2500k lapping results

Printable View

02-14-2011, 09:58 AM
dnottis

2500k lapping results

Keep in mind the pre-lap has had about 2 weeks of thermal interface curing. The Post lap were mins old. So over the next week my temps will probably drop a few more degrees as the TIM cures. But here are my results of lapping a 2500k. It was definitely worth the effort. Every chip I've owned I've lapped the last few years, it's always worth the effort imo.

Pre-Lap:
http://3dxtreme.net/other/Sandy%20Bridge/IMG_0022.JPG

Idle Pre-Lap:
http://3dxtreme.net/other/Sandy%20Br...dle-prelap.jpg

Load Pre-Lap:
http://3dxtreme.net/other/Sandy%20Br...oad-prelap.jpg

================================================== ==========

Post Lapping:
http://3dxtreme.net/other/Sandy%20Bridge/IMG_0025.JPG

Idle Post-Lap:
http://3dxtreme.net/other/Sandy%20Br...le-postlap.jpg

Load Post-Lap:
http://3dxtreme.net/other/Sandy%20Br...ad-postlap.jpg
02-14-2011, 10:03 AM
K404

Every 5 degrees colder is a good 5 degrees colder :)
02-14-2011, 10:07 AM
Eldonko

Wow, a lot bigger delta than I expected.
02-14-2011, 10:11 AM
dnottis

Thats my thought too. I'm trying lower voltage now and it seems I can IBT @ 1.35v where before I needed 1.365v. So not only lower temps, but less voltage. This is consistent with every chips I've lapped though. Its always a benefit imo.
02-14-2011, 10:12 AM
Hondacity

no 59 67 66 66
lapped 59 65 65 63

not much difference...

what is your tim?
02-14-2011, 10:23 AM
eva2000

i don't see the difference if you look at load-idle delta temps in load screenshot alone ? only accurate way to measure, unless your room/case ambients are temperature/climate controlled fixed ?

cpu0, cpu1, cpu2, cpu3
load-idle delta temps

pre-lap
27, 34, 36, 32 = 32.25C delta average

vs

lapped
30, 35, 29, 34 = 32C delta average

0.25C difference in delta temps averaged across 4 cpus
02-14-2011, 10:27 AM
Ket

Temp difference isn't just about TIM, it heavily depends on how concave the IHS is. In my case and dnottis case lapping the IHS is very much worth it as we both have horrible IHS surfaces. I have a 12c difference at idle/load between my coldest and hottest cores.
02-14-2011, 10:30 AM
eva2000

12C pretty wide variance, mine are about 5-6C apart

H50-1 cooled

http://fileshosts.com/intel2011/giga...blend_load.png
02-14-2011, 10:32 AM
Hondacity

Quote:

Originally Posted by Ket

Temp difference isn't just about TIM

i was just curious about the tim... coz he mentioned "cure" time
02-14-2011, 10:34 AM
dnottis

Quote:

Originally Posted by Hondacity

i was just curious about the tim... coz he mentioned "cure" time

as5
02-14-2011, 10:35 AM
dnottis

Quote:

Originally Posted by Hondacity

no 59 67 66 66
lapped 59 65 65 63

not much difference...

what is your tim?

not sure what u were looking at load temp were:

59, 66, 66, 65 - pre

54, 61, 59, 59 - post

btw, one side of the ihs was higher than the other.
02-14-2011, 10:42 AM
eva2000

look at max load temps, not current load temps :)
02-14-2011, 10:43 AM
Ket

Even 200MHz faster than I run 24/7 atm your temps are still way lower than mine :p:

4.6GHz 1.35v;

Idle: 35-41-29-36
Load: 63-73-69-66

Thats just with a few IBT runs, let it go 15-20 times difference gets up to 12c. How come your GFlops is so high in IBT? My CPU deffinately isn't throttling yet it's only just over half of your GFlop score.
02-14-2011, 10:47 AM
dnottis

Quote:

Originally Posted by Ket

Even 200MHz faster than I run 24/7 atm your temps are still way lower than mine :p:

4.6GHz 1.35v;

Idle: 35-41-29-36
Load: 63-73-69-66

Thats just with a few IBT runs, let it go 15-20 times difference gets up to 12c. How come your GFlops is so high in IBT? My CPU deffinately isn't throttling yet it's only just over half of your GFlop score.

I'm using Windows 7 SP1 and IBT with the new LinPack libraries so it's using AVX which runs a lot hotter too. I've got a LinX install here without the new LinPack Libs my GFlops is 70 when not using the AVX instructions.
02-14-2011, 10:48 AM
dnottis

Do you suggest AS5 or Arctic MX-3?

I've got both here, just been a long time user of AS5 I always find I get slightly better OCs and lower temps with AS5. I usually just use MX-3 on my GPU.
02-14-2011, 10:49 AM
dnottis

Quote:

Originally Posted by eva2000

look at max load temps, not current load temps :)

Just remember the pre-lap were cured too. The post are runs right after I reinstalled my block.
02-14-2011, 10:52 AM
Ket

Quote:

Originally Posted by dnottis

I'm using Windows 7 SP1 and IBT with the new LinPack libraries so it's using AVX which runs a lot hotter too. I've got a LinX install here without the new LinPack Libs my GFlops is 70 when not using the AVX instructions.

