Ivy Bridge 3770k testing checklist - help me to not forget anything

[Idontcare]

Ok folks, I finally got my IB 3770k set up and functioning

It went less smooth than anticipated, lost a day trying to figure out why my MIVE-Z would not recognize/accept the latest bios build for the board (lesson learned - update every step of a bios, do not skip any bios builds when upgrading from an old bios to the latest bios) and then lost yet another day after that trying to debug why the system crapped out as soon as I installed Afterburner (which then became an issue with the Nvidia drivers, and ultimately came to realize I had to do a total clean fresh install because it was the Intel chipset drivers causing havoc on everything).

Anyways, that's all behind me now, so now I/we can get to the fun stuff.

Now I have every intention of eventually delidding this IB, and along the way I intend to do a complete analysis of the temperature/power/clockspeed profile for my 3770k as I did for my 2600k in this thread and this thread.

Hardware-wise we have the exact same hardware for the 3770k as I did for my 2600k. In fact when I was done testing my 2600k for the above threads I left the entire system in place, turned off and mothballed, in anticipation of re-using everything as a bonafide apples-to-apples comparison between IB and SB.

Now my concerns are that I will forget to test something and by the time I realize it then it will be too late. For example once I lap this chip there is no going back to testing "stock" conditions, etc.

Right now I have the chip installed with stock everything (stock tim, stock HSF, stock settings in the BIOS, etc). I have my intended test to perform but in the meantime if any of you folks want to chime in with any tests you want me to do then please chime in and I will attempt to oblige all of them (if they are reasonable, not testing salted versus unsalted butter as a TIM, apologies in advance ).

TIM's I have in-house at the moment are AS5, Ceramique, some older style TX-2 stuff, NT-H1 (my preferred, so easy to get repeatable mounts and cleanup is a breeze), and Indigo xtreme.

I know a number of you prefer MX4 and the IC diamond stuff but I don't have that here. I can be convinced to get some MX4 if it really is all that different than NT-H1. addendum: screw that, just saw it was $10 shipped from newegg, ordered.

For HSF's I have the stock Intel HSF, my lapped NH-D14, and my lapped H100.

So...before I pull this stock HSF off and remove its stock TIM...any requests for temperature/clockspeed/power tests? I'm going to do a baseline, obviously, just to keep apples-to-apples comparisons as we lap and eventually delid this sucker, but help me make sure we don't forget something at this early stage in the game.

 

[nine9s]

Ok folks, I finally got my IB 3770k set up and functioning

It went less smooth than anticipated, lost a day trying to figure out why my MIVE-Z would not recognize/accept the latest bios build for the board (lesson learned - update every step of a bios, do not skip any bios builds when upgrading from an old bios to the latest bios) and then lost yet another day after that trying to debug why the system crapped out as soon as I installed Afterburner (which then became an issue with the Nvidia drivers, and ultimately came to realize I had to do a total clean fresh install because it was the Intel chipset drivers causing havoc on everything).

I just built a new system with a 3770K CPU and Asus P8Z77 V Pro motherboard. I installed the Intel Chipset drivers. Everything works fine. Should I be worried? What was the problem? Should I have not installed those drivers?

 

[Ferzerp]

Once you gather all your replaced-TIM data, can you sand down the IHS's supporting side so that it sits lower against the die like I've done (just be careful)?

I didn't get any before and after stats on mine for that, so I'm curious if it actually did anything.

 

[2is]

I just built a new system with a 3770K CPU and Asus P8Z77 V Pro motherboard. I installed the Intel Chipset drivers. Everything works fine. Should I be worried? What was the problem? Should I have not installed those drivers?

I have the same board using Intel drivers. No problems. You're fine.

 

[Don Karnage]

IDC let us know how you remove the IHS. I want to do it but just don't want ti f up.

 

[Ferzerp]

IDC let us know how you remove the IHS. I want to do it but just don't want ti f up.

It's pretty easy. Get a razor blade and be gentle. The advesive is a black rubbery adhesive that you can slice through.

The biggest tip is to take it slow, and never exert force that may get translated downwards in to your proc.

