Overclockers Australia Forums

OCAU News - Wiki - QuickLinks - Pix - Sponsors  

Go Back   Overclockers Australia Forums > Manufacturer-specific Forums > Intel x86 CPUs and chipsets

Notices


Sign up for a free OCAU account and this ad will go away!
Search our forums with Google:
Reply
 
Thread Tools
Old 16th March 2017, 12:17 AM   #556
Jonbanez
Member
 
Join Date: Dec 2007
Location: Newcastle
Posts: 267
Default

Quote:
Originally Posted by mr_mordred2095 View Post
There's 10x the surface area between IHS and waterblock, so there's 10x less BENEFIT (benefit, not cooling performance) compared to regular decent paste. Combine that with 10x the surface area, which means it's expensive to apply.
I see, thanks for the info. Slightly off topic, but would the same principle apply when using conductive pastes between a GPU die and water block?
__________________
Trading Since 2008. PM me or check previous posts for details.
Jonbanez is online now   Reply With Quote

Join OCAU to remove this ad!
Old 16th March 2017, 12:27 AM   #557
mr_mordred2095
Member
 
Join Date: Mar 2004
Location: Brisbane
Posts: 3,813
Default

Quote:
Originally Posted by Jonbanez View Post
I see, thanks for the info. Slightly off topic, but would the same principle apply when using conductive pastes between a GPU die and water block?
Correct. I see minimal improvements with Liquid Metal TIMs compared to AS5 or the like (less than 5c) on graphics cards.

Last edited by mr_mordred2095; 16th March 2017 at 12:31 AM.
mr_mordred2095 is offline   Reply With Quote
Old 16th March 2017, 6:50 AM   #558
bennyg
Member
 
Join Date: Dec 2005
Location: Melbourne, Oztraya
Posts: 2,711
Default

Quote:
Originally Posted by Jonbanez View Post
I'm wondering why noone seems to use the liquid metal compounds between the IHS and CPU block? I realise there is a risk of shorting the components, but surely this can be avoided if you're careful. Are there other disadvantages?
I had it on top of the IHS before I delidded and just did it again afterwards. On a "laptop" that gets moved around. With the liquidest metal, Conductonaut. If you do it properly it doesn't run. When I first did it I left about 2mm from the edge of the IHS and I taped off around where the IHS contacts the heatsink since that's underneath the CPU as it sits. So if it ran, gravity is taking it away from the socket anyway.

You can take the heatsink off repeatedly to check the fit and slowly build up the right amount to avoid it running.

I got big temp drops on my 980Ms. I cannot get them over 80C at 25% OC drawing 130W+. Maybe the compactness of MXM format amplifies the benefits - each is a 970 on a 10cm x 8cm PCB
__________________
P870DM-G 6700K delid+LM @4.5GHz -45mV - 32Gb DDR4-2666 - 4K QFHD 17" - 980M SLI
+ a bunch of Ivy Bridge laptops - mobile CPU socket + anti-proprietary standard-loving basketcase
Trades: >70, Value: I can't count that high

Last edited by bennyg; 16th March 2017 at 6:54 AM.
bennyg is offline   Reply With Quote
Old 16th March 2017, 10:35 AM   #559
cbwolf
Member
 
cbwolf's Avatar
 
Join Date: Dec 2001
Location: Melbourne
Posts: 2,107
Default

Quote:
Originally Posted by Jonbanez View Post
I'm wondering why noone seems to use the liquid metal compounds between the IHS and CPU block? I realise there is a risk of shorting the components, but surely this can be avoided if you're careful. Are there other disadvantages?
I'm assuming you mean between the IHS and water block/heat sink.

If so, there's little chance of shorting anything. The real risk of shorting stuff is actually from using it between the CPU core and the IHS because of the surface contacts on the pcb. Luckily on Skylake/Kabylake the core is quite small in the middle of the pcb and the exposed metal contacts fairly far away. So even if the liquid metal did run a bit it's unlikely it would make it to the contacts.

