TR Forums

derFunkenstein · Thu Jun 06, 2013 2:56 pm

ClickClick5 wrote:
My fellow Gerbils, welcome to nerd porn.

I admit to being mildly aroused.

ColeLT1 · Thu Jun 06, 2013 3:00 pm

chuckula wrote:
ColeLT1 wrote:
Chuckula, I read an interesting article, and trying my best to find it again, that they tested all forms of different TIMs, and the intel's ranked up with the best. They posted temps and results, then culprit was not the TIM, but the black epoxy gluing the IHS to the board creates too large of a gap between the IHS and core. They removed all the epoxy and put the IHS back on without touching the TIM, then removed the IHS and tested with high quality TIMs between the IHS and core, and saw equal temps to high quailty TIM, and much lower temps than stock.

Granted, solder would be better than any TIM.

I'd be very interested to give that a read if you locate it! I'm definitely onboard with the theory that it is the glue that introduces a gap between the IHS and the die.. and air gaps are killer no matter what kind of TIM you are using.

I would love to find the old forum post (from last year maybe on extremeoverclocking), with the person who found this, and some confirmations from other users, but it really is a needle-in-haystack situation. I was interested in de-lidding my ivy, I used to delid all my athlon 64 and x2 chips (939 days), and get a whopping 2.5ghz off of 2.0 chips (crazy I know), and was forum lurking. I ended up buying intel's $20 OC insurance for my I5 and not cutting it, because I was going to run 1.4v and run at 4.8GHz until it burned, but then wussed out and backed it off to 4.7 at 1.32v.

Just some quick googling of "haswell overclock guide" and I came up with this:

Speaking of temperatures, surely people will be saying “Haswell has the same problem as Ivy Bridge,” referring to the thermal paste issue. Yes, Haswell has thermal paste, but from Very Authoritative People, the TIM is not the problem. As has been posited across the net when people de-lid Ivy Bridge chips with great results, it’s really the black adhesive that’s the culprit. When you cut out that adhesive, it allows the IHS to sit closer to the CPU die, meaning there is less thermal paste through which the heat has to travel, leading to significantly lower temperatures. Intel’s TIM is really quite good, but the manufacturing process leads to that glue being just a little too thick, which is why you see such temperatures.

Source:
http://www.overclockers.com/3step-guide ... el-haswell

ULYXX · Thu Jun 06, 2013 3:14 pm

Thank you for sharing this with us chuckula.

Hehe

Airmantharp · Thu Jun 06, 2013 3:19 pm

It's refreshing to know that people are quickly identifying conventional methods at dealing with Haswell's increased thermal output at higher clock speeds. By all accounts it should be good for 5.0GHz+ with good air or an integrated water-cooler, and that's definitely something I'd like to see as games are already bouncing up against the 'available' CPU power from any overclocked CPU on the market.

elmopuddy · Thu Jun 06, 2013 4:33 pm

You sir have a set of big ones

Great pics.

Waco · Fri Jun 07, 2013 3:20 pm

Whenever our second 4770K arrives I'll figure out which one runs hotter and pop the lid.

I've been itching to do it for years since the wild success of my IHS removal back on my old Opteron.

chuckula · Fri Jun 07, 2013 4:45 pm

Good news! Even though the first motherboard died, the delidded 4770k is still alive and runnning in a new mobo! More info to come over the weekend.

Ditiris · Fri Jun 07, 2013 5:17 pm

Excellent! Glad to hear it didn't take the chip with it. You've got me a bit nervous now about installing the X60 though. I think I'll finally have enough time to put everything together tomorrow.

vargis14 · Sat Jun 08, 2013 6:13 am

Chuck!!!!! You should use Coollaboratorys liquid metal instead of past between the CPU and the IHS. It is the closest thing to solder.

Whatever you do just do not get it on anything aluminum. It's will literally destroy aluminum.

