Innovation Cooling's Diamond 7 TIM test results

[123bob]

Maybe that's where the twist came in at Anandtech. If you start on one side it might tend to push the paste in the opposite direction and give an uneven spread. Do you remember which side you started on?

I've got a busy week this week but if I find time I'll see if I can model it with a glass slide and a heat sink tomorrow.

I put the lower left screw in first on my pic. That's the corner that was missed. I put no real pressure on it. That didn't come until I started the screw in the opposite corner. Makes sense. It levered down and squished the paste to the upper right. Next attempt, I'll give it a good squish down before I put in any TRUE screws....Occam's Razor...

EDIT: You can see TRUE screws in Bail's pics below. They are a spring loaded affair that require significant pressure to get the second screw started. The third and fourth screws aren't as bad to put in. Thx Bail!!

Bob

[Bail_w]

Update: Air cooling results. Same quad core chip from watercooling with IC Diamond 7

Setup:
Asus P5k Motherboard
Q6600 @ 3.7Ghz with 1.5v
Thermalright Ultra 120 Heatsink with dual fan @ 7v (Custom backplate Bracket, to apply more pressure)
***4 screws were screw all the way down with custom backplate bracket.







Ambient Temp : ~19C




Idle:

Core 0: 34
Core 1: 34
Core 2: 28
Core 3: 32



Load (18 mintues WCG 100% load)

Core 0: 62
Core 1: 62
Core 2: 58
Core 3: 59



After disassemble:





The spread was very good because i trim the extra thread of the thermalright ultra 120 bracket to apply more pressure.

Will update later with AS5

[123bob]

Bail, did you squish down the sink before you started the screws? I didn't on mine above.

BTW, good mod on the backplate. I think I'll do that on my next go.

Regards,
Bob

[Bail_w]

Bail, did you squish down the sink before you started the screws? I didn't on mine above.

I did squish the heatsink down, then turn the screws simultaneously.

[tastymannatees]

Good, mount, good spread, how long on the cure?

When you modded the backplate you increased the pressure? Did your temps drop after the change?

[Bail_w]

here are my AS5 testing that i just did couple minutes ago.

Same setup, configuration and method of applying TIM.





Idle:

Core 0: 39
Core 1: 39
Core 2: 34
Core 3: 36



Load (~4 mintues WCG 100% load, maybe need time to cure but i dont think it will drop more than 2-3C )

Core 0: 66
Core 1: 66
Core 2: 61
Core 3: 62



After disassemble:





*Good spread as the IC Diamond 7 but the load temp were a bit higher. Clearly, for me air cooling and lower vcore benefits the most out of IC Diamond 7.

Good, mount, good spread, how long on the cure?

When you modded the backplate you increased the pressure? Did your temps drop after the change?

At first when i did the mod it was intend for my IHS-less E2160, obviously i will benefit better temperature because the screws now can go further down to spread the TIM more evenly. Btw, for those who have the themalright ultra 120 extreme mounting bracket, highly recommend to trim the extra thread out of the backplate bracket.

[tastymannatees]

I put the lower left screw in first on my pic. That's the corner that was missed. I put no real pressure on it. That didn't come until I started the screw in the opposite corner. Makes sense. It levered down and squished the paste to the upper right. Next attempt, I'll give it a good squish down before I put in any TRUE screws....Occam's Razor...

EDIT: You can see TRUE screws in Bail's pics below. They are a spring loaded affair that require significant pressure to get the second screw started. The third and fourth screws aren't as bad to put in. Thx Bail!!

Bob

I may have some similar springs at work, I see if I can get a reading on them. I have one heat sink for an HP workstation with springs for a 90 psi mount I wonder why Intel's spec is @ 43 seems a little light to me

[123bob]

I may have some similar springs at work, I see if I can get a reading on them. I have one heat sink for an HP workstation with springs for a 90 psi mount I wonder why Intel's spec is @ 43 seems a little light to me

It's actually a bit more complex than the spring tension alone. The actual sink hold down is an "X-bar" affair that clamps the sink. Real clear picture, step #4 here.

