C0 - 72.4C

Now keep in mind these are the temperatures that is recommended NOT to exceed, doesn't mean thats where the CPU starts throttling. As far as GPUs are concerned, nvidia cards will throttle at 127C. They don't have a safe operating temperature specification like CPUs, but obviously the lower the better, not so much for the life of the GPU, but more for the life of the surrounding parts, like the memory ICs. When you see artifacts 9/10 chances its not GPU but the memory causing it.

Those temperatures are taken at Tcase, not Tjunction. There is no way to measure the true Tcase except if you can machine a slit in the IHS and stick a probe in there to measure. Tjunction always runs hotter than Tcase.

Tcase max specification by Intel is debated at this point. An alternate view: It is said that the specification means that the chips they tested did not exceed that temperature at Tcase while running at their default voltage and clock speed under their controlled environment. The prevailing view: It is said that this temperature should never be exceeded.

I am of the opinion that 1) The specification itself is very conservative and 2) That the 'Max Tcase' is ambigous. It can be interpretated as the relationship of Tcase to TDP. Proof would be in Intel's own specification guide.

http://download.intel.com/desi.../datashts/31327807.pdf

Page 80

1. Thermal Design Power (TDP) should be used for processor thermal solution design targets. The TDP is not the maximum power that the processor can dissipate.
2. This table shows the maximum TDP for a given frequency range. Individual processors may have a lower TDP. Therefore, the maximum Tcase will vary depending on the TDP of the indivudal processor. Refer to thermal profile figure and associated table for the allowed combinations of power and TC.

To me, this is proof that Tcase is related to TDP. The higher the output, the higher your Tcase will be. To me, this seems obvious. But you have 95% of people out there saying that the max Tcase is not to be exceeded, which is not true at all. It simply was the Tcase that Intel measured when they tested a CPU at a certain TDP. The lower the TDP, the lower the Tcase because the CPU itself is generating less HEAT.

Any way you slice it, the fact is, individual CPU's vary (according to Intel!) and thus, even their specification guide is just a 'guide', nothing more. That totally explains why some B3's run cooler than some G0's. But generally, a G0 will run cooler than a B3. Even so, the TDP determines the Tcase.

I believe someone at HardOCP also went over this to a great extent. He called Intel to confirm that the prevailing view is incorrect on regards to their specifications. If I have time, I will try and dig that post up.

BUT, before you come in here argueing, do your research and read the ENTIRE Intel Specification guide, don't just read a website and report here. I have spent hours researching the Intel Thermal PDF's which would be the only reliable source (besides Intel engineers) for determining what is deemed 'to hot' for an Intel CPU.

One more comment: Intel would be completely retarded to set their supposed Tcase max below that which their Tjunction respectively throttles. In other words, if Tcase to Tjunction is a 10c delta, then why in the hell would Intel throttle their B3's at 82c (~72c Tcase) when Tcase has already been exceeded? To me, this is clear - because Tcase Max does not mean "Maximum Tcase for a given CPU". Rather, it means "Maximum Tcase that was reported in an Intel lab with a TDP of XXX"

 

[JAG87]

ArchAngel, obviously you missed most of the things I said.


First of all, I said the thermal spec proposed by intel is NOT the maximum Tcase, but simply a recommended temperature not to exceed.
Somehow you made this argument that B3s throttle at 82c which means Tcase is at 72c (assuming 10c delta) which is higher than the proposed safe Tcase, and that is a contradiction. Isn't that just a reiteration of what I said? Throttle =|= safe Tcase


Second of all, obviously the higher the TDP the higher the Tcase, but Intel doesn't expect you to have higher TDP (you aren't supposed to overclock), so they give you a safe temperature based on their stock TDP.
The whole trick is to stay within that safe temperature at a higher TDP, thats why we buy after market heatsinks duh?


And last but not least, I dont know what you are smoking with the probes and all, but its pretty easy to measure Tcase.
In fact Intel placed a sensor right under the center of the IHS, and that sensor can be read by many programs including Everest, Speedfan, ntune, and the list goes on.
So its pretty easy to check if you are going over the safe Tcase proposed by intel.