Sat Jun 08, 2013 10:07 am

vargis14 wrote:
Chuck!!!!! You should use Coollaboratorys liquid metal instead of past between the CPU and the IHS. It is the closet thing to solder.

Whatever you do just do not get it on anything aluminum. It's will literally destroy aluminum.

That would be the gallium mentioned in the MSDS.

Sat Jun 08, 2013 10:10 am

Just remember Bill Paxton (or is that Chuckula?) holding the smoking wreckage of a Z87 motherboard in that second video. :wink:

chuckula · Sat Jun 08, 2013 10:37 am

vargis14 wrote:
Chuck!!!!! You should use Coollaboratorys liquid metal instead of past between the CPU and the IHS. It is the closet thing to solder.

Whatever you do just do not get it on anything aluminum. It's will literally destroy aluminum.

I have heard about that stuff (including the aluminum warning). I'm both fascinated and terrified by it, but I might give it a shot in the future.

chuckula · Sat Jun 08, 2013 1:52 pm

OK! Initial hardware build is complete and Thunderbirds are GO!

I have updated the Photo set to include pictures of the full build.

A few notes from the photos:
1. You'll note that the SSDs are installed on the back-side of the case. I have them held in place Scotch Heavy Duty fastener strips (think industrial strength velcro). That's why the bottom drive cages in my Fractal Define XL R2 are missing, and it makes room for the 280mm radiator that isn't officially supported by this case :-)

2. I was a little leery of the Kraken pump, but it appears to work fine. As a precaution, I took the three-pin header from the pump and plugged it directly to a molex connector from my PSU instead of into the motherboard. The roof, rear, and floor case fans are plugged into motherboard headers though.

3. You can't see it but the radiator is actually in a push-pull configuration with 140mm fans. The pictures show the rear pull fans, but there are two more fans in push mode in the front of the case.

4. The GPU is a Gigabyte GTX 770 if anyone is wondering.

Firestarter · Sat Jun 08, 2013 2:07 pm

26C at idle, nice!

Thanks for taking the dive and making pictures along the way!

chuckula · Sat Jun 08, 2013 2:11 pm

Firestarter wrote:
26C at idle, nice!

Thanks for taking the dive and making pictures along the way!

It is pretty nice. I'm not making any promises about load temperatures and I'm not promising miraculous overclocks, but it is a positive sign. (P.S. --> My man cave isn't anywhere near as cold as I'd like, so I'd say that 26 is probably only 2-3C above the ambient room temperature).

vargis14 · Sat Jun 08, 2013 2:27 pm

I like the placement of the radiator, especially since you did it on intake pushing and pulling air into the case. Helps the CPU stay cooler since in guarantees the coolest air possible and the air does not get heated much from the radiator at all. A tower air cooler say like a cooler master 212 throws some heat off big time like a gpu cooler.

Also the SSD placement is sweet.

Btw even though my 2600k rig is still awesome I have to say I am a bit jealous and happy for you. I would be less jealous if I had both those SSDs and your graphics card:)

chuckula · Sat Jun 08, 2013 2:56 pm

vargis14 wrote:
I like the placement of the radiator, especially since you did it on intake pushing and pulling air into the case. Helps the CPU stay cooler since in guarantees the coolest air possible and the air does not get heated much from the radiator at all. A tower air cooler say like a cooler master 212 throws some heat off big time like a gpu cooler.

Also the SSD placement is sweet.

Btw even though my 2600k rig is still awesome I have to say I am a bit jealous and happy for you. I would be less jealous if I had both those SSDs and your graphics card:)

If you knew what I was running now, you'd be less jealous and feel more pity for me!

This machine is replacing a Core 2 duo (not quad) system that I built in February of 2008 and has served me very well for over 5 years. Some people upgrade hardware all the time, whereas I do it much more infrequently. This system will probably last a good 5 years, especially with the relative slowdown in new CPU designs. I do intend to upgrade the video card and the storage over the lifespan of the machine. I think it will definitely be able to drive strong discrete GPUs for quite some time.

elmopuddy · Sat Jun 08, 2013 3:33 pm

Thanks for the nerd pr0n.. awesome stuff!

sschaem · Sat Jun 08, 2013 4:00 pm

It it possible to place the heatsink directly on the chip without the protecting lid ?
Or its to much risk from the pressure and it would crack ?

