lapped my IHS on a Q6600 (pics and results)

[jason4207]

Just to clarify...

If I have a B3 x32x0, or Q6600 I can safely run it up to ~77* as seen in CoreTemp.

If I have a G0 Q6600 I can safely run it up to ~86* as seen in CoreTemp.

If I have a G0 x32x0 I can safely run it up to ~100* as seen in CoreTemp.

I had thought everyone on here was trying to keep their CoreTemp readings at or below the Thermal Spec (as listed on Intel's website). This is very good news indeed! Can I get some more confirmations please! I don't want to fry anything!

If this is true I'm going to have to tell my buddy to crank up his B3 x3210 a notch b/c I was trying to keep it under 62* w/ p95 running. I was getting 65-66* at 2.4GHz (1.2v) w/ p95 running, but was still under 70* at 2.66Ghz. He's still running the stock cooler BTW.

 

[Brolloks]

Bing would be the best guy to clarify IMO

 

[graysky]

@Jason: dude, you're temps are too hot according to Intel's own specs. Look up your processor at the processor finder to see what the max temps should be.

 

[jason4207]

@jason: dude, you're temps are too hot according to Intel's own specs. Look up your processor at the processor finder to see what the max temps should be.

Those aren't my temps man...you need to read further up the thread.

Those are "what if's" based upon what GotNoRice posted.

Bing, can you answer this question about TCase and TJunction temps?

 

[GotNoRice]

@jason: dude, you're temps are too hot according to Intel's own specs. Look up your processor at the processor finder to see what the max temps should be.

From that very link, if you look up your processor and then click on "Thermal Specification" so that the details box comes up, it says:

"Thermal Specification: 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."

Notice how it does NOT say measured using the Digital sensors on the core? That would be the TJunction temp, which is hotter than the temp at the IHS, typically by 15C under load.

Note: There is a separate issue that is confusing due to the fact that is also involves a 15C difference. Older versions of Coretemp report TJunction temperatures based on an 85C max. B3 and I would assume G0 processors also have a TJunction max of 100C. Since TJunction temps are calculated by subtracting the sensor value from the TJunction max, if the program thinks your max is 85 when it is really 100 it will report temperatures 15C too low. You can see how on the surface it would appear to correlate to TCase temperatures since you’re talking about a 15C difference in both cases, but there is no connection. Version .95 of coretemp should report your TJunction max correctly.

 

[jason4207]

I definitely believe you GNR. It makes total sense to me. It just seems from my reading that everyone else was trying to get their CoreTemp .95 reported readings to be less than or equal to Intels thermal specification when running p95 or Orthos. I may have interpreted posts wrong, but this is the general feeling I had based on all my readings.

If someone says they are running their G0 Q6600 at 80* everyone would say that is too hot. You are saying that is not the case, and they should be fine as long as their voltage isn't too high. Because too much voltage at any temperature can fry a chip.

I am aware of the problem w/ CoreTemp vs. speedfan temp reporting, and understand it is a separate issue. You have shed some light on a few things I didn't completely understand yet, though...thanks!

 

[Neuromancer]

I just use the little black plastic cover that comes with the CPU. It's not a bad idea to q-tip the backside with rubbing alcohol after lapping as well. LGA775s are soooo much easier to lap than socket types that have pins.

word

Even doing all the protective measures after 70-90 minutes of rubbing on sand paper... i had long lasting dents in my fingers where I grabbed the corners of the cpu.

I didnt "wet" sand until I had actually sanded my fingertips till they were bloody LOL (I was on 1500 grit by then too)


I didnt do a Quad, although I definitly would after seeing how easy it is on the E series

 

[Brolloks]

I definitely believe you GNR. It makes total sense to me. It just seems from my reading that everyone else was trying to get their CoreTemp .95 reported readings to be less than or equal to Intels thermal specification when running p95 or Orthos. I may have interpreted posts wrong, but this is the general feeling I had based on all my readings.

If someone says they are running their G0 Q6600 at 80* everyone would say that is too hot. You are saying that is not the case, and they should be fine as long as their voltage isn't too high. Because too much voltage at any temperature can fry a chip.

I am aware of the problem w/ CoreTemp vs. speedfan temp reporting, and understand it is a separate issue. You have shed some light on a few things I didn't completely understand yet, though...thanks!

Would be great if we can get some real confirmation on what max temp the q6600 can take, if it is 80 C then I can pobably get to 4 G's, that would be awesome

Edit...also what is max voltage for this CPU?

 

[jason4207]

Would be great if we can get some real confirmation on what max temp the q6600 can take, if it is 80 C then I can pobably get to 4 G's, that would be awesome

Edit...also what is max voltage for this CPU?

Can anybody else confirm?

 

[bing]

Finding CPU Throttling Temp sometimes will boost your OC morale !