I don't use liquid metal on top of the IHS for a couple of reasons:

1. As mordred said there is little benefit over regular TIM. Maybe 2 - 3 degrees at most.

2. It damages the IHS and block/heat sink surfaces. Not much, but enough to stain or etch the surface. I had liquid metal on the IHS of my last 7700k for a month and when I took it off, it stained my heat sink and removed all of the print on the IHS - not good for resale value.
__________________
i7 7700k | Asus ROG Z270E Strix | MSI GTX1080 Gaming X | 32GB Trident Z 3200Mhz DDR4 | 256GB + 1TB Samsung 960 m.2 SSD | Philips 43" 4k IPS Monitor
Over $5,500 in forum trades and counting!
cbwolf is offline   Reply With Quote
Old 16th March 2017, 12:25 PM   #560
SneakyBastd
Member
 
Join Date: Mar 2005
Posts: 224
Default

Quote:
Originally Posted by cbwolf View Post
replaced the TIM between the core and IHS with Coolaboratories Liquid Ultra with Noctua NTH1 between IHS and Heatsink
[*]7700k @ 5.1Ghz @ 1.440v
I'm going to use exactly the same stuff. I'm surprised to see 1.44v though??? I've been reading 1.4v is about the limit... is that wrong? Wondering what is safe for 24/7 use. Cheers.
__________________
Sales to: noremac12321 FireWire1394
SneakyBastd is offline   Reply With Quote
Old 16th March 2017, 12:29 PM   #561
mr_mordred2095
Member
 
Join Date: Mar 2004
Location: Brisbane
Posts: 3,813
Default

Quote:
Originally Posted by SneakyBastd View Post
I'm going to use exactly the same stuff. I'm surprised to see 1.44v though??? I've been reading 1.4v is about the limit... is that wrong? Wondering what is safe for 24/7 use. Cheers.
I've read that Intel class 1.5v as the 'safe max'.

https://www.google.com.au/search?q=k...hrome&ie=UTF-8
mr_mordred2095 is offline   Reply With Quote
Old 16th March 2017, 12:36 PM   #562
SneakyBastd
Member
 
Join Date: Mar 2005
Posts: 224
Default

Quote:
Originally Posted by Jonbanez View Post
No worries mate. Another thing to look out for - I cleaned the existing TIM from the IHS a little too vigorously. I ended up not having any distinguishing features to show me where to spread the conductonaut on the IHS...
I read that putting liquid metal on the underside of the IHS is fine, but not needed. I wasn't planning to do it myself, just cover the die with a thin layer and replace. Will let you know how it goes.
__________________
Sales to: noremac12321 FireWire1394
SneakyBastd is offline   Reply With Quote
Old 16th March 2017, 12:43 PM   #563
mr_mordred2095
Member
 
Join Date: Mar 2004
Location: Brisbane
Posts: 3,813
Default

Quote:
Originally Posted by SneakyBastd View Post
I read that putting liquid metal on the underside of the IHS is fine, but not needed. I wasn't planning to do it myself, just cover the die with a thin layer and replace. Will let you know how it goes.
I find doing it to the underside of the IHS is massively beneficial, else it has a kind of liedenfrost effect when you push the 2 together.
mr_mordred2095 is offline   Reply With Quote
Old 16th March 2017, 3:23 PM   #564
cbwolf
Member
 
cbwolf's Avatar
 
Join Date: Dec 2001
Location: Melbourne
Posts: 2,107
Default

Quote:
Originally Posted by mr_mordred2095 View Post
I find doing it to the underside of the IHS is massively beneficial, else it has a kind of liedenfrost effect when you push the 2 together.
Agree with this. Absolutely apply to the core and also the underside of the IHS.

Also remember that liquid metal goes a long, long way and you need less than you think you do.
__________________
i7 7700k | Asus ROG Z270E Strix | MSI GTX1080 Gaming X | 32GB Trident Z 3200Mhz DDR4 | 256GB + 1TB Samsung 960 m.2 SSD | Philips 43" 4k IPS Monitor
Over $5,500 in forum trades and counting!
cbwolf is offline   Reply With Quote
Old 16th March 2017, 5:52 PM   #565
Ratzz
Member
 
Ratzz's Avatar
 
Join Date: Mar 2013
Location: Cheltenham East 3192
Posts: 4,659
Default

Quote:
Originally Posted by SneakyBastd View Post
I'm going to use exactly the same stuff. I'm surprised to see 1.44v though??? I've been reading 1.4v is about the limit... is that wrong? Wondering what is safe for 24/7 use. Cheers.
My 7700K is not a great chip, needs 1.47v for a stable 5.1GHz but will do 5.2 @ 1.5v. Its a bit too flaky at 5.2 for benching though. Even at 1.5v I only occasionally go to 80 or so, it mostly sits in the mid 70's.. of course with a custom water loop and delidding/conductonaut.

