New Rig - Questions

[p4ck3tl055]

I FINALLY got all my parts (See Signature) and I've built my new i5 system (at least to the point of getting it test worthy). I was about to start OCCT to burn it in (no overclock yet), but before I did, I noticed this info from CoreTemp:

Notice how Core2 is 7° C cooler than the closest other core? Is that something I should be worried about? Like maybe an uneven coating of thermal paste?

This is the first multi-core CPU system that I've built, so I want to make sure that I don't toast it the first night.

 

[Phrost Phantom]

The first core might be hotter due to how close it is to any controllers or other parts of the processor.

 

[Shephard]

nothing to worry about. My Core 2 is high as well.

As of typing this post...

Core 1 - 20
Core 2 - 30
Core 3 - 26
Core 4 - 20

 

[p4ck3tl055]

One thing to note...I'm currently running on the embedded Intel Graphics 4000 instead of the GTX 660. Maybe this is the reason Core 0 is hotter than the others?

I ran a quick 10 minute Linpack wih OCCT and cores 0,1 and 3 evenend out under load, but Core 2 was always cooler by several degrees C. Nothing borked or crashed.

I think I'm going to double check my BIOS settings and then either run Prime 95 or another Linpack for a few hours while I sleep.

Will report back tomorrow.

 

[TuGuX]

Worst thing could be the cooling paste (not the same amount on the surface)

 

[p4ck3tl055]

I think I'm going to double check my BIOS settings and then either run Prime 95 or another Linpack for a few hours while I sleep.

Machine came through 6 hours of Linpack with no problems - Core No. 2 remained constantly cooler than the other three according to the graphs (which I don't have uploaded yet, sorry).

Not that it matters, but when I went through the BIOS settings I discovered that my RAM was in Single Channel mode instead of Dual. This was because I had installed the RAM sticks side by side instead of skipping a slot. (ASRock's manual has a diagram that led me astray on that one.) Easy to fix.

Worst thing could be the cooling paste (not the same amount on the surface)

This is what I suspect. Tonight when I get home, I'm going to check this. It's worth a drop of thermal paste for my peace of mind.

 

[BTRY B 529th FA BN]

It's a normal, don't worry about it.

 

[Vectronic]

Like a few have said, it's normal.

(for reference, I'm not using zero-based indexes)

If you are using the iGPU, core 1 will be warmer than it could be, core 2 and 3 will always be warmer than 1 and 4 (within reason) and core 4 will almost always be the coldest core.

[iGPU][C1][C2][C3][C4][MC]
[iGPU][L3][L3][L3][L3][MC]

[MC] doesn't put out any heat at all really, neither does [L3]... so basically it's just:
[iGPU][C1][C2][C3][C4]

Core 2, and 3 are stuck between 1 and 4, so not only are they heating up, but they are getting heated up by the other cores, and only have a limited surface area to output their heat.

If you are using the iGPU Core 1 has that same limitation, if you aren't using the iGPU then Core 1 will drop in temps, especially at load... Core 4 is only about 1/3rd surrounded, so it's always the coldest.

It's pretty normal to have a 10C difference between coldest and hottest cores because of this, and because the TIM between the die and IHS is... well TIM instead of solder which doesn't even out the heat very well.

With some CPUs the difference is a lot, and people have to de-lid and re-apply the TIM just to get them somewhere near balanced again.

Your core 3 is suspiciously low (or core 4 is suspiciously hot)... which might just be because of what apps you are running... it could be poor TIM under the IHS... at the moment, with your temps, not an issue.

 

[p4ck3tl055]

Worst thing could be the cooling paste

I got home tonight and removed the CM Hyper 212 Evo. Here's what I found:

Bottom of the Evo:

Top of CPU:

So it looks to me like my first attempt missed about a quarter of the CPU with the paste.

I cleaned and reapplied the paste. I'm hoping that with the second attempt, I did a better job. I did not pull the cooler again to see, I just fired the rig back up. I received similar temps according to CoreTemp.

Because of that I'm about half-tempted to pull the cooler again. But everything seems to functioning fine, so I'm also about half-tempted to just load the machine up and let'er rip.

My main concern about these temps is that I do plan on a mild overclock (like 4.2GHz, maybe even 4.6 if I can pull it off on air) and I'm concerned that once again, the paste isn't even and I'm going to cook my CPU when I try to overclock it.

Oh, and for those intrested, on the ASRock Extreme 6 with the CM Hyper 212 Evo, there is JUST enough room for you to use all four memory slots:

I was very concerned about this when I spec'ed the cooler, but now I (and you) know.

 

[Blain]

So it looks to me like my first attempt missed about a quarter of the CPU with the paste.

That's why I coat the smaller of my components (CPU or HS) with a thin layer of TIM.
I coat the smaller of the mating surfaces because it will be contacting the larger surface at 100%, and not waste TIM on non-contact areas.

 

[FridayThe13th]

I've always wondered about this too... the point of not spreading it is to not have air bubbles in the paste.

would it be better to have a drop in the middle spread out with pressure or to apply thin layer on the whole chip?

im thinking of redoing mine

 

[p4ck3tl055]

would it be better to have a drop in the middle spread out with pressure or to apply thin layer on the whole chip?

When I built my first systems - Pentium 4 days - I always spread the paste. But I never was satisfied with the job and looking back I always used too much.

