Test report Apogee GTZ vs. Koolance 350 vs. HeatKiller 3.0

[LOUISSSSS]

if the gtz is considered (kind of) a restrictive block, and it did so well in this test w/ multiple blocks and restrictions, i'll call it a winner!

the performance + the ease of mounting won it over the EK Supreme / K350 when i was deciding on my purchase. (the HK 3.0 didn't have reviews out at that time)

[santiagodraco]

if the gtz is considered (kind of) a restrictive block, and it did so well in this test w/ multiple blocks and restrictions, i'll call it a winner!

the performance + the ease of mounting won it over the EK Supreme / K350 when i was deciding on my purchase. (the HK 3.0 didn't have reviews out at that time)

Glad you like your GTZ.

Of course most realize that this particular test was suited towards the strengths of the GTZ over other blocks, as demonstrated by other tests that show different results. For example if you put greater flow against the 350, and (as stated previously) put it in a single CPU loop, then the results would be quite different. Of course we won't see a manufacturer conducted test that is that unbiased, which is why it's good we have other tests to refer to that are more... applicable to real world implementations.

[LOUISSSSS]

well gabe's test is not useless at all. many of us here indeed have multiple blocks and multiple radiators. such as me, i have a gpu + 2 radiators... would the gtz be the better block for me over a k350? i think so. the HK3.0 maybe? but its damned expensive $85 last i checked.

[santiagodraco]

well gabe's test is not useless at all. many of us here indeed have multiple blocks and multiple radiators. such as me, i have a gpu + 2 radiators... would the gtz be the better block for me over a k350? i think so. the HK3.0 maybe? but its damned expensive $85 last i checked.

Sorry, I don't mean to say it's useless. Under certain conditions I'm sure the GTZ will outperform the other blocks, say the conditions of this test. Your conditions probably apply, hell mine do too maybe But there are also plenty of other conditions that apply to users here, such as say single high pressure loops where this would not be true, as other tests have demonstrated.

But I'd bet that a single multi pump multi rad loop is much less common than say a multiple loop system with 1 cpu loop and a gpu/other loop, using the same components. It might make for an interesting poll anyway. I wonder if anyone has taken any statistical data on what types of configs XS users are using?

[hellcamino]

A single pump would have made a lot more sense for this test and been much closer to the average system...even on XS.

[ColonelCain]

A single pump would have made a lot more sense for this test and been much closer to the average system...even on XS.

You see, that is where the problem lies. IIRC, last test that gabe conducted, he was critisized because he used a "standard" set-up, and not an extreme one.

Extreme set-up, is best if you are trying to eliminate all other variables.

Though, for 90% of us "average" users on XS, this is not what we want to see.

We want to see what you are describing above. Unfortunatly, a great % of the people in the LC section who are the most "active" don't have "average" set-ups.

[/END OFF TOPIC]

[nikhsub1]

Could you explain the differences please?

By "Delrin block", are you talking about the LT, or the cheaper LC?

Please explain...

As far as I knew, the 3.0 Cu and the 3.0 LT were the same, except for the tops.

And the LC version has fewer fins.

Is there further difference b/t the Cu and LT?

Yes I misspoke... I thought the delrin top was ONLY the LC but I was wrong. They do this just to confuser me!

[Freaky Freezer]

Glad you like your GTZ.

Of course most realize that this particular test was suited towards the strengths of the GTZ over other blocks, as demonstrated by other tests that show different results. For example if you put greater flow against the 350, and (as stated previously) put it in a single CPU loop, then the results would be quite different. Of course we won't see a manufacturer conducted test that is that unbiased, which is why it's good we have other tests to refer to that are more... applicable to real world implementations.

I don't see how a CPU only loop is more applicable to real world implementations.

A single pump would have made a lot more sense for this test and been much closer to the average system...even on XS.

Agreed.

You see, that is where the problem lies. IIRC, last test that gabe conducted, he was critisized because he used a "standard" set-up, and not an extreme one.

Extreme set-up, is best if you are trying to eliminate all other variables.

Though, for 90% of us "average" users on XS, this is not what we want to see.

We want to see what you are describing above. Unfortunatly, a great % of the people in the LC section who are the most "active" don't have "average" set-ups.

[/END OFF TOPIC]

His first comparisonof these blocks was on a less "Xtreme" bench and yes he was indeed criticized for it.

[Martinm210]

One way to overcome the whole test with this setup or that setup is to test each block with varied amounts of pumping power. Maybe something like the delta T between water and core average vs Water Horsepower. You could then label specific points of water horsepower to identify specific pumps like one D5, one DDC with top, two DDC's.

Here is the water horsepower equation,
http://www.ajdesigner.com/phppump/pu...orse_power.php

all you need is the pressure drop at the block and the flow rate. And if you already have the pressure drop curve, then all you need is the as tested flow rate and you should be able to convert this data and simply test at several flow rates.

