Innovation Cooling's Diamond 7 TIM test results

[123bob]

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

Hmmm, good idea, I'll try it. It normally goes against my grain to have anything slow my machines down. I automatically disable speed step, etc when I set up a machine. Good catch, thx.

Bob

EDIT: Great idea T-bird. Temps now swing 6+ degrees. It's driving CPUz and coretemp nuts, but hey this is science...Thx

[Movieman]

First impressions:
WOW!
Specs:
Q6600 @3600 (9x400)
Asus P5K Premium(Black Pearl)
memory:2x1 gig Crucial Ballistix Tracers DDR2-8500
PSU:Corsair HX550
Cooling TRUE w/zalman 55CFM 120mm fan

Ambient temps are 68F
"Old" temps running WCG with all settings the same using AS5
49-49-48-47
New temps running WCG:
45-45-45-44
and this is not burned in..
3 mins after initial boot and right into WCG after checking system temps,etc.
TIM applied as I always do on the Quad: A medium "line horizontally and the same vertically and "seat" and move the HS 15 degrees side to side.
That's a average 3.5C decrease across the cores from what I thought was a very respectable temp.
I'm sold.

Forgot: idle temps at 3600mhz were in the high 20's.. Amazing!

[phelan1777]

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

Imagine if the whole team was reporting in? I bet you would have to do a week with out sleep, just to record half the data :P.

There is reason that less then 300 effective, can hold with teams of 1K or more members.

Even if some of us only have one Qcore running atm (me for example).

[Emerica]

First impressions:
WOW!
Specs:
Q6600 @3600 (9x400)
Asus P5K Premium(Black Pearl)
memory:2x1 gig Crucial Ballistix Tracers DDR2-8500
PSU:Corsair HX550
Cooling TRUE w/zalman 55CFM 120mm fan

Ambient temps are 68F
"Old" temps running WCG with all settings the same using AS5
49-49-48-47
New temps running WCG:
45-45-45-44
and this is not burned in..
3 mins after initial boot and right into WCG after checking system temps,etc.
TIM applied as I always do on the Quad: A medium "line horizontally and the same vertically and "seat" and move the HS 15 degrees side to side.
That's a average 3.5C decrease across the cores from what I thought was a very respectable temp.
I'm sold.

Damn MM, I wish you had a picture of the way you apply TIM because it looks like it worked out for you. I still can't picture it though. So you made like a Red Cross sign with the TIM?

[Movieman]

Damn MM, I wish you had a picture of the way you apply TIM because it looks like it worked out for you. I still can't picture it though. So you made like a Red Cross sign with the TIM?

Are you catholic?
Cross yourself!
Horizontal line and then a vertical line.

[tastymannatees]

I ordered the pressure film today. I should have it by Friday.

They are going to send me a digital color chart which we can post and participants can scan a picture, take a picture for comparison.

Alternatively, and I would probably do this if I get enough people, is send the samples back to the company and get an in depth digital analysis of each test(Not quite ready to spring for the full Monty test equipment- $12,000) But the company analysis is quite reasonable if the qty is large enough.

[Bail_w]

I ordered the pressure film today. I should have it by Friday.

They are going to send me a digital color chart which we can post and participants can scan a picture, take a picture for comparison.

Alternatively, and I would probably do this if I get enough people, is send the samples back to the company and get an in depth digital analysis of each test(Not quite ready to spring for the full Monty test equipment- $12,000) But the company analysis is quite reasonable if the qty is large enough.

well how many people do you need and have so far?

[tastymannatees]

well how many people do you need and have so far?

I have about 3 or 4, I would like 60 to 80 which is why I will probably revisit some of the other forums and get a full cross section. Maybe I can put a Library of images with corresponding results on our website so people can take a test on their own then log on to compare to see what most closely matches their own test result

[Deadeye550]

Sign me up for the pressure test as well. I haven't had a chance to test the TIM yet, but I've got Friday off and I will do the testing then.

Wesley

[tastymannatees]

Damn MM, I wish you had a picture of the way you apply TIM because it looks like it worked out for you. I still can't picture it though. So you made like a Red Cross sign with the TIM?