I'm running totally stable 5GHz @ 1.45v 24/7.. although it will boot @ 1.37v at that speed, its a little flakey, and since temps @ 1.45v are only in the 50's I'm pretty comfortable with leaving it at that for now. I'll work on finding an ideal voltage for 5GHz eventually, I'm guessing around 1.42v for my chip.

If you try to use 1.52v on my little ITX board, regardless of temps, it will halve the clocks, so I'm guessing there is an enforced limit of 1.51v before it wants to throttle.

I haven't tried doing only the core with conductonaut but it works well with doing both, and without rubbing it in well, it doesn't stick, so I'd wonder if something bad might happen if it were to flow along the IHS having not been rubbed into it. I'd be applying it to both surfaces as recommended in the instructions personally.

Don't forget to insulate the metal contacts closest to the core either, some discussion about this earlier in the thread FYI. I used liquid electrical tape.
__________________
I always say 'morning' rather than 'good morning'. If it was a 'good' morning then I would be on a yacht in the West Indies smoking a 10" reefer and getting a massage from Lexa Doig...

My Rig

Last edited by Ratzz; 16th March 2017 at 6:02 PM.
Ratzz is offline   Reply With Quote
Old 16th March 2017, 6:32 PM   #566
headin2001 Thread Starter
Member
 
headin2001's Avatar
 
Join Date: Apr 2007
Location: Sunshine Coast
Posts: 2,853
Default

Quote:
Originally Posted by Ratzz View Post
I'm running totally stable 5GHz @ 1.45v 24/7.. although it will boot @ 1.37v at that speed, its a little flakey, and since temps @ 1.45v are only in the 50's I'm pretty comfortable with leaving it at that for now. I'll work on finding an ideal voltage for 5GHz eventually, I'm guessing around 1.42v for my chip.
What sourcery is this. In the 50's with 1.45v. Think you sensors are out. If you are running a low VCCPLL it will be incorrect for sure. Some CPUs I've had needed 1.12-1.15v for it to read correct.

Otherwise, how's the ambients in our southern states now? I'm still rocking 31deg and after temp here in sunny QLD.
headin2001 is offline   Reply With Quote
Old 16th March 2017, 7:03 PM   #567
Ratzz
Member
 
Ratzz's Avatar
 
Join Date: Mar 2013
Location: Cheltenham East 3192
Posts: 4,659
Default

Quote:
Originally Posted by headin2001 View Post
What sourcery is this. In the 50's with 1.45v. Think you sensors are out. If you are running a low VCCPLL it will be incorrect for sure. Some CPUs I've had needed 1.12-1.15v for it to read correct.

Otherwise, how's the ambients in our southern states now? I'm still rocking 31deg and after temp here in sunny QLD.
Yeah about the same here mate. Today was a little cooler (mid 20's?) but humid as fuck, but although summer is over we've pretty much had over 30 all week.

I'm still learning how to use this stupid MSI BIOS and as you know I'm hardly expert at what I'm doing anyway. I'm in the 50's for general use, not 100% loads, but I'm confused anyway as the damn thing is IDLING at 48 in the BIOS so I would have expected a lot more once I actually got a few things happening than the 52 I am currently showing.

I'm not doing a lot mind you, I have a couple windows open, this one and a movie, and downloading a few torrents, but I would have expected a bit more heat than that.

Maybe the sensors are lying, for sure. Its a nice board but in the end, its still just an ITX with all the various limitations that go with that.

The GPU is at stock volts right now, one thing I have noticed is that my 970 even pushed as far as I could push it would rarely add more than 2 to the CPU temp, but if I push this 1080 I can add more like 7 to it.

