Help with CPU overclocking on z77 OC formula

[parsec]

Re: Help with CPU overclocking on z77 OC formula

Thanks for the Linux info, I'll check it out.

Why Intel went with standard TIM instead of solder in IB CPUs is a good question. I wonder if the IB die could not survive being bathed in solder, which frankly surprises me that any CPU die could survive that. Or with all the RoHS standards, they can't use solder with lead in it, and the lead free solder melts at a higher temp, which is well known, and the die cannot survive that. Don'tcha love progress sometimes...

So are you talking removing the CPU heat spreader, and going naked die? That is hardcore, good luck with that. At least it will be easier to remove the IHS without the solder under it, much easier.

[profJim]

Re: Help with CPU overclocking on z77 OC formula

The web is full of details about users CAREFULLY removing the heat spreader and applying top quality TIM after they cleaned off the old TIM.
One of our forum members nuked his Ivy Bridge cpu when it wouldn't post after his TIM surgery session.

[synack]

Re: Help with CPU overclocking on z77 OC formula

Thanks for the Linux info, I'll check it out.

Why Intel went with standard TIM instead of solder in IB CPUs is a good question. I wonder if the IB die could not survive being bathed in solder, which frankly surprises me that any CPU die could survive that. Or with all the RoHS standards, they can't use solder with lead in it, and the lead free solder melts at a higher temp, which is well known, and the die cannot survive that. Don'tcha love progress sometimes...

So are you talking removing the CPU heat spreader, and going naked die? That is hardcore, good luck with that. At least it will be easier to remove the IHS without the solder under it, much easier.

I haven't yet decided if I will go naked die yet or not. But yes, removing the heat spreader for sure. Going naked die will involve having to disassemble the retention clip. The die alone doesn't come out far enough. I also don't want to worry about bending the CPU's PCB as the pressure will be solely on the middle of the chip.

For what is reported as *maybe* 2-3 degrees cooler going naked die, it's not worth the risk. I'd rather put on good quality TIM on and replace the IHS. Well, not worth it to me anyway. If I could find a suitable shim to put in it's place.. maybe... but there are people out there using just the HSF or water block to hold the processor in... yikes.
The biggest gain I have read, is in the replacement of the horrible quality TIM Intel chose to use.

I'm not normally a conspiracy believer.. but I'm beginning to wonder if Intel did do it on purpose. People who have done the TIM change are getting SB type temps and OC's. That makes for a boat load of performance over SB processors. Pretty much every review I read, the change meant at least +100MHz easy on lower voltage.
it just doesn't make sense... they ALREADY have the Fab process for solder for SB procs. The heat specs are very similar, why would solder on IB melt the core where SB would not? why switch to bad grease and silicone? if it's an issue with solder and temps... at least use a decent quality TIM for Pete's sake. From the vids I've seen, it looks like plaster flaking off the core.

Well I'll get a better sense of it tomorrow and will report my findings here.

Syn

[synack]

Re: Help with CPU overclocking on z77 OC formula

The web is full of details about users CAREFULLY removing the heat spreader and applying top quality TIM after they cleaned off the old TIM.
One of our forum members nuked his Ivy Bridge cpu when it wouldn't post after his TIM surgery session.

OH NO! That would SUCK.
Hopefully I won't be reporting: "waiting for my new CPU arrive" lol

[profJim]

Re: Help with CPU overclocking on z77 OC formula

and hopefully you won't be one of the ones waiting for a new cpu to arrive.

Conspiracy theory?
What makes you think that Intel is getting secret payments from the water cooling suppliers?
.....and they are selling a few more cpu's to the failed TIM surgeons.

[synack]

Re: Help with CPU overclocking on z77 OC formula

and hopefully you won't be one of the ones waiting for a new cpu to arrive.

Conspiracy theory?
What makes you think that Intel is getting secret payments from the water cooling suppliers?
.....and they are selling a few more cpu's to the failed TIM surgeons.

LOL! no, the conspiracy that intel intended to cap the OC ability and temps that SB had. Some say to clear out old stock. I think it's more like, gives them an easy and cheap solution to put out a faster processor by changing how they implement the IHS.

