Haswell Delidded

Re putting the lid back on:

- First, read a lot of the howto's
- Check the lid fit on top of the bare die. See if it'll rest on the die. (Use some thermal grease and see if it transfers.)
- With the lid on the die, look at the gap, if any, between the lid and the rest of the chip's pcb.
- Put a putty-like bead around the lid and the pcb
- LEAVE AN AIR GAP at least at one point around the lid perimeter. (As the die heats up, you don't want the lid to pressurize. It'll bulge. Or try to.)

Just my .02. (I'm a month or so out from buying and delidding a 4770k, so the above points are only from what I've read, not what I've done. Yet.)

Ken
 
Based on how popular it is to do this on Ivy Bridge and Haswell, I am surprised a company hasn't started selling aftermarket shims that fit around the core and the capacitors.
 
Based on how popular it is to do this on Ivy Bridge and Haswell, I am surprised a company hasn't started selling aftermarket shims that fit around the core and the capacitors.

Enthusiast crowd is a small percentage of the market, couple that with an even smaller amount of people actually delidding their CPUs. Wouldn't be much profit for such a small segment of the marketplace that would actually pay something that isn't seen as a necessity.
 
does normal cpu coolers not work after delidding? sorry if i missed something..
 
does normal cpu coolers not work after delidding? sorry if i missed something..



If you use the vice method and are brutal about it you run the risk of damaging the alignment of the IHS. After you delid you can put the IHS back on and carefully mount any cpu block as long as you have the IHS in place.


I prefer direct die no IHS mount which works best with light weight water block IMO.
 
SonDa5, did the microcode update help you get anything further out of your chip?




I haven't had alot of time to play with it but it feels smooth. Running a new beta Xtreme BIOS designed for liquid nitrogen overclocking, seems to be geared to pushing memory with unlock voltages that help memory. I focused on memory and was able to get 2600mhz CL9 from my Gskill 2400mhz CL9 kit. With my Z77 Mpower and 3770k I could never get this kind of performance from this kit.

10microcode.jpg



4.7GHZ, 2600mhz CL9.

My system setting freaked out CPU-Z validation: http://valid.canardpc.com/2873741
fastmem.jpg


Not completely stable yet though but much more stable then anything I have done up until this point. Still tweaking.

Also I added the chipset and VRM water blocks to my MB.



Z87X-OCblocksonline.jpg
 
I delid 4770k tonight I when from 91c to 68c using the same 4.5ghz 1.2v however core 2 is 10c hotter then all the other cores anyone know why?
 
I delid 4770k tonight I when from 91c to 68c using the same 4.5ghz 1.2v however core 2 is 10c hotter then all the other cores anyone know why?

Which paste did you use? It's heavily recommended that you use CLP or CLU, if you haven't.

If you did use the latter, make sure you didn't add too much of it.
 
I delid 4770k tonight I when from 91c to 68c using the same 4.5ghz 1.2v however core 2 is 10c hotter then all the other cores anyone know why?

There will always be one core that is hotter then the other 3..It is the same on IB..That core is simply in a hotter area of the die..I wouldn't stress about it, as long as you are stable..
 
Which paste did you use? It's heavily recommended that you use CLP or CLU, if you haven't.

If you did use the latter, make sure you didn't add too much of it.

I did CLU on die to lid and cpu to block. Was a pain in the ass. lol
 
I did CLU on die to lid and cpu to block. Was a pain in the ass. lol

Can you elaborate? I'm toying around with de-lidding my 3770 k and have some liquid pro already that I'd be using. Curious what I'm in for If I go for it.
 
Can you elaborate? I'm toying around with de-lidding my 3770 k and have some liquid pro already that I'd be using. Curious what I'm in for If I go for it.

Last time I delid a cpu was back in the AMD x2 days which was pretty easy in comparison. The glue on the 4770k was much harder using a razor blade but once you get under the lid its a bit easier.
I suggest that you don't start on the side of the arrow and the big . when you start cutting. I started on a corner and move the blade around the cpu keeping the blade off as much as I could. The problem I had with CLU was I put too much so I had a bit of a hard time getting it off without wiping it off and starting over.