Know where the die is. (look at pictures and draw the outline on top of your IHS if you think it will help you)

Get some dental floss or the like to test your cut and make sure that all all the rubber is cut through. (make sure you can run the floss up against the die on all 4 sides). Once you do that, then you can start to wiggle the die around by hand. I never used anything to pry at all, just my fingers. The stock TIM will hold it on pretty tightly.


edit: Honestly, cleaning the rubber glue off after you've gotten the IHS free is much harder than removing the IHS itself. It won't dissolve in alcohol, and I didn't really know what acetone would do to the proc, though I think the IX solvent is just acetone. I ended up scraping it off *very* carefully with a soft plastic toothpick (like the kind that have dental floss on one end that are so soft that you can't really use as a toothpick) after *very very* carefully slicing off the thickest parts with a razor blade.

 

[Avalon]

IDC let us know how you remove the IHS. I want to do it but just don't want ti f up.

As Ferzerp stated, the biggest thing is to know where your die is. That way you mitigate the risk of cutting into it.

What I had the hardest time with was getting the razor blade under the IHS and to actually cut into the rubber bonding. It's such an insanely small gap that I had a hard time with my razor. What really helped me there was tapping the back of the razor blade with a screwdriver head. This gave me the leverage I needed to be able to continue slicing by hand.



@IDC - Don't forget when you de-lid, to save the Intel TIM under the IHS! I know a few people were interested in how it performed.

 

[Idontcare]

I just built a new system with a 3770K CPU and Asus P8Z77 V Pro motherboard. I installed the Intel Chipset drivers. Everything works fine. Should I be worried? What was the problem? Should I have not installed those drivers?

I doubt you will have issues.

My problem was ASUS, not Intel per se, in that they only have the ancient Intel chipset drivers available on their website for downloading for my specific mobo (mive-z). So I foolishly grabbed those thinking "of course ASUS is keeping its driver-update page up to date for its premiere ROG mobo, right?" and didn't check to see if Intel had a newer version released.

And naturally that bit me in the butt. So I went to Intel and got their latest chipset driver for the Z68 and all was fine. It just took me forever and a day (well, mostly just day) troubleshooting the rig to finally realize it was the chipset driver that was the root of my issue.

1. yep, will definitely document with pics my IHS removal process. More info always helps.

2. Thanks for jogging my memory and reminding me to test the replacement TIM while keeping the stock IHS hieght before sanding down the edges to minimize the gap :thumbsup:

3. Save the stock under-IHS TIM for retesting as a HSF TIM to give some indication of its overall performance as a TIM.

 

[Bill Brasky]

@IDC - Don't forget when you de-lid, to save the Intel TIM under the IHS! I know a few people were interested in how it performed.

Yes! Yes! It might not be possible if the TIM is flaky like other posters were reporting, but some good testing on the stock TIM help a ton!

 

[MacGyverSG1]

Once you gather all your replaced-TIM data, can you sand down the IHS's supporting side so that it sits lower against the die like I've done (just be careful)?

I didn't get any before and after stats on mine for that, so I'm curious if it actually did anything.

Ferzerp, do you have any idea on how much of the IHS you removed (or how large the gap is)? When I delid my i7-3770K I definitely want to decrease the gap between the core and IHS.

 

[Ferzerp]

you will be able to tell by how thick the TIM is. I didn't really look closely.

It isn't big, but *any* gap is detrimental.

 

[Tempered81]

Request to lap the inside of the IHS and reapply it and test for results of lapped/unlapped die contact to copper. Second, lap the base of the IHS & use with thinnest possible tape to make tighter contact to cpu package substrate

 

[Xpage]

saw 1.5g of the IC diamond for sale on amazon for 5.20, shipped for $9.15 if you were interested.

$5.20 + $3.95 shipping
In Stock. Sold by Microbarn


I found that IC diamond was quite thick, i followed the protocol to boil it before applying, for longer than they recommend to get it to be more easily usable.

I'd be curious what the max clock would be when you undervolt it, along with power draw since that is supposedly what intel is gunning for. Like around 0.8v

 

[blastingcap]

Request to lap the inside of the IHS and reapply it and test for results of lapped/unlapped die contact to copper. Second, lap the base of the IHS & use with thinnest possible tape to make tighter contact to cpu package substrate

Good ideas, I second Temp's suggestions. Also, if you can save the Intel TIM scrapings, and test it alongside a few other TIMs (toothpaste not necessary) so we can get an idea of where in the TIM spectrum of goodness-to-badness that the Intel TIM is.

