OCUK Official IC Diamond/ Perihelion Test Results

IC Diamond said:
OK, so the data was from a previous comparison? That's no problem just trying to understand the test
Click to expand...

Yes, I monitored temps with MX-2 from mid Nov 2010 to mid Dec then changed to AS5 (MX-2 temps had risen to high 60's), began comparing the results & by mid Jan had a month's worth of data for the 2 under the same conditions. When your offer came along to test Diamond/Perihelion it seemed logical to simply switch TIM's & compare the results against the other 2.
 
Actually pretty much the way we do our comparisons, multiple test dies for months and comparisons made on the same die. We record the data every morning to get a trend line up or down. have to be rigorous about ambient's

You and WingZero30 should be in R&D as your interest lies in that direction.
 
Going over the numbers and the return is pretty good so far with 60+% reporting in with good commentary/observations.

I will post up the ongoing results in a day or two, takes time to wade through couple hundred posts. Good Job all.

Though some will make it to the Wall Of Shame the return is good enough to date to open up another 20 samples for giveaway.
 
I'll post what results I have at the moment in that case, but I want to do more mounts to get an average but I'm busy with work so I don't have much time to sit down at the moment and do extended thorough testing:-

Test system
i5 2500k @ 4.4GHz
CPU cooled by Apogee XT
580GTX in loop
Passive water cooling

AS5
Credit card method
Ambient - 19.5C
Idle - 30C
Average peak load 56C
Delta = 36.5C

IC Diamond
Blob method
Ambient - 20C
Idle - 30.5C
Average peak load 58.25C
Delta = 38.25

Peak load temperatures were measured using OCCT CPU stress tool and recording peak temperature values in Realtemp for the 4cores and taking an average. Both measurements are after 2hours of burn in using OCCT.
 
Informational request-

Going over the numbers and found an interesting trend. Most, not all of you included a heat sink description - manufacturer, model, type of cooling etc. To get the full picture I need to fill the blanks if you could edit your posts or post your sink info would be a great help.

Much appreciated

Andrew
 
Chart below has a statistical cluster marked in red. Points 38 and 39 are zero but I marked them 0.1 to highlight them otherwise they are not visible.

The red marked data points all have a something(s) in common - 5 free ICD24 tubes to the first one who who can guess what I am looking at.

you get one guess each

overclockersUKDatasort.png
 
IC Diamond said:
Chart below has a statistical cluster marked in red. Points 38 and 39 are zero but I marked them 0.1 to highlight them otherwise they are not visible.

The red marked data points all have a something(s) in common - 5 free ICD24 tubes to the first one who who can guess what I am looking at.

you get one guess each
Click to expand...


Is it different cpu type, i.e red being AMD and blue being Intel?

If any more samples are released for testing and reporting back, please include myself, missed out first time lol.
 
Last edited:
Jokester said:
So is the chart showing that the spread of results for watercooling is much tight than for say aircooling?
Click to expand...


No I do not believe so, C/W is C/W and not going to change.

I have another data point that's within range -3C I did not post because I wanted to highlight the cluster.

You can see from the data results that as a group there was a pretty rigorous effort to get it right as far as application/pressure this probably goes for the water users as well.

May be a weakness in hardware. I was thinking of flipping the coin and looking to see if there are standout heat sinks with exceptional hardware mounting hardware for increased delta's which is why I asked for users heat sink details.

There are pros and con's for each product, for example ThermalRight makes great sinks but the ones with the bowed base in our giveaways often tests great and sometimes it does not(not sink, our compound - more on that later) although it has the best overall pressure I have seen. I have Ideas about it but some were concerned enough in Germany to do a petition for TR to do away with it, Often these things are subtle and will try the patience of the best of mechanics.

http://www.ipetitions.com/petition/thermalright-convex/

Each giveaway has it's own variables, Power users vs the undervolters, some forums will use 70% AS5 others might use mostly stock compounds or they might favor a particular sink.

OCUK favors a higher % of water cooling and this is after 18-20 of these surveys. In the past I would get a couple water samples per giveaway and some were good and some were not. I tried to do some focused sampling but still inconclusive as size was small.

Sample size here is small but it may indicate where the trend is going - I believe it does.

In any event here is some H50 C/P data from a guy who was seeing only a degree or so difference. Contact was light, pressure was on the edge etc. Not a good test, but many do see a good result. It is hard to tele troubleshoot these things and at the same time many have a good result from the H50

test3-h50.png
 
Hardware: I7 920 DO [email protected], XSPC RS360 Rad+Vipers, EK Supreme HF CU, EK 5870 FC Block, D4 Vario on Min, 1/22 pipe, 800D.

All tests with full fans also made sure temps started around 30-33'c

LinX Load Testing 15K Problem size, 20 runs ( 18 Mins )

Tested 4 TIMS Noctua NT H1, MX4, Perihelion, IC24

Noctua NT H1 - FAIL
LinX produced GFLOP ranges from 35 - 48. This I believe was caused by Thermal Throttling. Everything appeared to be OK after removing the block and I had been using the system for about 4 months.

