EK Supreme HF CPU Block performance

Scoob

New member
Hi all,

As part of my water cooled build I have the EK Supreme HF CPU block in Acetal. Nice looking block and quite highly regarded, at least in the reviews I read.

It's in my loop now, cooling a 2500k @ 4.5ghz. My 2500k isn't one of the better ones, needing 1.366 vCore under load to be fully stable at that speed. Still, far from "bad" so I'm happy.

I had an Antec Kuhler 620 before - you know, the closed loop WC kit with a thin 120mm rad. I ran it in push-pull with a proper cold air feed supplied by some corrugated hose like you have on older tumble driers - worked well.

Anyway, the issue I have is that the EK Supreme HF CPU block appears to perform no better than the 620 did! My loop is overkill, having a Phobya 1080 rad to cool the CPU and two 570's - well one at the moment, but that's another story.

The GPU's are both cooled VERY well, I don't see above 40c except on the hottest of days and that was with both cards clocked at 850 - up from stock 732. The CPU however will still hit 70c under load in IBT testing - a little down on the 620, but not much.

Like I said, my loop works well for the GPU's - I use parallel flow, so the coolant goes through the GPU's before going to the CPU. Still, the thermal capacity of the water should be more than enough for this not to be an issue. Indeed, dropping to just ONE GPU had no effect on the CPU temps what so ever. So, two overclocked 570's saw the same CPU temp as ONE 570 at stock.

So, I thought maybe I'd bodged mounting the CPU block or something, very unlike me as I'm pretty anal when it comes to that. Regardless, I drained the loop and popped the CPU block off so I could clean and re-seat it. On removal I saw a perfect spread pattern of TIM on the CPU and block - so no problems there - but I re-did everything anyway, just in case. For the record I fully cleaned and treated both the CPU and block with cleaning fluids, before re-applying TIM and re-seating the block. It made no difference at all.

Basically, despite being 100% sure my CPU block is properly mounted, there are no air locks, flow is excellent, loop cooling performance is total over-kill, my CPU runs hotter than I'd expect.

Everything works fine, rig is near-silent, but I cannot quite get over the CPU temps, especially as those GPU temps impressed me so very much.

I am wondering if I should lap the CPU, maybe the 2500k's aren't very flat vs. the block. I did so with my old Q6600 on air cooling to great effect years ago, but had assumed Sandy Bridge was made with finer tolerances so lapping was a thing of the past...maybe not.

Any advice welcome, happy to answer question regarding my loop, but it really is overkill for what I'm cooling - as was my intent. I've not posted my build log yet - my first rad was faulty and I wanted to wait - but it's basically an external water box (rad chassis) holding a Phobya 1080 (that's 360x360) rad with a D5 pump and res on the side and 4x 180mm Phobya fans in push.

F.Y.I. Tom and I actually had a little chat about my loop last year. Unfortunately, despite having a very comprehensive parts list, I couldn't start my build at the time due to injury. He later posted a great video using many of the bits we'd discussed, here it is:

http://www.youtube.com/watch?v=Am7ypdKUgfU&feature=plcp *

great minds and all that lol. Watching this caused me to change a couple of bits when I eventually came to order, but my external loop is basically the same as the one Tom created. As I used QDC's I can easily transfer the entire cooling assembly to another PC. The pump, fans etc. are all powered by a little external PSU, so it's all very neat, with no wires from the PC - just the two hoses :)

Cheers,

Scoob.

* the new forum doesn't like "nice" links does it?
 
Hi,

No thoughts or advice anyone?

I read elsewhere that due to the nature of the screws that clamp the EK Supreme HF down that with some CPU's it's not actually pressing down that hard. Sounds a little strange, but there must be some reason for my high temps.

I did an IBT yesterday and was hitting 70c on the hottest core - that's daft, and near the same as my Antec Kuhler 620 was getting. It's like the heat isn't getting to the block - I'm using MX-4 and it's applied nicely.

I had thought about lapping the CPU, but if there is an issue with clamping force - even when fully tight - that'd just make things worse.

Anyone else out there using the EK Supreme HF Nickel+ EN with Acetal top CPU block and having issues?

Cheers,

Scoob.
 
Damn, I thought I replied to this once!

I don't really know to be honest - everything you've said rings true. You've got a good component selection and the GPUs are being cooled very well, as well as I'd expect. The CPU block is meant to be very competitive all round so the only thing that springs to mind is checking the inside of the block itself to see if there is any furring or blockage from manufacturing (I can remove the copper base plate on my RASA and I guess you can do something similar).

If the block isn't pressing down hard enough then have you got some spacers lying around which you can use in front of the screw heads?
 
Hey M&P,

Thanks for your reply.

Yeah, I'm a little confused. The block is new and performance has always been like this. I gave the entire loop a good flush before using it in anger so any residue from manufacturing should have been flushed out. Loop really hasn't been running long enough for any furring up or anything, plus I run EK's own pre-mix which should prevent that.

I have thought about using some spacers as you suggest, it's just a pain to pull the thing apart and drain the PC bit of the loop once again you know. I think I will have to bite the bullet and just do it mind...

Spent a while building this loop, only to have my first rad be duff and need to be replaced, then I had a compression fitting fail killing one of my 570's - I just want to use the thing now! lol.

Cheers,

Scoob.
 
Ouch it doesn't sound like you've been having any luck!

The only thing I'd check for inside the water block would be to see if something hadn't been formed correctly in the manufacturing. I think it's highly unlikely but you never know.

I recently changed my motherboard and CPU without draining my loop because I was lazy and actually it worked fine, although I had to be careful given that I've got two 7950s hooked up too. You could try just removing one screw at a time and fitting in a washer if you've go the space to work? It's a little lazy though but I think that's what I'd do! There's no need to replace tim or drain the loop that way.
 
Lol, no, really not!

Might do that, very much last thing to check though I'd think.

Heh, to be honest, draining my loop etc. isn't so hard. I can isolate the PC from the external rad / pump / res assembly using QDC's - it's the time spent leak testing that frustrates when I want to game.

When I get some time I will likely pull the loop apart again and triple check things. Every thing does run fine, and I only see 70c+ in IBT, gaming is usually around the mid-50's on hottest core. The Prime 95 blend test also hits mid-to-high 50's on the hottest core and stays there too bar the odd blip higher every now and again - which sounds fairly healthy.

Maybe IBT generates so much heat that is IHS cannot shift it too the block quickly enough - after all IBT really is one of the most extreme CPU stress tests out there...a full 10-15c hotter than Prime 95!

Cheers,

Scoob.
 
For the most part, I only see a 5-8*C difference between IBT and Prime - maybe you're right about the heatsink not being able to dissipate the heat from IBT quickly enough.
 
Hi Josh,

5-8c isn't much at all, I've always seen a massive difference between IBT and Prime regardless of the cooling. Friends of mine report the same. Still, I'm not going to worry too much for the moment, though I will inspect things properly next time I drain the loop - likely at the weekend knowing me.

IBT is extreme and gaming performance is fine as well as being near-silent, so I shouldn't complain really. If I just used Prime for stability testing I expect I'd not have been concerned about temps at all...

I need to also remember that my 2500k isn't one of the good ones and needs more vCore than most for 4.5 1.366v - under load. Going to just 4.6 needs a BIG jump, seeing me peeking at over 1.4v in IBT - too much for me! lol.

Cheers,

Scoob.
 
Back
Top