As long as your temps are in check, I wouldn't worry about it.

They are much better for sure, Thanks.
 
Can you elaborate? I'm toying around with de-lidding my 3770 k and have some liquid pro already that I'd be using. Curious what I'm in for If I go for it.

If you aren't super steady handed, I would suggest you use a razor blade and a small hammer that you use to make light, steady pressure on the glue..

It is quite easy to control the amount of force you apply this way, and greatly reduces the chances of you pushing through the glue and digging into the PCB..
 

Why are you worried about pushing the ram speed in lieu of the cpu clock? You know that memory speeds do very little with SB and newer cpus, especially since you already have a premium rated 2400Mhz Kit..I would much rather run with slower ram speeds if it means you can eek out another 100-200mhz outta the cpu!

I gotta say, when I first saw that board, it really looked ugly to me based on the stock pictures you see at e-tailers, but looking at yours it looks much better then the stock photos..I am still not a huge fan of the orange, but if the orange was changed to white...:cool:

Where did you get your other blocks from? They fit nicely with the DT cpu block..I wish you had been able to take before and after VRM temps..I am thinking of getting a set for Extreme 4 from a German company I found..I am sure they would help since I am planning to push past 5Ghz for a 24/7 daily o/c as soon as I get my second 360mm radiator installed..
 
CCityInstaller I ordered the block set from EK direct. It's a new kit that wasn't available any where except direct from EK.


FrozenCPU is selling them now. http://www.frozencpu.com/products/2...t_-_Acetal_EK-FB_KIT_GA_Z87X-OC_-_Acetal.html

I didnt care for the Orange in the mb color either but it is growing on me. I'm probably going to get some bright orange tubing... :D

I will have to ask around about the Z87 chipset temps and VRM temps on my MB to see how much the blocks help cool down stuff.

I think Gigabyte still has alot of work left to get the BIOS for my MB worked out.

4.7GHZ with CL9 2600mhz is a blast for gaming and some benchmarks.


I'm working on stable 5GHZ with 2400mhz Cl9.
 
CCityInstaller I ordered the block set from EK direct. It's a new kit that wasn't available any where except direct from EK.


FrozenCPU is selling them now. http://www.frozencpu.com/products/2...t_-_Acetal_EK-FB_KIT_GA_Z87X-OC_-_Acetal.html

I didnt care for the Orange in the mb color either but it is growing on me. I'm probably going to get some bright orange tubing... :D

I will have to ask around about the Z87 chipset temps and VRM temps on my MB to see how much the blocks help cool down stuff.

I think Gigabyte still has alot of work left to get the BIOS for my MB worked out.

4.7GHZ with CL9 2600mhz is a blast for gaming and some benchmarks.


I'm working on stable 5GHZ with 2400mhz Cl9.

Now that is what I wanna see bro! 5Ghz stable with super fast ram will be sweet..That is a ~5.2Ghz IB or a ~5.4-5.5Ghz SB:eek::eek::eek:!!

Instead of bright orange tubing, I would go with some white, and do white fans (you can easily paint the ones you have now)..It would be an awesome cream-sickle style...Talk about sweet :cool:!!
 
it's bare silicone. It will chip easily. stay away if you're clumsy.

The old bare die chips were often epoxy coated which really helps to protect them. But they didn't have near the W/m^2 output so I prefer the current arrangement (despite people who think intel's solution was "greedy," I call it a sound business solution that still benefits enthusiasts.)

Right but back in the day we installed Heatsinks with by pressing down on a small tab (STAB AT THE MB with a screwdriver, opps I mean heat sink tab lol) with a lot of force and hook the SOB onto the socket, hopefully without slipping. With newer mounting systems (especially for water blocks) it doesn't seem like it would be nearly as much of an issue. but I have never delidded, I do however think I may be deliding my future haswell.

Now I just need to figure out which waterblock will be best for a delided chip.
 
