Build Log - Corsair Obsidian 900D

[Sneakytiger]

You over clocked ur gfx cards yet Tl as I have a 760 so interested in you're results

[BrotherTobious]

I recently did a new build as well: 4770k @ 4.4 on standard voltages with Corsair H100i cooler and it sits between 26 and 30 degrees when not gaming and I've never seen it above 40 when gaming. I didn't do the de-lid so I can only assume Intel did a better job with later chips.

Whiitey what is the noise level off the fans like for that mate?

[TeaLeaf]

You over clocked ur gfx cards yet Tl as I have a 760 so interested in you're results

My cards are 770s though Sneaky, so they'll not be a great reference for what a 760 will do.

I have not overclocked anything yet though, I'm slowly working my way through my business and other software and getting it updated/reinstalled to Windows 8 versions.   Once I get it all installed and up to date then I'll be taking an image of my system and pushing that to a NAS box, then and only when I have a nice working system image will I be pushing the overclocking up.    Don't expect a radical overclock from me though, my priority was 'fast, stable & quiet' as I use this as a home office PC as well as my main gaming rig.   The water-cooling was primarily to keep the system very quiet rather than to enable a high overclock.

[kregoron]

My cards are 770s though Sneaky, so they'll not be a great reference for what a 760 will do.

I have not overclocked anything yet though, I'm slowly working my way through my business and other software and getting it updated/reinstalled to Windows 8 versions.   Once I get it all installed and up to date then I'll be taking an image of my system and pushing that to a NAS box, then and only when I have a nice working system image will I be pushing the overclocking up.    Don't expect a radical overclock from me though, my priority was 'fast, stable & quiet' as I use this as a home office PC as well as my main gaming rig.   The water-cooling was primarily to keep the system very quiet rather than to enable a high overclock.

Don't make me come over there and do it for you! full power!!

[Chaosphere]

Bringing this thread back to ask a question.

I just got a 4770k and gtx 780 ordered.

Now, I am a bit of a mentalist when it comes to modding and OCing, but I have to admit the idea of de-lidding my bran new cpu scares me a little. I've never broken anything (expensive) before and don't fancy doing it now, as no way I can afford a replacement if I do.

However, I just... want to... What sort of difference did you see, TL? And, the liquid metal stuff you got for the inside, where did you get it and why did you choose that? (over say, a non-conductive TIM)

Also, I mention graphics card, as I wonder if with a single 780, I am being mental to even bother pushing the CPU. I mean, I don't exactly it would ever be the bottleneck anyway! I hear of people pushing 3 titans on sandy bridge systems and the CPU still not causing any bottleneck issues... In which case, maybe I will just undervolt the CPU as much as possible and see what temps I can get that way.

[TeaLeaf]

I can't give a difference as I did the delid before I built the system, so any comparison I have raised is based on a comparison of the temps that others have reporting for their i4770K.      From the reading I did at the time my temps were approx 20 degrees lower at idle and more under load.

My motivation for the delid was to get the quietest system possible, hence I went for water-cooled and wanted to delid to reduce core temps to minimise the fan speeds needed within the system.     Having done the delid and seen the way the HS is mounted I can well understand why there is a significant improvement in performance when you remount the IHS.  The black mounting 'goop' seems to lift the HS a decent way and whilst the 'contact patch' with the IHS was fairly good on my CPU (see below picture), it was not as thin as it could have been as  the TIM had to fill the gap between the core and the IHS - and thickness of TIM hits temps harder than almost anything else - and that leads onto why I chose Cool Laboratory Liquid Pro Liquid Metal TIM.  
[ATTACH=CONFIG]2329[/ATTACH]

I did a lot of reading into what got best results and I always came back to the same result - everyone was reporting at least 10 degrees better results (sometimes more) with Cool Laboratory Liquid Pro Liquid Metal TIM than they were getting from other high performance TIMs.   Many of the reports I read were direct comparisons done on the same system by the same person in a back-to-back comparison.    Cool Laboratory Liquid Pro Liquid Metal TIM seems to spread so thinly that it allows the most effective contact with the CPU/IHS and therefore significantly better heat transmission and lower temps.   The only risk is the 'pumping' effect from constant heating/cooling cycles pushing the conductive TIM onto the contacts on the surface of the CPU - hence the nail varnish application to protect against that.