The people I seen do this just replace the thermal paste with a more conductive one, and they seem to get ~20c drop in temp reading.
Direct contact seem like it would do much better...

I wonder if its the voltage regulator that are the real problem ?

derFunkenstein · Sat Jun 08, 2013 4:06 pm

So here's a dumb question - do you put the cap back on without the adhesive layer so that it's all closer together, or do you just apply your heatsink/water block to the naked chip?

chuckula · Sat Jun 08, 2013 4:21 pm

derFunkenstein wrote:
So here's a dumb question - do you put the cap back on without the adhesive layer so that it's all closer together, or do you just apply your heatsink/water block to the naked chip?

Not a dumb question! I delidded, cleaned off the CPU PCB and the IHS, then put the IHS back on (with a very very small dot of AS5 between the die and the IHS). So going from bottom to top it's: Die --> uber-thin-TIM --> IHS --> More TIM on top of IHS --> Water block/heatsink. I guess it's possible to put a heatsink on the bare die, but I'm not that insane

P.S. --> I have a very minimal Linux install running and the newest version of the mprime torture test (which includes AVX optimizations) is pushing all 4 cores to about 60C at stock... I'm not sure if that's good or bad, but I likely have some headroom for overclocking. I think I can do 4.6GHz as a 24/7 type overclock. I'll post updates!

Waco · Sat Jun 08, 2013 5:16 pm

AS5 isn't the best choice for under the heatspreader IMO. I have some Liquid Ultra on the way to put under mine...apparently most pastes will either dry out or pump out over time and your temps will skyrocket.

chuckula · Sat Jun 08, 2013 6:14 pm

Waco wrote:
AS5 isn't the best choice for under the heatspreader IMO. I have some Liquid Ultra on the way to put under mine...apparently most pastes will either dry out or pump out over time and your temps will skyrocket.

I'm going to be trying some liquid Ultra down the line. That stuff is scary, but it appears to be the best think short of a real solder joint.

vargis14 · Sat Jun 08, 2013 6:17 pm

chuckula wrote:

If you knew what I was running now, you'd be less jealous and feel more pity for me! This machine is replacing a Core 2 duo (not quad) system that I built in February of 2008 and has served me very well for over 5 years. Some people upgrade hardware all the time, whereas I do it much more infrequently. This system will probably last a good 5 years, especially with the relative slowdown in new CPU designs. I do intend to upgrade the video card and the storage over the lifespan of the machine. I think it will definitely be able to drive strong discrete GPUs for quite some time.

I hope to also get another 5 years out of my rig also, on top of the 2+ years it has been running like a champ. I am getting to the point that i might even change the TIM between my AIO water loop now. Actually i think i am just going to get a new one in the next 6 months and replace it all together since the pump has over 2 years of 24/7 use on it. plus i want a 240mm rad anyways since it will fit so nicely in the top of my HAF922 case. I hope my 560TI SLI setup will live on for another year at least. I am hoping the next AMD graphics generation year from now will be awesome.

The 1st thing you have to do is change your signature now since you now are in the top 5% bud

Then order the Coollaboratory Ultra Liquid Metal stuff. I have watch videos on you tube showing application and it looks very easy to work with for using it between the IHS and CPU die. Heck I would probably use it between the IHS and the Pump also since you will have plenty left and your AIO cooler should have a copper coldplate along with nickle coated recessed screws. So you wont have to worry about re-seating your cooler since metal does not dry out that i know of. Then if you happen to have any Gallium left over put a drop on the top of a used soda can and watch the top disintegrate.

chuckula · Sat Jun 08, 2013 7:41 pm

vargis14 wrote:
chuckula wrote:
The 1st thing you have to do is change your signature now since you now are in the top 5% bud Then order the Coollaboratory Ultra Liquid Metal stuff. I have watch videos on you tube showing application and it looks very easy to work with for using it between the IHS and CPU die. Heck I would probably use it between the IHS and the Pump also since you will have plenty left and your AIO cooler should have a copper coldplate along with nickle coated recessed screws. So you wont have to worry about re-seating your cooler since metal does not dry out that i know of. Then if you happen to have any Gallium left over put a drop on the top of a used soda can and watch the top disintegrate.