Check the slide application pics.

http://overclockers.com/articles1529/

Reasoning behind it is basically as the paste spreads when it hits the middle between the 2 corners it now has a place to go due to lower resistance.

The corner to corner spread is thicker and actually forms a visual "X" due to the increased distance/resistance it has to flow. Some people do an X corner to corner and some like the + middle to middle between corners which is my preference

[brot]

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....

Well, i could help you with that. But you would need to give a linux test installation a try, but i am sure you have one harddisk lying arround.

in linux you start and stop services with init scripts, wich is /etc/init.d/boinc in that case. to start, you will have to run as root "/etc/init.d/boinc start", to stop its "/etc/init.d/boinc stop".

now linux has somthing called cron, to quote wikipedia: "cron is a time-based scheduling service in Unix-like computer operating systems."

so you just add the following lines to the /etc/crontab file:

0 * * * * /etc/init.d/boinc start
30 * * * * /etc/init.d/boinc stop

Which results in boinc getting started every 0th Minute, and stopped every 30th minute, which should be enough time to cool down and heat up.

If you want to try this and need help, feel free to ask
(ps: maybe there is the possibility to write such a script which remote controls other clients, and thus only needs to run in a virtual machine, i could try out that tomorrow )

[123bob]

Well, i could help you with that. But you would need to give a linux test installation a try, but i am sure you have one harddisk lying arround.

in linux you start and stop services with init scripts, wich is /etc/init.d/boinc in that case. to start, you will have to run as root "/etc/init.d/boinc start", to stop its "/etc/init.d/boinc stop".

now linux has somthing called cron, to quote wikipedia: "cron is a time-based scheduling service in Unix-like computer operating systems."

so you just add the following lines to the /etc/crontab file:

0 * * * * /etc/init.d/boinc start
30 * * * * /etc/init.d/boinc stop

Which results in boinc getting started every 0th Minute, and stopped every 30th minute, which should be enough time to cool down and heat up.

If you want to try this and need help, feel free to ask
(ps: maybe there is the possibility to write such a script which remote controls other clients, and thus only needs to run in a virtual machine, i could try out that tomorrow )

Thx Brot! I may be onto something with this thermal cycling idea. I don't know if the script will be required yet. The 50% method may be working.

I've got good AS5 Cure results to post about later tonight. I'm using the 50% CPU utilization profile and T-bird's recommendation on turning EIST (speed step)back on. Coretemp shows about a 5-6 degree variance between cycles. It may actually be more since CT reads at 1 second intervals.

It appears to be cycling the temp enough to cure the AS5 test in about 6-8 hours of cycling time. It dropped the AS5 load temps about 6 degrees!! I couldn't believe it so I restarted the machine, to see if it was a fluke. No it wasn't. I rechecked everything out and it was a valid test.... More detail and pics later.

I'm running D7 on that machine now. It's initial temps are better than AS5's initial temps. We'll see what cycling does to it by tonight....

BTW, I made SURE I had a good blob on this time and was careful about how I put on the TRUE. It should be a good mount with the D7 this time.

Regards,
Bob

[Jaco]

That avatar always cracks me up
Brot is back ,firing Linux bullets from his mini-gun at Bob !

[brot]

That avatar always cracks me up
Brot is back ,firing Linux bullets from his mini-gun at Bob !

I decided that minigun is old-school, compared to a light saber

[Martijn]

I decided that minigun is old-school, compared to a light saber

LOL! No way

[64dragon]

thanks to tasty for posting the link to overclockers, i've learned that my razor blade method of spreading TIM is bad. so i decided to redo my AS5 (while i wait for D7 and Shin Etsu to arrive for more comparison) and use the pea blob. heres a few pics of the razor spreading and pea spreading with some twisting.

