Temp Reading

[epion2985]

Since there seems to be be quite a few problems with core temp and other programs not reading/calculating temperature on all cpu's or at least in sub zero range very well, as it seems from some threads here, this has been suggested by a friend engineer to me when we were chatting. Is this reasonable? Seems to me a pretty good way to go about it.

Make small hole in IHS and use a K type probe with a non conductive TIM to get a direct reading on the core. Granted you should know the internal layout to land the probe on it and not miss

[Brettbeck]

It would probably work if done with extreme precision. But I think there would be very few people who would actually drill a hole in to their CPU's IHS !

[epion2985]

It would probably work if done with extreme precision. But I think there would be very few people who would actually drill a hole in to their CPU's IHS !

Thats the point, to drill a hole in the IHS and apply the probe directly to the core. Or did you mean "some people may drill a hole in the cpu core" ?

[Vuco]

How tall is the HS ? For the probe to get in, you should have a hole 1-2mm in diameter, right ?

The only way to drill would be to get the HS off the CPU. Many people wouldn`t do even that...

Or did you mean to drill the hole while HS on the CPU ?
But I agree, this would be the best way to monitor the core temp of any CPU.

[Viss]

I think it has been done with A64 already but cannot find the pic anymore.

[Brettbeck]

Thats the point, to drill a hole in the IHS and apply the probe directly to the core. Or did you mean "some people may drill a hole in the cpu core" ?

I mean I doubt many people would want to drill holes in to their CPU, because the slightest mistake and you're likely to have a wrecked CPU!

[DetroitAC]

Not a problem with a milling machine. A good machinist can spot drill first, and then drill to a precise depth.

[epion2985]

How tall is the HS ? For the probe to get in, you should have a hole 1-2mm in diameter, right ?

The K type probe is under 1mm in diameter I believe.

The only way to drill would be to get the HS off the CPU. Many people wouldn`t do even that...

No you can easily make a hole without taking it off. You simply have to know where the core is on your cpu and not drill deep as the IHS is only what 1mm thick.

Or did you mean to drill the hole while HS on the CPU ?
But I agree, this would be the best way to monitor the core temp of any CPU.

Yeah thats the idea.

I mean I doubt many people would want to drill holes in to their CPU, because the slightest mistake and you're likely to have a wrecked CPU!

Well given the core (s) are usually quite a distance away form the IHS wall its not that dangerous at all. Just need a decent drill press or a mill. I wouldn't try to wing it with a hand held drill though lol

Not a problem with a milling machine. A good machinist can spot drill first, and then drill to a precise depth.

Indeed. I am going to try this on my new Xeon 3060. Anyone got a picture of a naked one?

[Gray Mole]

Seen it done here and there, and it would be a really nice tool to meter the temps of the IHS and extrapolate the core temp.

But at the same time...

Well, for one, that's the warranty gone. I mean most of us that would pop the ihs off don't care much about a warranty I suppose, but it just seems like it's an expensive hole that's being drilled.

Beyond that, is knowing the core temp critical?

If a phase or whatever system is cooling the cpu and working properly, the overclock is stable, and the temp that's being read is stable (unless it's simply not reading anything at all) then I'm not sure I see much point in being able to know accurately what the temp of the core is unless you're testing a cpu for it's exact response to a particular form of cooling.

Very nice for someone doing research, not really necessary for someone that just wants to overclock.

One of these days I'll be making another load tester for the power supply that's got a couple of features included with it, including a probe that's at 'IHS depth' and another that's on a secondary plate to simulate 'cpu core depth' but that's still only a simulation.

I feel, however, that once a relationship can be drawn between a particular cpu and a load simulator to determine it's actual wattage, then all of the susequent information can be extrapolated without having to drill a hole in every IHS on every cpu.

Just my thoughts on it.

Cheers

Gray

[johann]

I agree, Im certainly not drilling a hole in my £350 E6700...

Reading the temp simply is not that important to me.

[[XC] gomeler]

Even my cheap E6400 won't ever see a drill press, too valuable to take a chance with the vibrations and metal shavings screwing up the processor. Goodluck though with your Xeon3060, hope you got a good clocker!

[pumbertot]

also with Kentsfield you need to push 4 probes in if you want to be accurate.

also see your own sig for reason not to.

[epion2985]

Well, for one, that's the warranty gone. I mean most of us that would pop the ihs off don't care much about a warranty I suppose

Thats what I was thinking, on the extreme end people pop them off anyway, which is a lot more dangerous.

Beyond that, is knowing the core temp critical?

If a phase or whatever system is cooling the cpu and working properly, the overclock is stable, and the temp that's being read is stable (unless it's simply not reading anything at all) then I'm not sure I see much point in being able to know accurately what the temp of the core is unless you're testing a cpu for it's exact response to a particular form of cooling.

Appears to be so to a lot of people. Or maybe I am just getting my panties wet over nothing. Noticed a lot of people here try to diagnose and replicate results etc etc, data base threads with what was achieved at what temperature, makes the information pretty meaningless if the temp readings are not accurate.

I feel, however, that once a relationship can be drawn between a particular cpu and a load simulator to determine it's actual wattage, then all of the susequent information can be extrapolated without having to drill a hole in every IHS on every cpu.

Good point, but someone has to drill the first hole

also with Kentsfield you need to push 4 probes in if you want to be accurate.

also see your own sig for reason not to.

Then 4 it will be, though once I have one "reliable" core temp reading I would be pretty comfortable assuming the rest are with in a small margin of that.

You failed to understand my sig.

[pumbertot]

Thats what I was thinking, on the extreme end people pop them off anyway, which is a lot more dangerous.



Appears to be so to a lot of people. Or maybe I am just getting my panties wet over nothing. Noticed a lot of people here try to diagnose and replicate results etc etc, data base threads with what was achieved at what temperature, makes the information pretty meaningless if the temp readings are not accurate.




Good point, but someone has to drill the first hole




Then 4 it will be, though once I have one "reliable" core temp reading I would be pretty comfortable assuming the rest are with in a small margin of that.

You failed to understand my sig.

no i do understand your sig indeed, wouldnt want to be the first to pioneer this great idea you have. yep I think thats your responsibility.