After reading this thread I see there are 2 methods of mounting the CPU after delidding, With and without the IHS back on top of the DIE.

Questions:

  1. How much (if any) better are the temperatures when not using the IHS?
  2. Will I be able to mount my Corsair H60 directly on top of the DIE or there will be a gap between CPU and block?
 
Now that is what I wanna see bro! 5Ghz stable with super fast ram will be sweet..That is a ~5.2Ghz IB or a ~5.4-5.5Ghz SB:eek::eek::eek:!!

Instead of bright orange tubing, I would go with some white, and do white fans (you can easily paint the ones you have now)..It would be an awesome cream-sickle style...Talk about sweet :cool:!!



I may end up swapping mother boards. I think my Z87X-OC has bad VRM design that is hurting my over clocking.
 
I may end up swapping mother boards. I think my Z87X-OC has bad VRM design that is hurting my over clocking.

That is interesting. I had a vi hero (since sold), since I'm playing around with a ud5h I got for cheap. I do plan on picking up another hero, to use for test benching purposes.

I can't say I'm terrible impressed with haswell, at the moment. I've yet to find that one good chip.
 
After reading this thread I see there are 2 methods of mounting the CPU after delidding, With and without the IHS back on top of the DIE.

Questions:

  1. How much (if any) better are the temperatures when not using the IHS?
  2. Will I be able to mount my Corsair H60 directly on top of the DIE or there will be a gap between CPU and block?

Personally I wouldn't mount an H60, or any other AIW WC that has the pump mounted to the cpu block, directly on die..There is a *Chance* the vibration could cause an issue without the IHS in place..This is actually good news for users like you, since you do not need to worry about adjusting for the gap difference..

Just carefully follow the de-lidding advice, put some CL Liquid Ultra on your die. I recommend:

1. Cover the exposed gold contacts next to the die like SonDa5 did with a dielectric coating or nail polish if you are cheap.

2. Heating it up a tiny bit of CL LU with a hair dryer and smoothing it out in a fine layer with an old CC or plastic spreader. If you "tear" the layer before it is smooth all the way over the die, then it's not hot enough, so heat it back up, and repeat..

3. Carefully place the chip back in the socket, place the IHS in place, and hold it with your finger while you the socket retention bracket to latch it in place..

4. Now, use something like MX-4, AS-5, etc on the IHS, and mount your block like normal.

5. Sit back, and enjoy that extra 2-300mhz+ (and insanely lower temps!) you should be able to net since you are no longer thermally limited on your o/c!

I may end up swapping mother boards. I think my Z87X-OC has bad VRM design that is hurting my over clocking.

What makes you think that? I am just curious, since I clearly haven't used one, but it is their flagship OC'ing board no? Also, Sinhardware gives their component choice high scores.
 
What makes you think that? I am just curious, since I clearly haven't used one, but it is their flagship OC'ing board no? Also, Sinhardware gives their component choice high scores.



There are 2 models of the Gigabyte Z87X-OC. One is the cheaper OC and the other is the FORCE which is double in price. The Force has many more features and the OC has very few features but does have good over clocking control and BIOS support.


Now the reason I am displeased with the Z87X-OC is because when I bought it I thought it had the same high quality VRM parts found on the Force model. I knew that the OC had half of the number of VRM phases and I was ok with that because I thought they were high quality VRM parts and the Force seemed over kill for my needs and the price was out of my budget.

So back to the VRMs on the Z87X-OC. The VRM parts for the CPU power are IR3553M.

8Phaselowres.jpg



That CPU power 8 phase VRMs are powered by IR 3553M which is the same VRM found in much cheaper non OC series mother boards like the Z87-UD3H. From what I have researched online the IR 3553M is designed more for a server system not an over clocked gaming RIG and has max current rating of 40A. The guy from Sins hardware posted elsewhere that the chokes on the Z87X-OC are rated for 60 amps and I'm not sure if that is true but if the VRMs are rated for 40amps in series with the 60 amp chokes there is a design flaw since 60amps is way too high to be matched with the 40amp VRM part. Sin stated that it would help clean up the power but to me it seems like Gigabyte just threw the cheaper IR3553m VRM part on the OC to save money.