I could have also used Liquid Pro between the IHS and my water-cooling block, but whereas the reports I read seemed to give Liquid Pro a significant advantage in the CPU>IHS cooling, there did not seem to be as significant an advantage (if any) in the IHS>Cooling Block cooling.   As using conductive Liquid Pro on top of the IHS would create a much greater potential for electrical shorts, and as any cooling advantage of using Liquid Pro between the IHS/waterblock seemed anecdotally to be minimal, I opted for a non-conductive TIM for that interface.   I chose Noctua NT-H1 as across all of the reports I read it seemed to consistently give the best real world cooling gains compared to others (although AS5 was close iirc).

I got both compounds from overclockers.co.uk.

Difficulty Factor: 1/10
Sphincter Factor: 10/10

The delid scared me too, but after actually trying it the process was remarkably simple and easy to accomplish - delids with the method I used do not seem to fail, whereas the scalpel method seems to have a much higher potential for failure (and injury).   The biggest advice I can give is to make sure that your vice will not move or twist when you hit the PCB with the hammer.    I had problems initially until I switched from a handheld wood vice with wooden faces (which was awful and really not what I should have been using) to a properly secured metalwork vice with metal jaws.    The decent vice meant that all of the hammer force went into shearing the seal between the CPU and IHS, instead of into flexing the vice jaws or moving the vice itself.     Use the correct vice and the delid is seriously easy.

Hope this helps.

[Whitey]

My advice would be to try it without doing the de-lid. I didn't do mine and have similar temps to TL.  Read previous posts in this thread for details

[TeaLeaf]

My advice would be to try it without doing the de-lid. I didn't do mine and have similar temps to TL.  Read previous posts in this thread for details

I think you're lucky if you have those temps stock, but you are also water-cooled iirc?   Experience seems to vary a lot from batch to batch, I went for the delid to ensure I got low and to save me the trouble of building the system, then having to strip it back down again to do a delid!

[suicidal_monkey]

standard claw hammer
...
3 raps
...
CPU popped off.
...
remarkably simple and easy, something anyone could do.

Difficulty Factor: 1/10
Sphincter Factor: 10/10

:crazy: crazy

[Chaosphere]

Aye, thanks TL. Will have a think on it. Set the PC up today. When I have finished my uni work, I will start to play about with the CPU and see what I can do with it atm. I notice already, with everything in the bios just left to auto, the CPU hits temperatures in the 70s when running prime95 for an hour or so. Admittedly that is with the fans at 800 rpm... but still, hot CPU considering the lack of overclock.

[Chaosphere]

@ Whitey,

I will monitor temps during some BF4 tonight and see what comes up!

[Chaosphere]

Well after running some tests, the chip is just too hot for my liking. At somewhere around 1.25vcore I am getting temps in the 70s for gaming, and the chip is hitting the 90s in stress testing and some benchmarks. 1.25vc is more than I need for the 4ghz its set to atm, but I wouldn't want to start upping the mhz or the voltages anymore with temps already this high.

So, I have ordered some liquid metal for under the IHS, and as5 for on top of it. I am tempted to have a try with the razor blade / scalpel (I have a fresh scalpel - the joys of studying medicine) method, and just seriously take my time. The idea of smashing the chip with a piece of wood terrifies me! But, I shall do more reading before, as you just made it sound so damn easy...

[TeaLeaf]

The idea of smashing the chip with a piece of wood terrifies me! But, I shall do more reading before, as you just made it sound so damn easy...

Trust me, the delid with hammer/wood/vice is the easy method, scalpel is the damned difficult method (I tried both).   The problem with the scalpel method (apart from personal injury) is the sheer quantity of black glue they have used - you have to put a significant amount of pressure on the scalpel to wedge the blade in between the IHS and the PCB and all it takes is one little slip and you go right through part of the circuit, or slice off a resistor or your finger.

The beauty of hitting the wood against the PCB is that it spreads the load evenly and strikes the PCB along it's strongest line.   Net result is that the force goes directly into sheering the black glue.

Over to you!

[Tutonic]

Whatever you end up doing, please video it so that we can applaud/laugh at your incredible success/crushing failure.

[TeaLeaf]

Last minute addition to the advice:

When I did the delid I had a cloth bag covering the exit route from the back of the vice (under the cpu level) and over the top of the CPU (so I could just see the PCB but not the central CPU area).   When the PCB with CPU sheers away from the glue you'll need something to catch it before it hits the ground and breaks - hence the bag!