You have an excellent point about my sig... time for another update! Based on your advice and comments from several other posters, my next mini-project is to get some of that Coolaboratory Ultra stuff and use it between the die & IHS. I'm getting decent-but-not-spectacular overclocks right (4.6GHz, mprime stable) now just doing my initial run and now I want to get to the next level.. or just get better temperatures with a sane overclock :-)

BIF · Sun Jun 09, 2013 2:07 pm

Everytime I see this thread in the "Hot Forum Threads", it reads to my eyes "I just Diddled a 4770k..." The first couple times, I would think "WTF!" but now I just smile.

ColeLT1 · Mon Jun 10, 2013 9:16 am

derFunkenstein wrote:
So here's a dumb question - do you put the cap back on without the adhesive layer so that it's all closer together, or do you just apply your heatsink/water block to the naked chip?

Putting the heatsink/waterblock directly to the silicon would be the best way without a doubt, but there is one problem, the heat spreader has 2 tabs on each side, and this is what the socket retention mechanism grabs to push the chip down onto the board. It would take a lot of pressure from your waterblock, pushing directly on the now exposed core to get the LGA pins pads to press hard enough to make good contact, I have seen it done, but it is not suggested.

Hey chuckula, the 4670k I have at the house is going strong as of this morning (14hr prime) @ 4.6GHz set at 1.290v. Temps are indeed hot, as of this morning, max core-temp was 82c recoded, but the current prime temps were in the low 70s, I am using this method ( http://www.overclockers.com/3step-guide ... el-haswell ) of setting the voltage, then finding the voltage/speed I want, then changing it to adaptive voltage. If 1.290v is stable for 24+hrs, I will set adaptive voltage to 1.3v. Their site says set to 1.25v and 4.6ghz to see where you are on the bell curve, if it boots you have a upper 50% chip, mine booted but failed prime, then 1.275v failed after 3 hours. If it fails prime by the time I get home tonight, the temp is too high for more voltage (imo) and it will be a 4.5GHz final clock with lower temps.

chuckula · Thu Jun 13, 2013 9:36 pm

OK quick overclocking update: I got the Coollaboratory Ultra paste and applied it both to the die-IHS and to the top of the IHS as well... lemme just say that you guys were TOTALLY RIGHT to suggest this stuff. Edit: Screenshots of a few sections of the test begin here: http://www.flickr.com/photos/97134463@N07/9039248034/

Before: using AS5 as the TIM, at 4.6 GHz with a 1.1 Vcore + 0.18V offset in Adaptive mode (meaning 1.28 max although adaptive voltage can exceed that level):

4.6Ghz: Using the 8K strides in the 64-bit Linux mprime torture test, I was hitting temps in the mid-90s(!). The test would actually pass although mprime did crash after about an hour on one of the other shorter-stride tests. From what I've seen with mprime, the short-length tests are much more grueling on the CPU than the long-length tests.

After: Using the Coollaboratory Ultra and the same voltage settings... with a higher OC to 4.7 GHz: Absolute highest temperature I saw was 82C (briefly) for my hottest core in the 8K section of the torture test. I ran the torture test for over 2 hours and the hottest core only briefly hit 80 - 81C in the other tests. In the long-length tests none of the cores broke 70C (hottest core was usually around 67C). I finished the test after over 2 hours, machine was 100% stable and I'll be posting some screenshots soon.

