Innovation Cooling's Diamond 7 TIM test results

[Bail_w]

So here's a longer list, and then you can apologize for accusing me of fabricating data.

http://www.frostytech.com/top5heatsinks.cfm
http://www.anandtech.com/casecooling...spx?i=3268&p=5

Ya, the Arctic freezer 7 Pro outperforms the TRUE. Like i am going to believe that.


http://www.xtremesystems.org/forums/...d.php?t=158187

Its linked, so just click it.

[Movieman]

Guys, this started out as being a good thread. It's quickly deteriorated. This discussion is not productive in any way.

If all of you want to talk about CPU coolers, I suggest you move the discussion to the Air Cooling section.

Thank you. I feel like a referee in a slugfest where people ignore the rules.

[tastymannatees]

This is 1 of a number of charts we will be throwing up on our website in the next couple of days. I thought you would be interested as you guys are included.

These are the average results of 5 forum give away/beta tests with 127 users tests reported. Some of the more exaggerated improvements over the average might be attributed to a poor initial mount although at least anecdotally as I have not pulled the numbers yet as I want a little larger sample seem to be caused by the pump or bake out of the previously installed compound.

On the other side of the average line the small , zero or negative results at least as far as I can suggest might be attributed to more competitive compounds, misfired mounts, misread temps or variations in hardware, software Etc.

As most of you guys are pretty rigorous about your testing I am taking the Ockham's razor approach, lex parsimoniae. 9 out of ten times when I have a problem with temps it is usually a contact/pressure problem. I'll rework the sink mount and get a 5 degree drop, But on a more subtle basis if I fine tune it I can get a few more degrees out of it under perfect conditions.

The more subtle aspects to getting that last mile in ideal temps is awkward at best and subject to nebulous debates but this is the applied effort you people have been driving for and I am much appreciating it.

The contact/pressure testing will give us some good data whether my guess proves right or wrong the info will I think be useful to users across the board whatever compound they use.

[David_L6]

This is 1 of a number of charts we will be throwing up on our website in the next couple of days. I thought you would be interested as you guys are included.

These are the average results of 5 forum give away/beta tests with 112 users tests reported. Some of the more exaggerated improvements over the average might be attributed to a poor initial mount although at least anecdotally as I have not pulled the numbers yet as I want a little larger sample seem to be caused by the pump or bake out of the previously installed compound.

On the other side of the average line the small , zero or negative results at least as far as I can suggest might be attributed to more competitive compounds, misfired mounts, misread temps or variations in hardware, software Etc.

As most of you guys are pretty rigorous about your testing I am taking the Ockham's razor approach, lex parsimoniae. 9 out of ten times when I have a problem with temps it is usually a contact/pressure problem. I'll rework the sink mount and get a 5 degree drop, But on a more subtle basis if I fine tune it I can get a few more degrees out of it under perfect conditions.

The more subtle aspects to getting that last mile in ideal temps is awkward at best and subject to nebulous debates but this is the applied effort you people have been driving for and I am much appreciating it.

The contact/pressure testing will give us some good data whether my guess proves right or wrong the info will I think be useful to users across the board whatever compound they use.

That's very close to my average improvement. On the two computers that I've tried the Diamond 7 on - both quad cores - one TRUE, one Zalman 9700 - the average is a 2.75°C improvement.

Will try to get test started on another computer tomorrow.

[littleowl]

That is about the same here! I am close to 3c drop in temps also.

[tastymannatees]

That is about the same here! I am close to 3c drop in temps also.

Like I said we tend to do better against the 1+ month cured pastes and about 3C against the fresh install comparisons which is pretty close to what I get on a bench test @ 100W obviously if you are running 150W the delta would be higher around 5 C and 50W 1-2 C.

There are some very competitive performance pastes out there but I do not expect them to really test any better performance wise I had one formulation of our grease that had performance of 10WmK and was only .5C better @ 100 W Plus when you measure the resistance across the compound with thermocouple in-sink and in-die the temp drop was only 1.1-1.2 C @ 100W - Not much left to improve upon except the reliability/longevity

On the initial completed giveaways the averages have not changed much since the initial sample of around 30. Averaging is great stuff and the data suggests solutions like application amount, methods for optimization, trouble shooting etc.

You guys (this group )have given me a good education of the current state of the art on mounting hardware and several potential options for refining our message and trouble shooting guide. You are prime candidates for doing a similar beta test on our upcoming Thermosyphon. Good stuff, thanks

[Martijn]

Like I said we tend to do better against the 1+ month cured pastes and about 3C against the fresh install comparisons which is pretty close to what I get on a bench test @ 100W obviously if you are running 150W the delta would be higher around 5 C and 50W 1-2 C.