"News Release December 2012: Just in time for the holiday season, Intel releases the Ivy Bridge Ultra Series. Features 'new' temperature management that reduces temperatures by 10c which will allow the processor to run faster on lower voltages while consuming less power"
I jest of course!

But it does kinda make you wonder... The people at Intel are pretty smart folks, and I don't believe someone would make such an amateur mistake. I doubt that nobody noticed the processor running 5-10 degrees warmer stock with such wild disparity between the cores. Nobody thought to check if the "new way" was any good? Nobody at QC said, hey hold on a second... is this acceptable?


But i'm not a conspiracy type guy... It just seems a little hard to believe this was by accident.

Anyway enough about that LOL. Back to dreaming about how awesome my CPU will be after this weekend! hahaha (and not how dead it might be if I mess it up!)

Syn

[profJim]

Re: Help with CPU overclocking on z77 OC formula

Food for thought:
Try lowering CPU PLL (1.80 volt standard setting) to see if your system runs cooler and is still stable.
A web search using: z77 lower cpu pll setting will provide information where some overclocked z77 brands/models ran a few degrees cooler and their systems (hopefully) still passed rigorous stress tests.

I know of one case where this worked with an MSI z77 motherboard where the OP had a devil of a time passing Prime95 v27.7 stress tests.
His system had no stability problems with the earlier 26.6 non-AVX version.
From Free Software - GIMPS:

This latest version of prime95 has been optimized for Intel's new AVX instruction set.
Owners of Sandy Bridge and Ivy Bridge CPUs (Core i3/i5/i7 2xxx and 3xxx) will see a substantial performance boost.

__Happy testing

[synack]

Re: Help with CPU overclocking on z77 OC formula

Food for thought:
Try lowering CPU PLL (1.80 volt standard setting) to see if your system runs cooler and is still stable.

GOOD CALL ON THE PLL VOLTAGE!!

I was able to drop from 1.8 to 1.6 volts and remain stable.
It didn't help idle temps all that much but dropped the full load temps by about 7C. (~69C to ~62C @ 4.7GHz)
That might be worth another 100Mhz :D

I'm a linux guy, so prime95 probably won't work on my system. It may not be a very good indicator of stability for me, as I would need to run it through WINE and that might have stability issues.

However, lucky for us Linux people, is that compiling the linux kernel is one of the most stressful things you can do in linux. And GCC (the compiler) will error out on ANY hardware failure. It has a very low tolerance to hardware issues. AND my GCC is setup to compile to ivy bridge processors, so it will hit just about every instruction in the set. :)

As an example: I had overclocked to 4.9Ghz. I was able to boot right up to and login to the desktop. network/audio/video, all hardware drivers loaded. not a bit of corruption anywhere. I was able to run a console box, ran the CPU monitor. Everything looked great!
Then I ran the Kernel compile, it failed in 2 seconds with an exception error! lol. I then proceeded to do a clean reboot. Linux is very forgiving of hardware issues... GCC is not at all. it's perfect or it fails.


Syn

[profJim]

Re: Help with CPU overclocking on z77 OC formula

NICE!!

As I recall, the MSI z77 was almost stable with CPU Vcore ~ 1.41 before he lowered CPU PLL to 1.65 volts.
His final stable cpu setting was ~ 1.37 volts for his 2500K cpu @ ~ 4.60GHz with the lower CPU PLL.

You can compute the dollar per degree of additional cooling cost of your koolance rp985 pump/res CPU/RAM/VRM wb water cooling setup.
If your wc hardware cost you $300 and it lowered your temps by 10⁰ (C), your cost was $30 per degree cooled.
. In this example, you now owe me $30 x 7⁰ (C) or $210
Normally I charge $$ per ⁰(F) of additional cooling, but I'm feeling generous today.

[parsec]

Re: Help with CPU overclocking on z77 OC formula

If Intel dared to make an improved version of IB CPUs, that did away with the mickey mouse TIM method they are using now (which would be good actually) I would be first in line saying, "Great, here's my old version (via RMA), I want the real deal!". That might sound childish, and is not my style, but... come on! That will never happen.