With only 8 phases for cpu I think Gigabyte cut costs big time here. I got suckered into the "OC" hype of this MB and was sold on this IR power unit which from what I am learning is different than what many online reviews have stated. Considering the VRM is much weaker than the FORCE and only has half the numbers of phases for CPU.

The FORCE uses 16 phases for CPU power and each of them are IR3550 parts which are listed from manufacturer to be designed for over clocked gaming systems. Sin has stated the IR3550 part is much more expensive.

The OC deserved at least 8 phases with top quality IR3550 IMO. I feel like Gigabyte cheapened down the build quality of the OC to save money in production and then has spent alot of time marketing it as a "OC" level mother board just to make money.

The FORCE has many other features besides the high quality VRM parts that make it worth at least double the price of the OC.
 
Last edited:
So do you plan on picking up the force? Very pricey for an lga1150 board, but it does look fantastic.
 
Personally I wouldn't mount an H60, or any other AIW WC that has the pump mounted to the cpu block, directly on die..There is a *Chance* the vibration could cause an issue without the IHS in place..This is actually good news for users like you, since you do not need to worry about adjusting for the gap difference..

Just carefully follow the de-lidding advice, put some CL Liquid Ultra on your die. I recommend:

1. Cover the exposed gold contacts next to the die like SonDa5 did with a dielectric coating or nail polish if you are cheap.

2. Heating it up a tiny bit of CL LU with a hair dryer and smoothing it out in a fine layer with an old CC or plastic spreader. If you "tear" the layer before it is smooth all the way over the die, then it's not hot enough, so heat it back up, and repeat..

3. Carefully place the chip back in the socket, place the IHS in place, and hold it with your finger while you the socket retention bracket to latch it in place..

4. Now, use something like MX-4, AS-5, etc on the IHS, and mount your block like normal.

5. Sit back, and enjoy that extra 2-300mhz+ (and insanely lower temps!) you should be able to net since you are no longer thermally limited on your o/c!

What makes you think that? I am just curious, since I clearly haven't used one, but it is their flagship OC'ing board no? Also, Sinhardware gives their component choice high scores.

After my parts arrive, before dellidding anything I will first check how much difference OC is going to make in terms of 3d rendering, if real life figures will be worth it then I am going for it.

Just curious why do you propose to use different TIM on top of the lid? If CLU is so good.
 
So do you plan on picking up the force? Very pricey for an lga1150 board, but it does look fantastic.


Funny you ask... I do like the mb alot but its still out of my price range and I was just going to settle with the Z87X-OC until about 20 minutes ago when I ran into a deal on a MSI Z87 X Power... It sells at newegg for $430 at Newegg right now. http://www.newegg.com/Product/Product.aspx?Item=N82E16813130689


I just ordered 2 of them from TigerDirect for $159.99 each!!! HAHAHAHAHAHAHA Please keep this on the low down till mine ship. I ordered 2 day delivery so I hope to get them in a few days!! :D


Tigerdirect link in case anyone is interested. Keep it on the low down. D

http://www.tigerdirect.com/applications/SearchTools/item-details.asp?EdpNo=8450495&CatId=13

I'm so thankful for this deal. :)
 
Funny you ask... I do like the mb alot but its still out of my price range and I was just going to settle with the Z87X-OC until about 20 minutes ago when I ran into a deal on a MSI Z87 X Power... It sells at newegg for $430 at Newegg right now. http://www.newegg.com/Product/Product.aspx?Item=N82E16813130689


I just ordered 2 of them from TigerDirect for $159.99 each!!! HAHAHAHAHAHAHA Please keep this on the low down till mine ship. I ordered 2 day delivery so I hope to get them in a few days!! :D


Tigerdirect link in case anyone is interested. Keep it on the low down. D

http://www.tigerdirect.com/applications/SearchTools/item-details.asp?EdpNo=8450495&CatId=13

I'm so thankful for this deal. :)

Checked the model number of that link: Z87I

So, don't get too upset if they ship you this:
http://us.msi.com/product/mb/Z87I.html
 
Thanks. I didn't see that very minor yet important detail. Looks like the whole deal could be a mis print. I still hope they ship me what I want but I will call them to verify and if necessary cancel order. Thanks.
 