I still have no idea why people continuously base their readings from core temperatures when we have no idea what is a safe core temp.
All we know is they throttle at x degrees, but we dont know whats safe since throttling temperatures must be far past safe, while on the other hand we do know what a safe Tcase is.
IMHO core temps should be checked just to make sure there is a reasonable delta between Tcase and Tjunction, and nothing more.


Make more sense now?

 

[ArchAngel777]

Originally posted by: JAG87
ArchAngel, obviously you missed most of the things I said.


First of all, I said the thermal spec proposed by intel is NOT the maximum Tcase, but simply a recommended temperature not to exceed.
Somehow you made this argument that B3s throttle at 82c which means Tcase is at 72c (assuming 10c delta) which is higher than the proposed safe Tcase, and that is a contradiction. Isn't that just a reiteration of what I said? Throttle =|= safe Tcase


Second of all, obviously the higher the TDP the higher the Tcase, but Intel doesn't expect you to have higher TDP (you aren't supposed to overclock), so they give you a safe temperature based on their stock TDP.
The whole trick is to stay within that safe temperature at a higher TDP, thats why we buy after market heatsinks duh?


And last but not least, I dont know what you are smoking with the probes and all, but its pretty easy to measure Tcase.
In fact Intel placed a sensor right under the center of the IHS, and that sensor can be read by many programs including Everest, Speedfan, ntune, and the list goes on.
So its pretty easy to check if you are going over the safe Tcase proposed by intel.


I still have no idea why people continuously base their readings from core temperatures when we have no idea what is a safe core temp.
All we know is they throttle at x degrees, but we dont know whats safe since throttling temperatures must be far past safe, while on the other hand we do know what a safe Tcase is.
IMHO core temps should be checked just to make sure there is a reasonable delta between Tcase and Tjunction, and nothing more.


Make more sense now?

Incorrect. That is not the true 'Tcase'. Read the OEM System builders Thermal Specification guide by Intel. You will see that they machine a slit on the TOPSIDE of the IHS in the GEMOTRIC center. That will give you all the correction you need. Otherwise, we can simply use the defination they give of the Tcase.

From Intel's Website:

The thermal specification shown is the maximum case temperature at the maximum Thermal Design Power (TDP) value for that processor. It is measured at the geometric center on the topside of the processor integrated heat spreader.

In case you missed it. It is clear you came in here running your mouth before reading any of their thermal guides. The only reason I decided to correct you is to stop the spread of misinformation. You, like many others, just run your mouth and talk down to people. I find that your attitude at AT sucks and would have banned you a long time ago if I had the powers.

For those wonder what he is talking about is the DIODE on the BOTTOMSIDE (actually, in between the cores) of the IHS, not the TOPSIDE. They are two different measurements. The Tcase that intel reports on their specification page is gathered by slitting a small line into the IHS and then placing in a very, very thin probe. This is used to measure the Tcase that Intel specifies. The diode underneath the CPU is NOT the same thing but it called Tcase in some cases but would be similar in temperature, but not the same.

Edit **

I found another post regarding a lot of this information. It should set the record straight.

OCFORUMS POSTER...

Edit 2 **

And Jag, if there is going to be any further correspance with you, then you will need to drop your attitude and provide proof of your claims. I can't say I dissagree with your first two paragraphs, but the attitude needs to be dropped. I also agree with your conclusion.

 

[ArchAngel777]

Originally posted by: JAG87
ArchAngel, obviously you missed most of the things I said.

Before I posted, I took you for your word on this. So, I decided to read your post I replied to and thought "Wow, I already read this, what did I miss?" So I searched the entire thread and don't see anywhere that you said any of those things... In fact, the post I quoted was the only thing you have mentioned in this thread... Care to clarify? Or... Did you just 'think' you mentioned those things?

 

[JAG87]

Originally posted by: ArchAngel777

Originally posted by: JAG87
ArchAngel, obviously you missed most of the things I said.


First of all, I said the thermal spec proposed by intel is NOT the maximum Tcase, but simply a recommended temperature not to exceed.
Somehow you made this argument that B3s throttle at 82c which means Tcase is at 72c (assuming 10c delta) which is higher than the proposed safe Tcase, and that is a contradiction. Isn't that just a reiteration of what I said? Throttle =|= safe Tcase


Second of all, obviously the higher the TDP the higher the Tcase, but Intel doesn't expect you to have higher TDP (you aren't supposed to overclock), so they give you a safe temperature based on their stock TDP.
The whole trick is to stay within that safe temperature at a higher TDP, thats why we buy after market heatsinks duh?