The screws actually bottom out on stops when they hit the post on the backplate, or in Bail's case, the mobo.

Bob

[Bail_w]

The screws actually bottom out on stops when they hit the post on the backplate, or in Bail's case, the mobo.

Bob

Thats correct, so if you trim that extra thread, you can apply more pressure.

Original backplate. *From sidewindercomputers

[123bob]

Thats correct, so if you trim that extra thread, you can apply more pressure.

I may have to pull mine a bit early, just to try it out...

[tastymannatees]

Thats correct, so if you trim that extra thread, you can apply more pressure.

Original backplate. *From sidewindercomputers

OK, got it. You guys are making me hustle 3X the work I had on tests in the other forums.

[123bob]

OK, got it. You guys are making me hustle 3X the work I had on tests in the other forums.

We are XTREME here..... This is normal for us.....

You might want to think about bedtime, given your address.....

Bob

[Bail_w]

OK, got it. You guys are making me hustle 3X the work I had on tests in the other forums.

We are not here to take your samples and run

[littleowl]

That is because we do things right here on xs!

[123bob]

On a more hair-brained idea, I was thinking about AS5 and it's cure directions. Got this right off their site, here.

"Due to the unique shape and sizes of the particles in Arctic Silver 5 conductive matrix, it will take a up to 200 hours and several thermal cycles to achieve maximum particle to particle thermal conduction and for the heatsink to CPU interface to reach maximum conductivity. (This period will be longer in a system without a fan on the heatsink or with a low speed fan on the heatsink.) On systems measuring actual internal core temperatures via the CPU's internal diode, the measured temperature will often drop 2C to 5C over this "break-in" period. This break-in will occur during the normal use of the computer as long as the computer is turned off from time to time and the interface is allowed to cool to room temperature. Once the break-in is complete, the computer can be left on if desired."

(My bolding)

Hmmm, thermal cycling. We crunchers don't do that...It's 24/7/365 for us. I took one of my WCG profiles, adjusted the "CPU usage" parameter to 50% and now have Farm-09, the one with the recent AS5 test, to cycling. The temp is jumping up and down as you would expect. About 3C, but jumping. I can only get CT to do 1 second samples, so I may not be seeing the full swing.

It would be more of an extreme swing if I suspended, resumed, suspended, etc.

The duty cycle by modding the profile is rather quick. If I could have 20-30 seconds on, 20-30 off, that would cycle the temp more radically. Hence, it may cure any paste of this type, including D7, faster.

My question to the SW jocks out there, could some kind of simple macro be cooked up that will send the resume/suspend command to BOINC, much like you can do with BoincView? Can it be done inside BoincView? If not, this standalone type program would have to supply the unique "GUI_RPC_AUTH" code to boinc, and have the requisite "REMOTE_HOSTS" info put in the target machine, if not run locally. I obviously don't know quite what I'm talking about here, but you I hope my Neanderthal description helps....

Of course, something like a primes or OCCT would be a better candidate for this, but at least with BOINC, we're still working while we cure, in both senses of the word , right?

A quick and dirty approach would also be welcome. I'd like to try this.

Thanks and Regards,
Bob

[D_A]

Errrm ... put it in 'maintenance/upgrade mode', go have dinner, put back in 'normal/crunch mode'. Rinse, repeat.

That quick and dirty enough? Your quoted text suggested letting the TIM drop to ambient, not idle, temps.

[123bob]

Errrm ... put it in 'maintenance/upgrade mode', go have dinner, put back in 'normal/crunch mode'. Rinse, repeat.

That quick and dirty enough? Your quoted text suggested letting the TIM drop to ambient, not idle, temps.

Yeah, it says to go power off ambient, but that ain't happening on this farm.....

As to manually doing it, that might be significant work....particularly if we are talking about my entire farm....I'm lazy by nature, that's why I became an engineer, to invent things to allow me to take a nap.