Incidentally, ambient temperature was about 23C during these tests, so if I had a room cooled to my liking at about 20C, I doubt that I would have broken 80C even with the hottest core.

Frankly, I think I could even back off the voltage a little bit and maintain stability at 4.7GHz while lowering the temps a bit... I may give that a whirl. For pure benchmarking purposes I'm extremely confident I could get this bad-boy up to at least 4.8GHz, with 5GHz probably being a possibility for semi-stable benchmarking. I'm going to be using 4.7GHz as my daily-use overclock though, and I think the delid was extremely helpful in getting more reasonable temps with solid stability. Remember too that I'm running 32GB of RAM at DDR3-2133, and that is probably not a super-friendly RAM configuration for overclocking.

Haswell definitely is not a cool-running chip, but frankly if Intel had done its job right and soldered the IHS to the die, there would probably be a whole bunch of 5GHz overclocks that are even stable for daily use. So in a nutshell: Haswell isn't an overclocking disaster, but if Intel would just do a few simple things that don't require billions of dollars in R&D investment, they could really improve the situation for enthusiasts. Are you listening Intel marketing? (I doubt it)

Shout out to ColeLT1 too! Lemme know how those OCs go for you!

kumori · Thu Jun 13, 2013 11:33 pm

ColeLT1 wrote:
Chuckula, I read an interesting article, and trying my best to find it again, that they tested all forms of different TIMs, and the intel's ranked up with the best. They posted temps and results, then culprit was not the TIM, but the black epoxy gluing the IHS to the board creates too large of a gap between the IHS and core. They removed all the epoxy and put the IHS back on without touching the TIM, then removed the IHS and tested with high quality TIMs between the IHS and core, and saw equal temps to high quailty TIM, and much lower temps than stock.

Granted, solder would be better than any TIM.

I think this is what he was referencing.

http://forums.anandtech.com/showpost.ph ... tcount=570

vargis14 · Fri Jun 14, 2013 2:39 am

Grats chuck! If you want it to last 5 years it would probably be a good idea to run her at 4.5 ghz and have a lower temp and voltage for 24/7. But if you need that 200+ mhz 24/7 by god its your rig and your under 1.3v.

I also have those cougar fans...great build quality, I dont know about lasting 300,000 hrs but they are nice

TR Forums

I just Delidded a 4770K (and lived to tell the tale!)

Re: I just Delidded a 4770K!

Re: I just Delidded a 4770K!

Re: I just Delidded a 4770K!

Re: I just Delidded a 4770K!

Re: I just Delidded a 4770K!

Re: I just Delidded a 4770K!

Re: I just Delidded a 4770K!

Re: I just Delidded a 4770K!

Re: I just Delidded a 4770K!

Re: I just Delidded a 4770K!

Re: I just Delidded a 4770K!

Re: I just Delidded a 4770K!

Re: I just Delidded a 4770K!

Re: I just Delidded a 4770K (and lived to tell the tale!)

Re: I just Delidded a 4770K (and lived to tell the tale!)

Re: I just Delidded a 4770K (and lived to tell the tale!)

Re: I just Delidded a 4770K (and lived to tell the tale!)

Re: I just Delidded a 4770K (and lived to tell the tale!)

Re: I just Delidded a 4770K (and lived to tell the tale!)

Re: I just Delidded a 4770K (and lived to tell the tale!)

Re: I just Delidded a 4770K (and lived to tell the tale!)

Re: I just Delidded a 4770K (and lived to tell the tale!)

Re: I just Delidded a 4770K (and lived to tell the tale!)

Re: I just Delidded a 4770K (and lived to tell the tale!)

Re: I just Delidded a 4770K (and lived to tell the tale!)

Re: I just Delidded a 4770K (and lived to tell the tale!)

Re: I just Delidded a 4770K (and lived to tell the tale!)

Re: I just Delidded a 4770K (and lived to tell the tale!)

Re: I just Delidded a 4770K!

Re: I just Delidded a 4770K (and lived to tell the tale!)

Who is online