Would be great if we can get some real confirmation on what max temp the q6600 can take, if it is 80 C then I can pobably get to 4 G's, that would be awesome

I guess some of you recognize me for one of the OcFer here that do not trust cpu absolute temp reading using software.

Now, this is about the "other" method to boost your OC morale on squezzing more performance from your current OC.

Use this method if you got scary on the temp reading you've seen when your OC-ed CPU on loaded, "but" you still don't believe that temp reading accuracy when using software.

Warning ! I never recommend this to everyone, read and undertand the risks & consequences.

If... again a big "IF" you have the gut, my suggestion is find the CPU throttling temp with these following method :

• Put your mobo on bench or table, not in casing, and use Intel stock HSF, also I suggest add few fans to blow on the mobo mosfet and chokes and other parts as well.
  
• Yank out the HSF fan and hold it with one of your hand right above the heatsink.
  
• Now, run your favourite temp monitoring program, throttle watching program like RightMark CPU Clock or ThrottleWatch and Prime95 to load the CPU.
  
• Once the Prime95 launched, watch the temp rising and be prepare another hand which is not holding the fan at the power switch for "emergency shutdown" just in case.
  
• While watching the program toasting the CPU, watch closely the temp reading and the throttling windows.
  
• Start swing or waive the fan "slowly" farther from the heatsink to slowly heat up the CPU until it starts throttling.
  
• Once you saw the throttling started, mark down the CPU temp, this will be your last hottest temp that the "TRUE" Intel Throttling Mechanism (TM) starts to kick in.
  
• Once that throttling temp is acquired, at this point it is up to you whether you want to turn it off thru the power switch or cooling the cpu by bringing the fan closer to the heatsink and watch the process of CPU cooling and where the throttling stops at.

This will last only for 3-5 seconds when the throttling starts, been there with 3 pre core Intel cpus and they're all still alive today & kicking !

Just "an example" if you captured the throttling temp for your CPU is at 93 C and your current maximum OC on loaded CPU is at 68 C,

Now, the last important question that you should ask your self as an "OC-er" :

"Do you have the gut to squeze more ? Maybe like until it reached 78C or even 80C ?"


Final Warning !
Do this at your own risk, I will not be responsible for any problems or damaged that might happened to your rig.

 

[GFORCE100]

Your problem may be two fold: 1.38 V is quite to bit of heat since the heat increases with the square of the vcore. My temps are @ 1.2625V (1.232 V in windows) which is about 150 mV less than your system -- 150 mV is a lot of juice. Here is a copy/paste out of this thread I wrote about minimizing vcore:

Here is a more detailed analysis of two difference vcore settings and the temps they produce on a Q6600 @ 9x266=2.4 GHz as well as @ 9x333=3.0 GHz. The two voltages I used were 1.112 V and 1.232 V (both of these are the load voltage, the actual BIOS settings were 1.1375V and 1.2625V respectively). 2x orthos ran for 30 minutes and the temperatures were averaged over the last 10 minutes of those runs (well after they stabilized). Room temps was 75-76 °F. Notice that the difference in voltage is ONLY 0.120 V or 120 mV, but this seemingly small difference brought the load temps up by an average of 6-7 °C per core!

Run1 (9x266 @ 1.112 V), Average temps (°C): 51,52,50,50
Run2 (9x266 @ 1.232 V), Average temps (°C): 57,58,57,57
Differences (°C): +6, +6, +7, +7

Now if I add a faster FSB, they increased further:

Run3 (9x333 @ 1.232 V), Average temps (°C): 61,61,60,60
Differences from lowest voltage (°C): +10, +9, +10, +10
Differences from same voltage (°C): +4, +3, +3, +3

You're a full 10 % faster on your o/c as well as being 150 mV over mine. Did you say you lapped the chip also? If not, I would highly recommend that you do. I dropped off about 10 °C by lapping my Q6600 as well as the base of my ultra-120 x.

I retested my temps at 2.66GHz, 1.24V actual Vcore, 1066MHz FSB.

Under 4x Prime95 24.14 Small FFT the general CPU temp is 45C while cores 0-3 are 62/61/59/58.

This is still with an unlapped CPU only lapped IFX-14 which in the above scenario now feels cool.

Ambient today is 24.5C.

With the above it would seem my IHS is pretty flat if temps can go down if one takes away 666MHz and the difference between 1.38 - 1.24V = 0.14V Vcore.

I have 2x 2400rpm 120mm fans that can go on the IFX-14 but they'll probably be too loud unless I get a fan controller and set them around 1800rpm max just before the hairdryer sound barrier kicks in.

 

[treatmentx]

I just wanted to post the pic of my g0 i lapped. Still waiting for my RAM to come in, but wanted to say thanks for the author of the guide. I really lapped the SOB. Note the reflection off the chip is the texture OF THE BOX, not the chip's surface... an idea how smooth it went.

Btw, I was able to rub it down all the way to the copper playing on the first 600grit, but it was really concaved, as was mentioned.