Doesn't AVX get used with IBT 2.5? I have W7 SP1 installed as well.
02-14-2011, 11:04 AM
dnottis

Quote:

Originally Posted by Ket

Doesn't AVX get used with IBT 2.5? I have W7 SP1 installed as well.

I had to rename and copy over the updated libs for the GFlops to jump to 120 so I don't believe so.
02-14-2011, 11:08 AM
Hondacity

eva and i looked at the registered max temps....

we'll wait for the "cured" "registered" max temps

thanks for sharing...
02-14-2011, 01:40 PM
Ket

Well, seeing this thread inspired me to lap my 2500k. Its made a bit of difference, but nowhere near the kind I was expecting. I might have to start looking at the orientation of the cooler being the biggest culprit. Problem is if I rotate the cooler round I then have a problem with the side panel going on because the side mounted 120mm fan catches the CPU HSF. Stuck between a rock and a hard place. :brick:
02-14-2011, 02:53 PM
dnottis

Quote:

Originally Posted by Ket

Well, seeing this thread inspired me to lap my 2500k. Its made a bit of difference, but nowhere near the kind I was expecting. I might have to start looking at the orientation of the cooler being the biggest culprit. Problem is if I rotate the cooler round I then have a problem with the side panel going on because the side mounted 120mm fan catches the CPU HSF. Stuck between a rock and a hard place. :brick:

What kind of drop did you see?

My prime temps are about 5 deg lower across the board. Still not all sensors are showing equal temp and 5 deg drop isn't the biggest I've had from lapping, still 5 deg is 5 deg. Still wondering if I should remove the AS5 and try MX-3 on the CPU...
02-14-2011, 02:56 PM
zalbard

Quote:

Originally Posted by Hondacity

we'll wait for the "cured" "registered" max temps

That'd be wise. :)
But I don't think SB changes anything in regards to lapping. 3-5C on average, sometimes a little bit higher gain, depends on how flat the heatsink is, and how bad the IHS initially was. :yepp:
02-14-2011, 03:16 PM
dnottis

Quote:

Originally Posted by zalbard

That'd be wise. :)
But I don't think SB changes anything in regards to lapping. 3-5C on average, sometimes a little bit higher gain, depends on how flat the heatsink is, and how bad the IHS initially was. :yepp:

Using glass to lap on, one side was down to copper the other wasn't. At the least, across the IHS was crooked has hell. The "heatsink" is a Dtek Fuzion water block.
02-14-2011, 03:32 PM
Ket

Quote:

Originally Posted by dnottis

What kind of drop did you see?

My prime temps are about 5 deg lower across the board. Still not all sensors are showing equal temp and 5 deg drop isn't the biggest I've had from lapping, still 5 deg is 5 deg. Still wondering if I should remove the AS5 and try MX-3 on the CPU...

Temps seem to be far more steady and a bit more even with eachother, the IHS was deffinately cratered. Even using extra course sandpaper the middle remained stubbornly nickel plated for a while. HSF is a Hyper 212 Plus which kept a Phenom2 955BE @ 3.8 to 60c with all 4 cores 100% loaded so the cooler is pretty good. The problem has to be orientation, to get the side panel on the cooler is currently mounted so the heatpipes go straight across the IHS, meaning only 1 of the 4 direct heatpipes is over the CPU cores. Mounting the other way I'd think would deffinately hold significant improvements but then my side panel would have to go without a 120mm fan and it wouldn't look as beautiful :(
02-15-2011, 05:50 AM
Ket

Has anybody tried popping the hood on a 2500k or a 2600k? I've just relapped the IHS to make absolutely sure its flat so its deffinately not going to be any flatter now. Unfortunately, the temp delta is still high and the only thing remaining to do is pop off the IHS, so any links if anyone has done it will be very helpful :D
02-15-2011, 05:54 AM
dnottis

Quote:

Originally Posted by Ket

Has anybody tried popping the hood on a 2500k or a 2600k? I've just relapped the IHS to make absolutely sure its flat so its deffinately not going to be any flatter now. Unfortunately, the temp delta is still high and the only thing remaining to do is pop off the IHS, so any links if anyone has done it will be very helpful :D

Maybe at this point it's just the inaccuracy of the temperature sensors too.

I guess we'd have to see someone with direct core contact to hear their results.
02-15-2011, 06:02 AM
zalbard

I bet it will be pointless.
It's been pointless since Intel started soldering IHS...
02-15-2011, 06:18 AM
Ket

Intel don't solder all of the IHS' of their CPUs, only the real highend ones last I heard. Its quite possible its just some TIM underneath the IHS of Sandy CPUs and it would explain how even after a super perfect lapping the difference in temps between my hottest and coolest cores at load is still 11c. Temps are more stable as a result of the lapping though so it wasn't entirely pointless.