There are some very competitive performance pastes out there but I do not expect them to really test any better performance wise I had one formulation of our grease that had performance of 10WmK and was only .5C better @ 100 W Plus when you measure the resistance across the compound with thermocouple in-sink and in-die the temp drop was only 1.1-1.2 C @ 100W - Not much left to improve upon except the reliability/longevity

On the initial completed giveaways the averages have not changed much since the initial sample of around 30. Averaging is great stuff and the data suggests solutions like application amount, methods for optimization, trouble shooting etc.

You guys (this group )have given me a good education of the current state of the art on mounting hardware and several potential options for refining our message and trouble shooting guide. You are prime candidates for doing a similar beta test on our upcoming Thermosyphon. Good stuff, thanks

Thermosiphon (alt. thermosyphon) refers to a method of passive heat exchange based on natural convection which circulates liquid in a vertical closed-loop circuit without requiring a conventional pump. Its intended purpose is to simplify the pumping of liquid and/or heat transfer, by avoiding the cost and complexity of a conventional liquid pump.

That sounds interesting. Care to explain, or aren't you allowed to disclose any information yet?

[SiGfever]

If the heatsink you use is not capable of moving enough heat, then the best TIM in the world will not have a chance to perform. I hope that's clear enough for you.

DDTUNG

This is the best analogy that I have heard. The heatsink should NEVER be the limiting factor, it should be capable of removing more than the test can throw at it so the TIM is the only factor in limiting the heat transfer. The numbers will be smaller but more accurate when it comes to top rated TIM.

So many have tested with AS5 as their standard and my past testing has shown me that MX-2 outperforms AS5 by a good bit so D7 should always triumph over it. MX-2 will give it a better run for its money and in my case it has bested D7 with each test but I will continue testing and follow through with the additional pressure test.

I also have some T-C Grease 0098 that I have been wanting to test and after the D7 I will run it under the same conditions.

And as has been stated this is a TEST Thread and not a personal battle ground for members. Please either keep on topic or do not post! Take heatsink discussions to the proper forum section as S_B mentioned unless it directly applies to your application of D7. Thank You.

[littleowl]

Like I said we tend to do better against the 1+ month cured pastes and about 3C against the fresh install comparisons which is pretty close to what I get on a bench test @ 100W obviously if you are running 150W the delta would be higher around 5 C and 50W 1-2 C.

There are some very competitive performance pastes out there but I do not expect them to really test any better performance wise I had one formulation of our grease that had performance of 10WmK and was only .5C better @ 100 W Plus when you measure the resistance across the compound with thermocouple in-sink and in-die the temp drop was only 1.1-1.2 C @ 100W - Not much left to improve upon except the reliability/longevity

On the initial completed giveaways the averages have not changed much since the initial sample of around 30. Averaging is great stuff and the data suggests solutions like application amount, methods for optimization, trouble shooting etc.

You guys (this group )have given me a good education of the current state of the art on mounting hardware and several potential options for refining our message and trouble shooting guide. You are prime candidates for doing a similar beta test on our upcoming Thermosyphon. Good stuff, thanks

I am also interested. can you tell us more?

[D_A]

Yes, please do. Though I suppose we should start a new thread for it, to be consistent.

[tastymannatees]

I am also interested. can you tell us more?

Off and on again project. Advantages are cost and size, we might make a few units for the performance market but that's not the target, performance as good as any tower and that's about it and all I can say at this point. This group with a beta test would be a good way to work out any kinks as you guys are the TIM and sink experts.

I also got the 24 carat tubes today and should be available by the end of the week.

[STEvil]

Been an interesting read.

I must point out that contrary to an earlier post Intel's thermal testing methods use a thermal probe inlaid into a channel milled into the top of the IHS of an engineering sample CPU.


I'm not saying that the methods being used by anyone are incorrect.

[tastymannatees]

Been an interesting read.

I must point out that contrary to an earlier post Intel's thermal testing methods use a thermal probe inlaid into a channel milled into the top of the IHS of an engineering sample CPU.


I'm not saying that the methods being used by anyone are incorrect.