Perihelion
Peak Load - After 24 Hours
64,60,61,55
Peak Load - After 48 Hours
64,59,61,55

Retested these temps are more like it!

MX4
Peak Load - After 24 Hours
63,59,61,54
Peak Load - After 48 Hours
62,58,59,53
Really like MX4, very easy to spread. Good even spread after removing the block.

IC24
Peak Load - After 24 Hours
62,58,60,54
Peak Load - After 48 Hours
61,57,59,53

In my system IC24 performs very similar to MX4 with IC24 having a slight edge. Perihelion now looks good after re testing. To me all the compounds are performing around the same. +- 2'c is pretty good.

End of the day I think care applying and mounting is key. If you get it wrong it doesn't matter how good the paste is.

Performance over time. I suspect the IC stuff will have the edge here.

Many thanks for allowing me to take part in the testing. I know it's a marketing excerise but it was still interesting if a tad time consuming.

All the best with a great product.
 
Thanks for taking the time to test it is much appreciated by us.

Yes it is a marketing promotion but is also a educational process for us that helps us refine our marketing message and improve on our troubleshooting on user problems. Application method and amount were determined from user tests and further refined by IC.

It continues to reveal valuable information even after 20 of of these exercises.

The Asgard -Thanks again for the extra effort
 
rawimgb.jpg
rawimga.jpg


From the onsite Review

Testing/Troubleshooting Notes

Picture the test results above as coming from 2 different website reviews with identical systems and identical heat sinks and run under the same conditions.

1st reviewer sees a 1 C difference

2nd reviewer sees 3 C difference

Not to dither here on the details of testing but which one is more accurate?

They are both equally accurate as the reviewers both reported what they saw.

But on a single test you have to Quantify your result to Qualify your result.

Both were tested @ 75lbs + (TR sinks have some awesome pressure!)

and if the reviewers both report contact area of 0.64 and 1.14 square inches respectively the tests are now comparable with the obvious conclusion that better contact improves performance. (Most here on this board are aware of this as experienced end users, I re-emphasize so everyone gets it, simple point but a key one)


In this test the lapping exercise improved performance for both compounds but ICD -2 C more so than the AS5 for overall of -3 C.

Low pressure and light contact area tends to homogenize the differences between compounds so that differences of perhaps maybe only +/- 1-2 C over an array of several tested compounds marginalizing test results within the range of the margin of error.

So to hammer the Point it is just good shop practice to insure good contact and tighten the sinks down
 
Latest update - Any errors or omissions on my part let me know and I will make the change.

Well so far we have an interesting set of data that separates OCUK from all the 20 or so groups tested by us.

In the past water cooling is usually less than 5% of the sampled results, on OCUK at this point we are 30% water cooling(marked in red), 70% air! You guys are unique.

What made it readily obvious was the statistical cluster of 20% water in the marginal-zero-negative result category. The other 10% is mostly the H50 in the performed as expected group. Might be linked to compound mix but I have a sense that it may have an overall edge in C/P for the H50. (Curse you water coolers for sinking my averages!,lol)

In any event The air cooler group (marked blue) has it over the water with pretty much positive results across the board. I highlighted the ThermalRight (yellow) sinks as they were generally lower but the sample is small and may not be telling us anything and as in water cooling the overall performance is good enough that they accept results as they are. I believe the TR convex base has merit but as you push limits you reach a hysteresis point where minor differences can put you off a degree or two but again probably mitigated by sink overall performance.

I would be interested in a sub test with the water group if I can get enough of a sample size using the contact and pressure paper same as what was used in the review. Any takers?

Perihelion will be revised separately







OCUKMARCH282011update.png


OCUKMarch27updatechart.png
 
Last edited:
IC Diamond said:
Thanks for taking the time to test it is much appreciated by us.

Yes it is a marketing promotion but is also a educational process for us that helps us refine our marketing message and improve on our troubleshooting on user problems. Application method and amount were determined from user tests and further refined by IC.

It continues to reveal valuable information even after 20 of of these exercises.

The Asgard -Thanks again for the extra effort
Click to expand...

No problem and my pleasure. :)
 
Some mistakes in the update for my result.

1. My paste was MX-2 not MX-3.
2. You left out the idle difference. MX-2 was 1 degree cooler than ICD24.
3. Load temps difference is also wrong. The average ICD24 is 47.25 degrees not 47 degrees meaning ICD24 was only 1 degree better, not 1.25 degrees.
4. Waterblock is a EK Supreme HF not the older Supreme.
5. You left out my cpu and speed as well. I5 760@4Ghz @1.228v.

This info is all in my results post.
 
You must log in or register to reply here.
Back
Top Bottom