And last but not least, I dont know what you are smoking with the probes and all, but its pretty easy to measure Tcase.
In fact Intel placed a sensor right under the center of the IHS, and that sensor can be read by many programs including Everest, Speedfan, ntune, and the list goes on.
So its pretty easy to check if you are going over the safe Tcase proposed by intel.


I still have no idea why people continuously base their readings from core temperatures when we have no idea what is a safe core temp.
All we know is they throttle at x degrees, but we dont know whats safe since throttling temperatures must be far past safe, while on the other hand we do know what a safe Tcase is.
IMHO core temps should be checked just to make sure there is a reasonable delta between Tcase and Tjunction, and nothing more.


Make more sense now?

Incorrect. That is not the true 'Tcase'. Read the OEM System builders Thermal Specification guide by Intel. You will see that they machine a slit on the TOPSIDE of the IHS in the GEMOTRIC center. That will give you all the correction you need. Otherwise, we can simply use the defination they give of the Tcase.

From Intel's Website:

The thermal specification shown is the maximum case temperature at the maximum Thermal Design Power (TDP) value for that processor. It is measured at the geometric center on the topside of the processor integrated heat spreader.

In case you missed it. It is clear you came in here running your mouth before reading any of their thermal guides. The only reason I decided to correct you is to stop the spread of misinformation. You, like many others, just run your mouth and talk down to people. I find that your attitude at AT sucks and would have banned you a long time ago if I had the powers.

For those wonder what he is talking about is the DIODE on the BOTTOMSIDE (actually, in between the cores) of the IHS, not the TOPSIDE. They are two different measurements. The Tcase that intel reports on their specification page is gathered by slitting a small line into the IHS and then placing in a very, very thin probe. This is used to measure the Tcase that Intel specifies. The diode underneath the CPU is NOT the same thing but it called Tcase in some cases but would be similar in temperature, but not the same.

Edit **

I found another post regarding a lot of this information. It should set the record straight.

OCFORUMS POSTER...

Edit 2 **

And Jag, if there is going to be any further correspance with you, then you will need to drop your attitude and provide proof of your claims. I can't say I dissagree with your first two paragraphs, but the attitude needs to be dropped. I also agree with your conclusion.

roger that


but lets be objective ArchAngel, Tcase measured on the bottom side of the IHS would only provide a larger margin for safety. IMHO it would only differ by a few degrees from a sensor placed on the topside, but even if it does differ, its always going to be higher, not lower.

so you cant really screw yourself by observing that reading, and its the closest thing you have to the real Tcase. the core temps are far higher, and the socket temperature reported in the BIOS is completely off the wall. most of the time it reads a lot higher simply for the smart fan options to kick in sooner rather than later.

do you agree?

Originally posted by: ArchAngel777

Originally posted by: JAG87
ArchAngel, obviously you missed most of the things I said.

Before I posted, I took you for your word on this. So, I decided to read your post I replied to and thought "Wow, I already read this, what did I miss?" So I searched the entire thread and don't see anywhere that you said any of those things... In fact, the post I quoted was the only thing you have mentioned in this thread... Care to clarify? Or... Did you just 'think' you mentioned those things?

no, I was referring to the fact that I said the Thermal Spec from intel is the recommended temperature, not the absolute max. I had already mentioned that, but you repeated it as if I didn't.

 

[ArchAngel777]

Originally posted by: JAG87

Originally posted by: ArchAngel777

Originally posted by: JAG87
ArchAngel, obviously you missed most of the things I said.


First of all, I said the thermal spec proposed by intel is NOT the maximum Tcase, but simply a recommended temperature not to exceed.
Somehow you made this argument that B3s throttle at 82c which means Tcase is at 72c (assuming 10c delta) which is higher than the proposed safe Tcase, and that is a contradiction. Isn't that just a reiteration of what I said? Throttle =|= safe Tcase


Second of all, obviously the higher the TDP the higher the Tcase, but Intel doesn't expect you to have higher TDP (you aren't supposed to overclock), so they give you a safe temperature based on their stock TDP.
The whole trick is to stay within that safe temperature at a higher TDP, thats why we buy after market heatsinks duh?