This time, after watching several people use the "rice" sized drop method (on YouTube) I thought I would give that a try. This machine was my first and second try.

I was shocked to see the CPU with my first attempt, thinking I would have gotten more of a centered spread instead of all the paste running to one side. I've not take the cooler off the second time. I'm getting to the point where curiosity is going to get the best of me. As I get closer to putting the machine into normal use (I'm typing on it right now), I'm really wanting to be done with building and getting on to playing. But at the same time, I'm wanting to make sure that the rig is going to last me for 3+ years. So I'm thinking if every thing goes well tonight, when I crack the case for hopefully "the last time", I'm going to pull the cooler one last time.

If the paste looks the same as I've shown here, I may give up on the drop method and go back to the spreading meathod - just to assure myself that the entire lid of the CPU is covered.

 

[crashtech]

It almost looks like an uneven seating of the HSF assy, or as if both attachments on the left were fastened before the ones on the right, causing the paste to run to the right before attachment was complete.

I tend to fasten or torque diagonally opposed fasteners in sequence. Is that what you do, p4cktl055?

 

[VirtualLarry]

It almost looks like an uneven seating of the HSF assy, or as if both attachments on the left were fastened before the ones on the right, causing the paste to run to the right before attachment was complete.

I tend to fasten or torque diagonally opposed fasteners in sequence. Is that what you do, p4cktl055?

That's what I was starting to think. Looks like one side was torqued down before the other, or torqued down more.

 

[p4ck3tl055]

I tend to fasten or torque diagonally opposed fasteners in sequence. Is that what you do, p4cktl055?

Yes I did tighten down the cooler in a diagonal pattern. (My father was a mechanic, so I learned this early on.) Maybe I had one side too tight to start with though...the 212 Evo is not the easiest cooler to mount.

Since I am planning an overclock attempt, I will give the cooler one last look before I button the build down.

My biggest problem right now (aside from my cooler worries) is the cable management. The Zalman Z9-U3 that I bought is a decent case, but it leaves a lot to be desired behind the motherboard tray.

I ran the machine under what would be a typical load (played Guild Wars 2 for about 4.5 hours) tonight and didn't have a single issue. The highest temp I saw was 57° C (28° C ambient) under 100% load.

All in all, it looks pretty good so far.

 

[crashtech]

Pictures can be misleading, but I still suspect the HSF was and perhaps still is not sitting level on the CPU. It's good that you are not seeing any problems, I guess if I was you it would bother me enough to try and determine if something was out of whack.

 

[2is]

The spread on your thermal compound is not that bad (in your case) the core is rectangular, not square and the long sides extend from that black label on the PCB, upwards. I doubt having that end covered would have pulled away that much more heat.

 

[p4ck3tl055]

I guess if I was you it would bother me enough to try and determine if something was out of whack.

Well, I took the cooler off again tonight (about an hour ago) and the second time around I did a much better job with the thermal paste (I'll load up some pictures once I can get them downloaded from my camera.). I suspect that in the first attempt, I either didn't have enough paste or, as several of you suggest, I tightened the cooler into place unevenly.

In any event, I'm glad I pulled the cooler after the second application and I'm confident that this last time (third time's the charm, eh?) I was able to get a good coating of paste.

Thanks for everyone's input and guidance. I really appreciate it!

Now to get everything else done on the machine...where's my external hard drive???

 

[p4ck3tl055]

I'll load up some pictures once I can get them downloaded from my camera.

Well, here is how the thermal paste worked on the second seating of the 212 Evo, I think I did a better job this time around (and i did it even better on the third time afterwards...)

As a matter of fact, upon the suggestions of everyone here, I've dismissed the cooler No. 2 core to normality and have finished the build. Check it out on my blog! (it was easier to point you guys/gals there than to link individual pictures. I hope it's not a no-no.)

 

[SpeedTester]

P4c,

That's quite a bit of excess paste, how much are you applying?

 

[crashtech]

Theoretically, we'd all like to see the material go right to the edge of the heat spreader and no further, in practice that is a bit more difficult to achieve. My applications generally squeeze out slightly, but usually less than shown in the OP's photo.

 

[p4ck3tl055]

how much are you applying

I tried to apply a small drop (slightly larger than a grain of rice) in the center...honestly, with the tube/syringe applicator it was more difficult to get a little than a lot. I barely pushed at all on the plunger on my first attempt and got a glob instead of a drop. Needless to say, I cleaned that off and tried again.

Either way, I don't think I did too bad for my 3rd (and now 4th) attempt.

 

[tweakboy]

A stripe on where the CPU cores are located.

 

[2is]

The whole thing does NOT need to be covered. There is and should be some metal to metal contact. Paste is there to fill in the gaps where there is NO metal to metal contact. Metal to metal will transfer heat much better then metal > TIM > metal. If you're applying so much TIM that the entire surface area is covered (even after installing the HSF and the excess is "squeezed out"), you're likely doing more harm then good as far as optimal heat transfer is concerned. Metal > Air > Metal is what you're trying to avoid.

 

[Grooveriding]

You're using far too much TIM. I get the best results using just the minimum, an amount that when I pull the block off the CPU, there is just a very thin coating - to the point I can see metal through it - remaining on the CPU.