I realize dropping to idle is not as good as ambient, however I don't think power cycling the machine would be too good for it, particularly those who run highly overclocked D9 RAM......

Bob

[DDTUNG]

Seeing that the results posted so far indicate that ICD7 lowers temperatures by 1-2C over its rivals, the following questions beg to be asked:

1. Does the use of ICD7, coupled with high end cooling equipment, allow higher or more stable overclocks?

2. Is there deterioration of cooling performance over time(of most interest to DC folks)?

DDTUNG

[123bob]

Seeing that the results posted so far indicate that ICD7 lowers temperatures by 1-2C over its rivals, the following questions beg to be asked:

1. Does the use of ICD7, coupled with high end cooling equipment, allow higher or more stable overclocks?

2. Is there deterioration of cooling performance over time(of most interest to DC folks)?

DDTUNG

Realize, this is my .02 USD....with the recent trade in dollars, tomorrow it may be worth more, or less.....

1. So far, it would seem so, when careful attention is paid to the application. As to the increased OC part of the question, that remains to be seen. By the nature of the test, we can't adjust OC because that throws a new variable into the test. I have two machines, running at STOCK because of this. I hate that.....It will remain that way until I determine a need to change the parameters. When I do, they will return to their normal production, and perhaps more if I can clock higher. I think I'm somewhat like you on this, I require 24/7/365 pain free clocks. I want these machines up always. Tortoise and the Hair....What looks fast at first, doesn't win the race.

2. I have seen no deterioration over time, as short as it has been, with D7. The times I have seen it were due to me, mechanically blowing the mount by playing "gorilla" on it, thinking I had not made contact. Long term versus good Shin Etsu mounts, I will report on. I have some machines I will be cracking into that are rock solid reliable with that paste.

Regards Victor,
Bob

[D_A]

I'd guess you'd be right, so long as there is a significant difference between idle and loaded temps it should do. Might just need a few more cycles. Given what I've read here about that RAM I can understand your reluctance to put even the tiniest provocation to it. "Touchy" really doesn't describe that stuff.

[littleowl]

Here is a little update to what was happening to my main rig.
This was my 10 min test
Test 4
room temp down just a little to 29.4
compound IDC7

Idle:
Core 0:38
Core 1:38
Core 2:35
Core 3:32

Load:
Core 0:47
Core 1:47
Core 2:45
Core 3:43

ok right now my room temp is 22.2 and here is what I had after 2.5 days of running on the IDC7
core 0:46
core 1:46
core 2:44
core 3:42

I shut down and took a pic of the mound but my camera is junk. I am getting a new one wed. (woot)
I had very good stread and it all looked great. so i changed the chip from a b3 quad to a G0 quad. put another pea sized drop on the new chip and slapped the fuzion on. My temps right now full load with room temp being 22.2c are
core 0:38
core 1:34
core 2:31
core 3:30

all I can say to that is wow.. everything seems ok. I will leave it running and one of my days off I will try and mess around and make sure all is well.

[Thund3rb1rd]

I would imagine if you leave speed step activated, and then idle, you'll get near ambient temps on your chip.

[G.Foyle]

Hmmm, thermal cycling.

I have thought about that, too. I am making 8 hour cycles, with 4hours of load and 4 hours of idle, measuring temps at the end of 4th and 8th hour. I'll heve results for Q6600 in several hours.

[littleowl]

I would imagine if you leave speed step activated, and then idle, you'll get near ambient temps on your chip.

I don't have speed stop turned on. it is disabled unless It changed it without me telling it to. i will have to recheck it to be 100% sure

[64dragon]

As to manually doing it, that might be significant work....particularly if we are talking about my entire farm....I'm lazy by nature, that's why I became an engineer, to invent things to allow me to take a nap.

thats why I went into engineering (mechanical) too, i'll be graduating in May, now i just need to find a job

back on topic, i'm gona play with my AS5 since my "razor blade" method isn't the best. i'll try the pea blob. imo it will be interesting to see the difference in spread from my ultima90 since it only uses 2 screws