That's what I use along with several other methods, the more the merrier

I even cut an IHS for the guys at HardwareLogic recently, they are using thermocouples now. Should be a good reference for the software monitoring programs

A well rounded thermal test set up in my opinion would encompass thermal diode, thermocouple mount on IHS and sink and synthetic die and what the heck add some pressure film in to boot

[64dragon]

Like I said we tend to do better against the 1+ month cured pastes and about 3C against the fresh install comparisons which is pretty close to what I get on a bench test @ 100W obviously if you are running 150W the delta would be higher around 5 C and 50W 1-2 C.

that is a huge statement, i completley ignored/forgot about processor wattage, my x2 4200 is a 65W version and your statement explains completely why i'm not seeing an improvement but no more than 1-2C worse during fluctuation (i'll summarize all my results as soon as i can from each paste - AS5, Shin Etsu and D7)

[tastymannatees]

that is a huge statement, i completley ignored/forgot about processor wattage, my x2 4200 is a 65W version and your statement explains completely why i'm not seeing an improvement but no more than 1-2C worse during fluctuation (i'll summarize all my results as soon as i can from each paste - AS5, Shin Etsu and D7)

I was going to ask the forum members when I first started these beta tests to supply the wattages but it is kind of a pita. I have sampled some of the wattages but will I go back later and do them all, the advantage being I can then generate a chart of wattage vs thermal performance for the average user.

acquiring the data is tough But the information is valuable i.e. I should be able to generate other graphs showing things like thermal performance vs reliability over time, thermal performance vs pressure etc.

My ultimate goal is to have the average user log onto our site view a few simple charts like the one I posted earlier and leave with a full understanding of how our compound works and what to expect if he would apply it in his own system. Maybe with that knowledge in hand I make another sale which is the ultimate goal.

[STEvil]

That's what I use along with several other methods, the more the merrier

I even cut an IHS for the guys at HardwareLogic recently, they are using thermocouples now. Should be a good reference for the software monitoring programs

A well rounded thermal test set up in my opinion would encompass thermal diode, thermocouple mount on IHS and sink and synthetic die and what the heck add some pressure film in to boot

What kind of thermocouple and is it controlled correctly?

The reason I ask is because thermocouples have a tendency to become more or less efficient depending on the temperature they are operating at. If you have a power supply which monitors and controls power output you can adjust to hold total power consumption constant which helps with that a bit (not perfect but more reliable than hooking it up and flying blind which can and will skew results).

[tastymannatees]

What kind of thermocouple and is it controlled correctly?

The reason I ask is because thermocouples have a tendency to become more or less efficient depending on the temperature they are operating at. If you have a power supply which monitors and controls power output you can adjust to hold total power consumption constant which helps with that a bit (not perfect but more reliable than hooking it up and flying blind which can and will skew results).

T Type and test to Intel spec. and have many metered power supplies. It is interesting that Intel recommends their method when doing any testing of consequence rather then their own internal diode. But they seem OK in a general sense lacking resolution in finer measurements on an individual basis they seem to average out within a couple of 10th's of a degree to what I am seeing here to what I get on an absolute number on a bench test.

Maybe I am simple minded but that closeness with the averaging has held from day one with just a couple of tenths fluctuation either way struck me as pretty awesome.

[fart_plume]

Anything works at the scale that the circuitry of these new cpus is amazing let alone being consistant, I mean 45-65 nm I can even think of an analogy that would be a good comparison,lol

[Martijn]

Anything works at the scale that the circuitry of these new cpus is amazing let alone being consistant, I mean 45-65 nm I can even think of an analogy that would be a good comparison,lol

The protein hemoglobin in a red blood cell is around 4.5nm. You've got millions and millions of these in your body. Gives you something to think about.

[G.Foyle]

About the wattage, I am measuring it with Zalman's MFC-2 fan controller; I have found that it fluctuates ±15 watts depending on WU type. I have not checked which WUs are most power hungry. The downside of this method is that I am measuring total power consumption, not the CPU alone.

[tastymannatees]

You guys are now famous although I do not think IC gets the page hits XS does

http://www.innovationcooling.com/article1a.htm

[tastymannatees]

I also had a thought on a poorman's or DIY contact test. coat the IHS with graphite(soft pencil ?) put a piece of paper, maybe tissue paper 1.5X 1.5 in. between the IHS and sink and clamp. Should show where the CPU is making contact and where it is not. Kind of like taking a rubbing off an old tombstone

Or better yet use some carbon paper from the old typewriter days and a clean piece of white paper

Might be interesting to compare to the pressure film result

[fart_plume]

The best type of carbon paper to use for this purpose is NCR paper. It's designed for pressure and heat it's used alot in trades to make tracings.

[tastymannatees]

The best type of carbon paper to use for this purpose is NCR paper. It's designed for pressure and heat it's used alot in trades to make tracings.

I'll try and see how it works

[fart_plume]

If you have trouble finding it I have a butt load of it I used to test gold foil imprint stamping.I know this stuff I have works from 30*f to 400*f.