Intel already announced that most (and all high end) SB CPUs will no longer be available after Q1, 2013. I can't believe they would cripple IB on purpose, that just increases sales of SB to enthusiasts. Of course we're kidding ourselves, the high end enthusiast market is a blip on their sales charts, or at least that's what we are led to believe. Or is this to make Haswell look better? Some of the coming 8-series chipsets have six SATA III ports (YES, about time!!), which makes a Haswell system in my future a done deal. I'm an I/O guy, CPU performance is not the bottleneck IMO, give me storage I/O speed!

Next year we will be seeing new SSDs with the latest ONFI and toggle NAND, at 400MB/s at the NAND chip level, vs 200MB/s at best now, and mostly 166MB/s now, that with synchronous NAND only of course. Of course, if 4K performance is not increased, the result will be a big zero in actual use. Time to shut up...

synack, in case you didn't see this, there is a nice review of your board here:

ASRock Z77 OC Formula Review: LGA 1155 Mainboard for Overclocking and Breaking Records - X-bit labs

It's tested in Windows of course, but I think you'll find some useful info in the review, regarding how the board actually works.

Also if possible, if you do it, I'd love to see some IB CPU pron, naked IB die!

[profJim]

Re: Help with CPU overclocking on z77 OC formula

A web search using:

• ivy bridge cpu tim or
• ivy bridge remove ihs

provides plenty of links for guides and viewing videos or pictures.



Increase the above photo's gamma setting from 1.0 to 1.6 and you can see how easy it can be to nick the electrical traces on the top surface of the pcb.









Removed the IHS from an Ivy Bridge i5 3570k - [H]ard|Forum

[parsec]

Re: Help with CPU overclocking on z77 OC formula

That actually looks pretty easy to do... not that I will be any time soon.

I could swear that the solder-TIM type construction has a much larger area of contact between the CPU die, and the CPU cover/IHS. That is, the area under the IHS is virtually full of solder. On this IB CPU, over two thirds of the IHS is doing nothing but providing structural integrity.

The size and shape of the die also shows why CPU coolers with direct touch heat pipes are a "hit or miss" situation. Given the wrong orientation, the contact of those heat pipes are simply stripes across the die, and could cause a very inconsistent heat transfer between areas of the die. A CPU die is also not consistent in its construction, the processing cores, cache, video processing core, etc, are in different locations. A heat pipe will be touching random areas of the die, and cooling parts of the die more than others, potentially missing the areas that need it most.

IMO, the TIM we can see is not simply filling in microscopic variations in the surface of the IHS and die (the latter appearing polished to a mirror finish) but as a relatively large gap filler, between the die and IHS. Compared to solder, the standard TIM interface is terrible, IMO.

[profJim]

Re: Help with CPU overclocking on z77 OC formula

I'm currently reading page 9 of the Hard|Forum link at the bottom of my previous post.
They're debating numerous thermal interface options and it's getting stale.
It appears that the biggest hurdle is trying to mount a water block to the naked die without destroying the pcb.
There is a promising link where there might be some good info at the end of the thread but I can't connect to it:
http://www.xtremesystems.org/forums/...rom-79C-to-71C
I'm hoping that page 11 is the end of the thread, I'm getting bored.

[synack]

Re: Help with CPU overclocking on z77 OC formula

So far so good.
I'm seeing 3-5C drop in temp.. and most importantly, I didn't fry the CPU.

I didn't go naked die yet. Risky, and I didn't feel like draining all my liquid today, to remove the motherboard, to remove the retaining clip.

it's about 59-62C now at 4700Mhz at full load, ambient temp 28c, water 31C.

These are as good if not better than the temps I was getting at 4500Mhz :)


The process of removing the IHS was quite painless in fact. Everything took maybe 20 minutes total. I happen to have a really good tool, and made the process like cutting through butter.
I took some pictures of the process and will post them once I've had a chance to edit out the serials and such.

Syn

[synack]

Re: Help with CPU overclocking on z77 OC formula

Oh snap.. spoke too soon...
Forgot to drop the voltage on the CPU voltage, now that I'm running cooler....
Just dropped to 1.3V. looking good and I'm NOW running 56-59C at full load.

Gonna go tweak some more.

Syn