Looking for 3770K Core position info

[red454]

My 3770K works well, and I haven't overclocked yet, but I see CoreTemp shows that core 0 and core 1 are typically 10 degrees F higher than the other two under idle and load. I am using a Corsair H100.

I remember seeing a picture or graphic that showed the relative core positions on the die so you can look at the CPU and point to each core under the lid.

I want to take a look at the contact between my H100 waterblock and the CPU surface.

 

[Homeles]

 

[lehtv]

It's common to see higher temperatures on one of the cores. Of course it can't hurt to check that the thermal paste is evenly spread, but my guess is that your CPU just is like that and there's nothing you can do about it, and that there's nothing to worry about

 

[Idontcare]

It's common to see higher temperatures on one of the cores. Of course it can't hurt to check that the thermal paste is evenly spread, but my guess is that your CPU just is like that and there's nothing you can do about it, and that there's nothing to worry about

Not just common, its practically unavoidable owing to the intrinsic asymmetric thermal dissipation pathways that arise from having an asymmetric die layout.

Of course you can have the very unlikely scenario pan out for your CPU and HSF combo in which the regions between the die, the IHS, an the HSF are so non-uniform in gap heights such that the thermal dissipation from the intrinsically cooler dies is compromised in a way that results in them actually running hotter than the other dies...but that is a contrived situation that is rather unlikely to occur in practice.

 

[red454]

Ok - thanks - I'm not worried about it so much, rather I just want to make sure I have the cooler seated properly against the CPU and wondered what the relative core position is.

Years ago in my hot rod days, we used thin wax or plastic to check bearing clearance. I think it was called plasti-gage.

You would put a piece of it on the bearing and assemble the bearing cap. Then disassemble and measure how much the plasti-gage squashed, which would give you the bearing clearance. Also used dykem - that would leave a dye patch transfer anywhere two opposing surfaces made contact.

Last week I experimented with some Bostik Never-Seez as a thermal paste. It is used to keep metal components from seizing up when you take them apart and it is loaded with copper. Good to 1800 degrees too. It did a fair job - about 2 or 3 degrees warmer than my stock paste. I then put the stock paste back on. Which leads me to wonder if I got the cooler base seated...

 

[Idontcare]

For heatsinks and waterblocks, there is a pressure-sensitive paper you can use to give you an idea of mounting contact uniformity.

I've never used it and the product name escapes me at the moment, but if you browse much over on XS forums you'll see lots of people posting pics when using it.

 

[red454]

For heatsinks and waterblocks, there is a pressure-sensitive paper you can use to give you an idea of mounting contact uniformity.

I've never used it and the product name escapes me at the moment, but if you browse much over on XS forums you'll see lots of people posting pics when using it.

I've seen that too - can't think of the name. Reminds me of the stuff the dentist uses to check your bite / how your upper and lower teeth mesh.

Really what started all of this is that I noticed that this processor (3770K) runs a little hotter than my 2600K did, particularly during folding. It will spike to the low 140's F. The 2600K spiked in the upper 120's using the same H100. Idle peak for the 3770K is 82 F.

I read all your info on delidding too.

 

[Idontcare]

LOL, I bet it is exactly the same thing, my dentist uses it too

Your temps sound fine for a non-delidded 3770k. Only delid for hobby interests or if you are serious about getting the absolute top OC out of your 3770k.

Yeah you save some $ in lower power usage if you delid, but if you were interested in saving money then you would have stuck with the 2600k in the first place and not paid for a 3770k

 

[Termie]

Quick check without reseating:

Is hottest core at idle different than hottest core at load? If so, my hunch is that it's not a mount issue. Valid, IDC?

My core 0 is hottest by 10 degrees at idle but coolest at load.

 

[red454]

Quick check without reseating:

Is hottest core at idle different than hottest core at load? If so, my hunch is that it's not a mount issue. Valid, IDC?

My core 0 is hottest by 10 degrees at idle but coolest at load.

They of course jump around, but idle is

Core
0 82
1 82
2 68
3 79

100% Load (folding)

Core
0 127
1 138
2 120
3 122

 

[Termie]

They of course jump around, but idle is

Core
0 82
1 82
2 68
3 79

100% Load (folding)

Core
0 127
1 138
2 120
3 122

Ok, generally speaking they are in the same order at idle and load. Could be mount, but it's ridiculously cool overall.

 

[Idontcare]

Quick check without reseating:

Is hottest core at idle different than hottest core at load? If so, my hunch is that it's not a mount issue. Valid, IDC?

My core 0 is hottest by 10 degrees at idle but coolest at load.

It is valid but only if we accept that we are making other assumptions in the process, and those assumptions are (unfortunately) on questionable ground.

First questionable assumption is that the error in reported temperature (which itself will be a function of temperature) across the cores does not result in the changing of the reported hottest core itself.

The hottest core at idle could also be the core which has the worst intrinsic error, an error that is less than the actual temperature delta between that core and the true hottest core when fully loaded, resulting in the "hottest core" award changing around on loading.

Second questionable assumption is that the heat generated by each core (a function of both conductor resistance and undesired leakage) is the same across all cores such that the disparate heating can only be a function of disparate thermal conductivity from a bad mount and not from disparate power-consumption.

We can't rule out either of these two assumptions, they are assumptions we are forced to make. But these two assumptions are tenuous at best which makes it equally tenuous to conclude that a flip-flop in the "hottest core" ordering when going from idle to load is the result of a poor HSF mount.

The challenge here is we could be right, but for all the wrong reasons; or we could be wrong for all the right reasons (and never know it, either way).

 

[Ferzerp]

Someday, people will realize that Intel idle reported temps are worthless when they're under 35-40C...

I mean, you see reported temps under ambient commonly. Why put *any* stock in it?

 

[LogikJ]

As with Termie my Core 0 is also 10-12 Degrees cooler at idle but under load it is the coolest or within 5 or less of the other cores.

 

[IntelEnthusiast]

You have to love when IDK explains these things. As he makes it so clear for the rest of us non-engineers.