You could always return it, if the board ends up being the mini itx msi board. An x-power for $157, would be a mega steal if true.

Back on the gigabyte-oc, I was always under the impression that board used the IR3550. Guess not.
 
tonyftw said:
Back on the gigabyte-oc, I was always under the impression that board used the IR3550. Guess not.

Me too. Its hard to find any reviews right now that mention the exact part used on it.

Makes me think Gigabyte deliberately asked he reviewers to not list such information.

It is by no means any where near the level of the Force. Z87X-OC VRM potential out put is the same as the UD3H because of the same count and part used on the VRMs.

Not saying its a bad part but I really think Gigabyte cut corners with it hoping to make max money with it.

I like the MB alot but just not pleased with this major detail.
 
Last edited:
That CPU power 8 phase VRMs are powered by IR 3553M which is the same VRM found in much cheaper non OC series mother boards like the Z87-UD3H. From what I have researched online the IR 3553M is designed more for a server system not an over clocked gaming RIG and has max current rating of 40A.

The IR3553 is just the 40A version of the IR3550, which is a 60A part. The IR3553 has a smaller footprint as well, but is still pin-compatible with the IR3550 footprint. That's why you see those extra pins on the motherboard next to each of the power stage ICs.

In reality, the 40A rating of the part probably isn't going to hold you back. Remember that you have 8 phases using these parts, which is still a lot of power. Are your voltages actually sagging significantly under load?

However the 60A part will be more efficient across the range and therefore generate less heat.

The guy from Sins hardware posted elsewhere that the chokes on the Z87X-OC are rated for 60 amps and I'm not sure if that is true but if the VRMs are rated for 40amps in series with the 60 amp chokes there is a design flaw since 60amps is way too high to be matched with the 40amp VRM part. Sin stated that it would help clean up the power but to me it seems like Gigabyte just threw the cheaper IR3553m VRM part on the OC to save money.

Definitely not a design flaw. Inductor ratings are upper limits, and all things equal you always want the highest current rating for the inductor. A higher current rating usually comes with thicker internal conductors, which results in lower electrical resistance and higher efficiency. The current rating doesn't change how much it 'cleans up the power'.

But yes, the cheaper IR3553 power stage was used because it's cheaper. My supplier shows about a $1 difference between the two parts. That's probably $0.25 per part or less at the quantities motherboard manufacturers are buying. That probably doesn't sound like much, but for a part that appears 8 times on a board, that's an absolutely huge cost delta in the world of electronics manufacturing, especially for a part that is 'good enough'.
 
I may end up swapping mother boards. I think my Z87X-OC has bad VRM design that is hurting my over clocking.

I'm curious to see if this makes any difference for you. I haven't seen any symptoms of power regulation issues on my 6 phase asrock extreme 4. My understanding is that all the stress is on the FIVR. I haven't even needed active cooling on the VRM. This is the first time I haven't used a fan blowing around the socket area while watercooling since Barton.

I'll need to remove the cap on my 4670k because it needs 1.36v for 4.5. I'm pretty sure I have the worst Haswell on record. How are you guys removing the adhesive after the lid is off? Does it just peel off or can I use goo gone?
 
I'm curious to see if this makes any difference for you. I haven't seen any symptoms of power regulation issues on my 6 phase asrock extreme 4. My understanding is that all the stress is on the FIVR. I haven't even needed active cooling on the VRM. This is the first time I haven't used a fan blowing around the socket area while watercooling since Barton.

I'll need to remove the cap on my 4670k because it needs 1.36v for 4.5. I'm pretty sure I have the worst Haswell on record....


I'd like to see how your chip performs in a MB with more VRM CPU phases and better quality VRM before blaming the CPU as being the worst.
 
Back
Top