PDA

View Full Version : Start Over?


jaydee
03-30-2008, 09:14 AM
Testing discussion was the end of ProCooling in my opinion a few years back. Re-reading some of those threads leads me to people (testers/reviewers) looking for different results and ultimately giving up.

I never lost the desire to test and review coolers but I did loose the desire to argue about how to go about it and what constitutes usable results and exactly what the target was.

I have thought about it a lot the last few months and gone as far as starting http://www.customcooledpc.com back up and start reviews going. I was thinking about submitting them here but I think the goals are different.

It seems to me analytical testing was the goal here and nothing else mattered or was good enough. The question being good enough for WHO? In my opinion the goal we were setting was far and away past review level testing. Sure we want to do a better review and we did get many reviewers to upgrade their testing methods and equipment. However the end results still remain pretty consistent. No matter what style or equipment used the end results still came out pretty much the same. The order of what coolers are best seemed to stay unchanged.

Also the time to do the testing we were discussing is not really an option for most people even without lifes. :D So what I propose is lowing the standards for reviews and start getting some done. Continue the more analytical testing as desired but separate it from the standard reviews. The gear some of us have is already superior to most review sites so why not make use of it?

One reason I feel less guilty for suggesting such is modern CPU's seem much less effected by better cooling. Especially in the 2-5C range which most liquid coolers differences fall in. So there really isn't any major necessity for such a high level of testing for standard reviews.

Here is a quote from Robotech (Lee Garbutt) about is new perspective on testing.


Editor’s note: Long time readers may remember that I used to do analytical waterblock testing where I used a custom built waterblock test bench to accurately measure each block’s thermal resistance using a thermal die simulator and a lot of fancy equipment. I also measured the pressure drop versus flow rate for each waterblock like you see above. Well I don’t do that anymore for several reasons. What started out as a challenge and learning experience turned into an obsession that required way more time, money and space than I was willing to keep investing. At about the same time, the incorporation of Integrated Heat Spreaders (IHS) on both Intel and AMD CPUs leveled the playing field considerably. On the test bench I could measure slight differences in performance between waterblocks but when placed on a real-world CPU, the results were virtually the same.

Yes, there are performance differences between different brands and styles of waterblocks. However, in addition to raw performance what really differentiates one waterblock from another are things like quality materials and construction, ease of installation and especially the mounting hardware. Actual waterblock performance will in most cases be affected more by the installed clamping force than the subtle design differences inside. So with that new perspective, our current reviews focus less on simulated performance and more on real-world application.
Source: http://www.pcper.com/article.php?aid=479&type=expert&pid=5

Also air coolers are now so good there should be plenty of air coolers to review without all the water cooling gear needed. Only reason not to get reviews and content going is lack of interest or laziness.

I am going to continue reviews on my site with my equipment. However it would be good to see ProCooling get more active with reviews and articles.

Thoughts?

bigben2k
03-30-2008, 12:10 PM
Well, I'm sorry to say that I took part in that too, and I probably didn't help any either. I have stacks and stacks of measuring instruments, but I do plan on putting it to use.

Honestly JD, I think you were on the right track with measuring Max Overclock.

Hopefully, what we went through has taught us that there is still a lot of variables, and what they are. For example, the cooling effect is going to depend on the exact position of the core on the chip (so you can't just swap processors, and call it all the same). Also, if you're going to do this on a regular basis, take some time and re-test some blocks that you tested previously, once in a while, to make sure that your measurements are still the same (processors do wear out).

I put my testbench on hold because of the expense, but also because I needed to refocus on what I was trying to accomplish.

jaydee
03-30-2008, 01:11 PM
Well, I'm sorry to say that I took part in that too, and I probably didn't help any either. I have stacks and stacks of measuring instruments, but I do plan on putting it to use.

Honestly JD, I think you were on the right track with measuring Max Overclock.

Hopefully, what we went through has taught us that there is still a lot of variables, and what they are. For example, the cooling effect is going to depend on the exact position of the core on the chip (so you can't just swap processors, and call it all the same). Also, if you're going to do this on a regular basis, take some time and re-test some blocks that you tested previously, once in a while, to make sure that your measurements are still the same (processors do wear out).

I put my testbench on hold because of the expense, but also because I needed to refocus on what I was trying to accomplish.
I have tried the CPU/Mobo but find it real difficult to unmount the blocks. The large IHS combined with the thermal compound keeps ripping the the CPU out of the socket when trying to pull it off. The pins on my test CPU are pretty worn just after a few mounts.

This last week I have spent a lot of time comparing my new die sim results with the results from CPU combo. Here are the results with the Danger Den Maze 4 and AquaXtreme MP-05-SP LE.

Maze 4
AMD Sempron A64 2500+ Stock 1400mhz 62watts
@2GPM 11C
@1.5GPM 11C
@ 1GPM 12C
@ .5GPM 13C

Maze 4
AMD Sempron A64 2500+ 1862mhz approx. 82watts
@2GPM 14C
@1.5GPM 15C
@ 1GPM 15C
@ .5GPM 16C

MP-05 SP LE
AMD Sempron A64 2500+ Stock 1400mhz 62watts
@1.5GPM 8C
@ 1GPM 9C
@ .5GPM 9C

MP-05 SP LE
AMD Sempron A64 2500+ 1862mhz approx. 82watts
@1.5GPM 10C
@ 1GPM 11C
@ .5GPM 12C

Maze 4
72 watt 14mm square aluminum die
@2GPM 16.1C
@1.5GPM 16.5C
@ 1GPM 17.2C
@ .5GPM 19.0C

MP-05 SP LE
72 watt 14mm square aluminum die
@1.5GPM 13.5C
@ 1GPM 13.7C
@ .5GPM 14.4C

The results are pretty damn consistent between the 2. Close enough for me to justify using the die sim not only because the results are consistent but also not ruining a CPU after a short time. Also mounting is easier on the die sim and more accurate with with my Cole Parmer (http://www.coleparmer.com/catalog/product_view.asp?sku=0850216) thermometer and YSI probes.

Therefor I think either a CPU/mobo combo or a die sim would be fine. I am thinking a Intel socket 755 pinless for the CPU/mobo combo though as those are pinless CPU's. On the bad side of that Intel has some serious thermal protection. The throttling may make your overclock an unreliable measurement if that is what you want to go by.

I think you are better off testing your CPU at various overclocks using the stock cooler that comes with the CPU and calculate the approximate wattage used and repeat those on every block/cooler tested. You can also find the max overclock for the hell of it.

In other words do it how you want as the results will be pretty much be the same in the end. The order of which block is best will not vary much. That is advice for reviewers. If you still want to design coolers then you will want to get a lot more precise I imagine. However at this point in the game trying to design new coolers is pretty pointless.