 

[Ferzerp]

Request to lap the inside of the IHS and reapply it and test for results of lapped/unlapped die contact to copper. Second, lap the base of the IHS & use with thinnest possible tape to make tighter contact to cpu package substrate

There is nothing to be gained from placing *anything* adhesive to hold the IHS on other than increasing the gap between the die and IHS

The TIM holds it on just fine (though it slips around). You just have to make sure to keep it lined up as you clamp down on it. Once that's done though, it's not moving.

 

[dma0991]

If you have a caliper or a screw gauge, measure the thickness of the CPU + IHS
1. Before doing anything
2. After lapping the IHS
3. After sanding the IHS base to reduce the gap
4. Measure how large is the gap between the IHS and die

 

[Idontcare]

If you have a caliper or a screw gauge, measure the thickness of the CPU + IHS
1. Before doing anything
2. After lapping the IHS
3. After sanding the IHS base to reduce the gap
4. Measure how large is the gap between the IHS and die

Sure, I already have a digital caliper (not the same model shown in the pic, but you get the idea) and it is supposedly accurate to +/- 0.01 mm

Request to lap the inside of the IHS and reapply it and test for results of lapped/unlapped die contact to copper. Second, lap the base of the IHS & use with thinnest possible tape to make tighter contact to cpu package substrate

I'll do it if I can figure out how.

The surface inside the IHS that you want me to lap is recessed from the protuding edges of the IHS that extend down. So lapping that recessed surface, and actually making it flat versus just making it polished and shiny, will be a real challenge.

Any ideas?

In the meantime I've been in troubleshooting hell with this system for the whole weekend. Could not get the system stable, thought it was my video card drivers, chipset drivers, windows updates...but nah, nothing makes an improvement.

The problem is that if I have a video card in slot1 on my MIVE-Z then it locks up, but as soon as I move that video card to any other slot (2, 3, or 4) then it works fine. The really weird part is the issue only surfaces when the IB CPU is installed. With my Sandy chip I can use the video card in any of the four slots.

So I don't know if my IB is borked (seems really really unlikely) or if my MIVE-Z has a bad PCIE slot (seem more likely, but then why does it work when I reinstall my 2600k?) or if this is some new incompatibility bug that ASUS needs to identify and fix with a BIOS rev.

 

[Ferzerp]

Any ideas?

You would have to make a little platform, somewhere around the size of the CPU die, and affix sandpaper to it. It would be exactly the height off the surrounding table that the height of the inside of the IHS is. As you removed material, you would have to make it higher. Perhaps spring-load it.

Maybe take pane of glass, cut a hole in it, and spring-load a tiny block that has sandpaper on it from down below?

 

[Ferzerp]

So I don't know if my IB is borked (seems really really unlikely) or if my MIVE-Z has a bad PCIE slot (seem more likely, but then why does it work when I reinstall my 2600k?) or if this is some new incompatibility bug that ASUS needs to identify and fix with a BIOS rev.

PCIe 3.0 vs 2.0 for the SB?

 

[Kenmitch]

So I don't know if my IB is borked (seems really really unlikely) or if my MIVE-Z has a bad PCIE slot (seem more likely, but then why does it work when I reinstall my 2600k?) or if this is some new incompatibility bug that ASUS needs to identify and fix with a BIOS rev.

Did you disable the onboard video? I know other members have had issues that seemed to be fixed by it. Might just be a temp thing that will be fixed with a later bios rev.

 

[Ferzerp]

Did you disable the onboard video.

You can't. You can only set one as primary.

 

[Kenmitch]

You can't. You can only set one as primary.

Don't feel like digging thru the threads. Was it Lucid hydra or something similar then?

 

[Ferzerp]

Don't feel like digging thru the threads. Was it Lucid hydra or something similar then?

Even without that stuff, there's no explicit setting to turn if off. Supposedly, it just powers down that section of the CPU if you're using a discrete card (unless you've installed the virtu crap).

 

[nyker96]

I am also curious how this process of removing IHS is for Ivy. I'm sure lots of people who own one is curious as well. If this process of removing IHS then reapply a new compound can be done safely, that would be good news for many.

I am also wondering if it is feasible just remove IHS then put the heatsink directly on die w/ some compound in the middle. would that yield better result? well eager to see how your project turn out.

 

[nitromullet]

Is it possible to run one of these cpu's without the IHS at all? I remember back in the original Athlon days we ran procs with no heat spreaders at all with no problem.

edit: for those of you interested in the removal process... http://www.youtube.com/watch?v=yCW508yl_Eg