I guess the higher clocks of the 1080, and maybe a more efficient cooler which is actually moving heat into the water better might have that effect? With everything turned up as high as I can get it, so far, the GPU doesn't get over 46 which is pretty impressive. I'm just assuming the factory waterblock is more effective?

Hard to compare a 970 to a 1080, but I think you get my general drift.

Its the same EX280 rad I've used in like a dozen builds now, nothing special, with 3 Redux PWM 1500's (no room for number 4 ) a Raystorm Pro, and the D5 vario is running at minimum due to the very small loop it has to feed. The case looks pretty crammed but actually has pretty decent airflow. (link in my sig)

Oh and because I can't help myself.. 'sorcery'
__________________
I always say 'morning' rather than 'good morning'. If it was a 'good' morning then I would be on a yacht in the West Indies smoking a 10" reefer and getting a massage from Lexa Doig...

My Rig
Ratzz is offline   Reply With Quote
Old 17th March 2017, 8:57 AM   #568
jhonnypolak
Member
 
Join Date: Oct 2008
Location: Sydney
Posts: 9
Default

Quote:
Originally Posted by wraith.50 View Post
That's an excellent temperature drop. What cooling are using air or water ?
Using a Kelvin S24 AIO

Quote:
Originally Posted by wraith.50 View Post
You wrote 1.24-1.32 volts, but what is your actual voltage at 5ghz, when running Prime 95 small FFT's test ?
It fluctuates because I'm using Auto Voltage on the CPU. I haven't had time to tighten it down.
jhonnypolak is offline   Reply With Quote
Old 17th March 2017, 1:31 PM   #569
SneakyBastd
Member
 
Join Date: Mar 2005
Posts: 224
Default

Just got my system assembled last night (7700K/IX Hero), and did the "cookie cutter" overclock suggested in this vid:
https://www.youtube.com/watch?v=wlR7CNuoOXk

Working well so far. Haven't done any proper lengthy stress tests yet (haven't had time), but the 15 minute XTU runs passed, cinebench, 3dmark, etc, all work fine so far...

Max temps on each core are 66/69/70/71 in a moderately warm room. Haven't delidded it yet. Really happy that the magic 5ghz seems to be a goer for me
When I get more time I'll try faster, and then lower vcore.
__________________
Sales to: noremac12321 FireWire1394
SneakyBastd is offline   Reply With Quote
Old 18th March 2017, 9:32 AM   #570
SneakyBastd
Member
 
Join Date: Mar 2005
Posts: 224
Default

I've found that 5hgz/1.3vcore/LLC5 is fine for passing 15 minute tests of Prime95/AIDA64/XTU.
But some games crashed within minutes. Clearly it's not stable. And then I tried RealBench. OMG, it finds a blue screen alright.

After some experimentation with upping the vcore to 1.34 in bios, I noticed in hwmonitor it never showed that being used. It showed 1.328v. It helped a bit but the blue screens continued...

I then tried LLC6, and instantly hwmonitor shows vcore at 1.344v. Realbench passed!
Gradually lowering vcore, I find 1.32 in bios still passes. 1.31 does not.
Note that in hwmonitor, the vcore shows as 1.312v whether I've set 1.31 or 1.32 in the bios. But only 1.32 in bios is stable. Weird?

The kicker is:
LLC5, 1.34 in bios = 1.328 in hwmonitor. FAILS tests.
LLC6, 1.32 in bios = 1.312 in hwmonitor. PASSES tests. (About 3-4 degrees warmer, cores max at 71-74C)

Question is: is this approach best? It seems LLC is more important than vcore?!? Should I have stayed at LLC5 and kept upping vcore in bios, hoping it would eventually draw what it needs?

Thanks for putting up with my nub overclocking attempts!
__________________
Sales to: noremac12321 FireWire1394
SneakyBastd is offline   Reply With Quote
Reply

Bookmarks

Sign up for a free OCAU account and this ad will go away!

Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT +10. The time now is 8:52 AM.


Powered by vBulletin® Version 3.8.9
Copyright ©2000 - 2017, vBulletin Solutions, Inc.
OCAU is not responsible for the content of individual messages posted by others.
Other content copyright Overclockers Australia.
OCAU is hosted by Micron21!