 

[graysky]

@GF100: I think you should lap your chip dude. If you managed to do your HS, the chip will be cake in comparison.

@treatmentx: looks like a nice job you did. It's hard to tell from your shot because it looks to me as though you focused on the wording on your box (i.e. the chip is out of focus due to the depth of field @ f/5.6); did you get the top of the chip flat enough? Seems like there is a little more material you can take off from the extreme sides of the chip. It might be an illusion of the photograph though... the flash fired at a pretty close-up distance can do that.

 

[GFORCE100]

@GF100: I think you should lap your chip dude. If you managed to do your HS, the chip will be cake in comparison.

@treatmentx: looks like a nice job you did. It's hard to tell from your shot because it looks to me as though you focused on the wording on your box (i.e. the chip is out of focus due to the depth of field @ f/5.6); did you get the top of the chip flat enough? Seems like there is a little more material you can take off from the extreme sides of the chip. It might be an illusion of the photograph though... the flash fired at a pretty close-up distance can do that.

I lapped thw QX6700 up to 2000 grit so now both the IFX-14 and it are lapped and guess what, temps haven't changed.

The only thing that's changed is that the temp variation between the cores is now only 1-2C 90% of the time.

At 3.33GHz default Vcore I'm still getting 84/84/86/86C in Coretemp 0.95 in Prime95 25.14. Let it run for a bit longer and it's 86/86/89/89 so temps are even worse.
At 3.00GHz 1.25V Vcore I'm getting 68/68/65/65C in Coretemp 0.95 in Prime95 25.14. A 20C delta between these for just 333MHz more? That's what peltiers do, not air coolers, just underlines there's a spanner in the works.

I checked if the S775 top retension bracket is sitting higher than the IHS voiding the latter contacting with the IFX-14 but now, there's a 1mm height difference.

At this stage I'm just lost and somewhat annoyed at what the problem is. Just about everything I could have done has been leaving little else.

I would haved expected temps to now be in the 70's at most.

I reckon the IFX-14 isn't performing up to scratch though why remains a mystery.

Something somewhere is not giving up the fight. If only I had some of that paper dentists use where you place between your teeth to check their texture. It would allow checking just how well the IFX-14 contacts with the IHS which now should be almost 100% ideal.

 

[graysky]

Wow dude, I don't know what to say about this... perhaps the IFX-14 is just an inefficient HS like some of the articles suggest... did you manually set the vcore? You said "default VID" which could mean it's automatically managed.

 

[El<(')>Maxi]

I just wanted to post the pic of my g0

Nice job treatment, love that shot

 

[GFORCE100]

Wow dude, I don't know what to say about this... perhaps the IFX-14 is just an inefficient HS like some of the articles suggest... did you manually set the vcore? You said "default VID" which could mean it's automatically managed.

Yeah I do think it's the IFX-14, however ilogical that may sound. Everywhere where I've read people lap their heatsinks/IHS has been proof that temps go down, sometimes 5C, sometimes 10C but they do go down. I've even got the back CPU cooler installed.

I have photos of my lapping job in progress but as I was doing it on a mirror I can't see how the IHS can't be 100% flat now. 2x 600 grit, 3x 800 grit, 1x 1200 and 1x 2000.

I would love to get this fixed, probably with another cooler but I don't want to go out spending loads on some other new cooler that might just perform the same, as it can't really get worse than 90C under load.

 

[bing]

GForce,

Have you tried stock Intel HSF (yes, those irky push pins) and run the CPU at "stock speed" with stock vcore (vcore = VID) to see those temp reading ?

No, I'm not joking and that is the only way to tell either your IFX has a major defect "or" your temp reading is way off.

 

[GFORCE100]

GForce,

Have you tried stock Intel HSF (yes, those irky push pins) and run the CPU at "stock speed" with stock vcore (vcore = VID) to see those temp reading ?

No, I'm not joking and that is the only way to tell either your IFX has a major defect "or" your temp reading is way off.

As bad luck as it may sound, I don't have an original Intel cooler I could try, unless I specially buy one.

I thought CoreTemp 0.95 always reports the right temps? Even say it did report bad temps, it would still not answer why, after all this lapping do the temps not move, or even get worse by a small margin.

The IFX-14 is screwed on very well, I can't twist it so there's pressure alright or so one's led to believe.

My understanding would be that any CPU heatsink that can't handle taking the CPU heat load off would be very hot before CoreTemp shows very high temps but this just doesn't seem to be the case I could touch the IFX-14 all day long and not burn myself, with 2x 1200rpm 120mm fans or not (though with the latter it does naturally feel a little warmer).

If only I had some of that dentist's paper I mentioned above to see how well it's all contacting each other I would then know where and how much pressure there is, who knows maybe there's a 0.5mm gap.

 

[bing]

Coretemp accuracy is the matter of luck.

Any OC-er friends around the neighbourhood ? Most never use them, maybe you could buy it cheap ?