specs: ihs and hsf lapped 800grit
amd x2 4200 @ stock 2.2ghz 1.25 vcore but lowered multi to 9 and raised fsb to 245 to OC the ram to 882

first off, a pic of the thermalright ultima 90 mounting mechanism. screws are all the way down. from now on, and as i did for the pea blob, i'm tightening both screws at the same time


razor spreading applied around x-mas 07
idle temps: forgot to record, when is my cpu idle since i've started crunching
WCG 100% load temps with ambient of 19C
core 1: 30-32C
core 2: 38-42C

heatsink


cpu



blob


i mounted the hsf, let it sit for a minute or 2 and then pulled it off to see if i applied enough and check how it spread

heatsink


cpu


spread looks very nice and way different from the razor method. i cleaned all that off and applied a new blob basicly the same size

pea spreading - cure time of 8hrs no thermal cycling
idle temps: (i dont think these changed from TIM spreading method)
core 1: 23C
core 2: 29C
WCG 100% load temps with ambient of 19C
core 1: 29-30C
core 2: 36-38C

just by changing my mounting method i've gained 2-4C!!!

[tastymannatees]

One other suggestion on application. I have been doing accelerated testing with glass slides thermal cycling. In the razor blade method you can see how the bubbles are introduced into the paste. These bubbles with heat and pressure expand and are a source of failure in any grease long term. I have seen some lapse into lima bean sized holes.

When you pull the sink off to check spread and then reattach the low points on the paste (when you reassemble) can trap air and look like the razor mount. best to just follow through

That is a great spread by the way

Also note on the old mount the mirrored bare spot on both sink and IHS - Pump Out?

[Movieman]

After 16 hours at 100% load on WCG I'm seeing another degree of drop in temp:
43-44-44-43
vs the "old" standard of:
49-48-48-47
Definately a winner and my thanks to you Mr. Tastymannatees
( love that name)
Took an inventive mind to think that one up!

[64dragon]

Also note on the old mount the mirrored bare spot on both sink and IHS - Pump Out?

do you mean the bare spot in the center or outside edge? what do you mean by pump out?

[D_A]

'pump out' is where the TIM gets gradually pushed out from between the CPU and cooler due to expansion of air bubbles with heat. Thermal cycling turns it into a 'pumping' process.
ie small bubble heats up and pushes out TIM a little. As assembly cools, rather than pull the TIM back in, the slight vacuum helps to pull the cooler closer to the CPU, closing the gap. When it heats up again the air expands again and pushes more TIM out, then when it all cools the assembly clamps closer together again with even less TIM. This leaves pockets of low pressure air between the cooler and CPU. The lack of TIM gives poor thermal contact but any pockets of vacuum are insulators. End result is a whole bunch of 'not good'.

[tastymannatees]

do you mean the bare spot in the center or outside edge? what do you mean by pump out?

D_A pretty well nailed it.

The bare spot in the sink and the mirror of the same spot on center. Usually when you pull the sink and IHS apart some will stick to each, you should be able to match up voids on each and have a full coverage. But when both are bare means no paste on that spot. It may have been there sometimes hard to tell from a picture.

[SiGfever]

Update: Original Mount: 04/06/08 12:45pm

04/08/08
17:59:00
23.2c ambient

Idle:
Core 0: 34
Core 1: 30
Core 2: 34
Core 3: 31

Load:
Core 0: 45
Core 1: 44
Core 2: 45
Core 3: 45

[tastymannatees]

Anymore takers on the pressure film test?

[Movieman]

24 hours in temps have stabilized across the cores at 44C
What truly amazes me is that all 4 cores are essentially reporting the same temps.. Yes, will see a back and forth of one degree but virtually all identical.
This says to me that this stuff truly "levels" out when seated.
To say I am impressed would be an understatement and as much as I hate pulling good working machines down I'm going to pull down the clovers and apply this stuff.
Mr. Mannatees, you have a winner and if you need someone to stand and say that publically just yell.

[SiGfever]

24 hours in temps have stabilized across the cores at 44C
What truly amazes me is that all 4 cores are essentially reporting the same temps.. Yes, will see a back and forth of one degree but virtually all identical.
This says to me that this stuff truly "levels" out when seated.
To say I am impressed would be an understatement and as much as I hate pulling good working machines down I'm going to pull down the clovers and apply this stuff.
Mr. Mannatees, you have a winner and if you need someone to stand and say that publically just yell.

"Elizabeth, I coming honey". <holds heart>