And last but not least, I dont know what you are smoking with the probes and all, but its pretty easy to measure Tcase.
In fact Intel placed a sensor right under the center of the IHS, and that sensor can be read by many programs including Everest, Speedfan, ntune, and the list goes on.
So its pretty easy to check if you are going over the safe Tcase proposed by intel.


I still have no idea why people continuously base their readings from core temperatures when we have no idea what is a safe core temp.
All we know is they throttle at x degrees, but we dont know whats safe since throttling temperatures must be far past safe, while on the other hand we do know what a safe Tcase is.
IMHO core temps should be checked just to make sure there is a reasonable delta between Tcase and Tjunction, and nothing more.


Make more sense now?

Incorrect. That is not the true 'Tcase'. Read the OEM System builders Thermal Specification guide by Intel. You will see that they machine a slit on the TOPSIDE of the IHS in the GEMOTRIC center. That will give you all the correction you need. Otherwise, we can simply use the defination they give of the Tcase.

From Intel's Website:

The thermal specification shown is the maximum case temperature at the maximum Thermal Design Power (TDP) value for that processor. It is measured at the geometric center on the topside of the processor integrated heat spreader.

In case you missed it. It is clear you came in here running your mouth before reading any of their thermal guides. The only reason I decided to correct you is to stop the spread of misinformation. You, like many others, just run your mouth and talk down to people. I find that your attitude at AT sucks and would have banned you a long time ago if I had the powers.

For those wonder what he is talking about is the DIODE on the BOTTOMSIDE (actually, in between the cores) of the IHS, not the TOPSIDE. They are two different measurements. The Tcase that intel reports on their specification page is gathered by slitting a small line into the IHS and then placing in a very, very thin probe. This is used to measure the Tcase that Intel specifies. The diode underneath the CPU is NOT the same thing but it called Tcase in some cases but would be similar in temperature, but not the same.

Edit **

I found another post regarding a lot of this information. It should set the record straight.

OCFORUMS POSTER...

Edit 2 **

And Jag, if there is going to be any further correspance with you, then you will need to drop your attitude and provide proof of your claims. I can't say I dissagree with your first two paragraphs, but the attitude needs to be dropped. I also agree with your conclusion.

roger that


but lets be objective ArchAngel, Tcase measured on the bottom side of the IHS would only provide a larger margin for safety. IMHO it would only differ by a few degrees from a sensor placed on the topside, but even if it does differ, its always going to be higher, not lower.

so you cant really screw yourself by observing that reading, and its the closest thing you have to the real Tcase. the core temps are far higher, and the socket temperature reported in the BIOS is completely off the wall. most of the time it reads a lot higher simply for the smart fan options to kick in sooner rather than later.

do you agree?

Originally posted by: ArchAngel777

Originally posted by: JAG87
ArchAngel, obviously you missed most of the things I said.

Before I posted, I took you for your word on this. So, I decided to read your post I replied to and thought "Wow, I already read this, what did I miss?" So I searched the entire thread and don't see anywhere that you said any of those things... In fact, the post I quoted was the only thing you have mentioned in this thread... Care to clarify? Or... Did you just 'think' you mentioned those things?

no, I was referring to the fact that I said the Thermal Spec from intel is the recommended temperature, not the absolute max. I had already mentioned that, but you repeated it as if I didn't.

I Agree 100% with your two replies to this. We are in total agreement at this point then.

:beer:

 

[JAG87]

Originally posted by: ArchAngel777

I Agree 100% with your two replies to this. We are in total agreement at this point then.

:beer:

perfect!

 

[TheOtherRizzo]

A question I've never seen asked or answered: why don't AMD tell us what the max Tjunction for their chips is? On my Toledo looking at case temps is useless because they were obviously never calibrated (difference of 15°C between cores) and looking at core temp is useless because nobody knows what the max temp is. I tried to find out once at what temp it would start throwing orthos errors but I chickened out when I reached 110°C (do AMD even have throttling?).

Anyone have any ideas about max tjunction?