Then you could compare the blocks fairly over this water horspower axis and see the complete picture in how pumping power affects performance and where the optimizations are. It's more work and may make it more difficult to understand, but it would provide the data to answer all the "What if I had x pumping power" types of questions. It would be kind of like a c/w vs flow rate, but something better in regards to using pumping power instead of flow rate. Using flow rate will inherently favor the more restrictive blocks, but using pumping water horsepower should fix that problem. Also using delta T (core-water) is something a bit easier to picture than c/w.

Anyhow, that's something I've always wanted to try, but never got around to it.

[hellcamino]

You see, that is where the problem lies. IIRC, last test that gabe conducted, he was critisized because he used a "standard" set-up, and not an extreme one.

Extreme set-up, is best if you are trying to eliminate all other variables.

Though, for 90% of us "average" users on XS, this is not what we want to see.

We want to see what you are describing above. Unfortunatly, a great % of the people in the LC section who are the most "active" don't have "average" set-ups.

[/END OFF TOPIC]

His last test was done using a bench system with all variables artificially controlled for tight testing conditions, he then "arbitrarily" set his flow/pressure to a variable that it would almost take an iwaki to achieve....He had me interested until he got to that point.

My initial suggestion was based on curiosity and made a lot of sense from a marketing perspective: use a Swiftech branded pump, radiator, reservoir, cpu block and chipset/gpu block such as an MCW60. Test the blocks on a bench like that and you have real world comparison where he can definitively show that his stuff is better (if it is) by switching out the cpu block only for testing.

Being as many people use Swiftech products I would think this test would make great sense and you could still get all the info you were after in the first place too. Using paired pumps with these blocks just makes the results closer than they should be and will even skew them somewhat from what the average person would see with their single pump system.

[RayderR6]

Gabe, you had me at "painless and repeatable" and it's 1 of 3 reasons why I bought your GTZ many months ago.

It's surprising how close (ballpark) your setup is to mine, but I couldn't run my varios at 5 or I'd have a washing machine in my res .

Thanks for your work. I am also curious about cpu block only, will be patiently waiting.

[voigts]

While I understand the hesitancy at relying on a manufacturer's own tests like this posted by Gabe, I think some people go to the point of having a conspiracy theory. Not that I'm a Swiftech fanboy, but Gabe could simply not post his testing results and leave out this info for us all. Also, where is Koolance or Dtek or any other manufacturer when it comes to being willing to put their own products on the line in their labs and publish the results? I think it is great for Hondacity and others to take of their time and money to run tests for us all to benefit from, and I also appreciate Gabe being willing to put his products on the line as well.

[silverphoenix]

One way to overcome the whole test with this setup or that setup is to test each block with varied amounts of pumping power. Maybe something like the delta T between water and core average vs Water Horsepower. You could then label specific points of water horsepower to identify specific pumps like one D5, one DDC with top, two DDC's.

Here is the water horsepower equation,
http://www.ajdesigner.com/phppump/pu...orse_power.php

all you need is the pressure drop at the block and the flow rate. And if you already have the pressure drop curve, then all you need is the as tested flow rate and you should be able to convert this data and simply test at several flow rates.

Then you could compare the blocks fairly over this water horspower axis and see the complete picture in how pumping power affects performance and where the optimizations are. It's more work and may make it more difficult to understand, but it would provide the data to answer all the "What if I had x pumping power" types of questions. It would be kind of like a c/w vs flow rate, but something better in regards to using pumping power instead of flow rate. Using flow rate will inherently favor the more restrictive blocks, but using pumping water horsepower should fix that problem. Also using delta T (core-water) is something a bit easier to picture than c/w.

Anyhow, that's something I've always wanted to try, but never got around to it.

Darn it Martin!!! lol I was quietly planning on a test setup to measure just that. Gathering info on equipment and such. Technically if only our brains could interpret it intuitively, a blocks performance should be graphed in 3 axis on a 3D graph representing, flow, temperature, and heat load, LOL. I was going to PM you on my ideas, well I still may. Testing equipment is not a cheap thing for a college student!!

[veedubfreak]

I recently switched from a GTZ to a koolance 350, and i have remounted it at least 3 times due to chip changes, and each time has been consistent. It's not that hard to get a good mount with it. Personally i'm still pissed that i killed my last chip due to getting a leak from the recessed rings on top of the GTZ and thats the reason i switched to the 350.

[nikhsub1]

One way to overcome the whole test with this setup or that setup is to test each block with varied amounts of pumping power. Maybe something like the delta T between water and core average vs Water Horsepower. You could then label specific points of water horsepower to identify specific pumps like one D5, one DDC with top, two DDC's.

Here is the water horsepower equation,
http://www.ajdesigner.com/phppump/pu...orse_power.php

all you need is the pressure drop at the block and the flow rate. And if you already have the pressure drop curve, then all you need is the as tested flow rate and you should be able to convert this data and simply test at several flow rates.

Then you could compare the blocks fairly over this water horspower axis and see the complete picture in how pumping power affects performance and where the optimizations are. It's more work and may make it more difficult to understand, but it would provide the data to answer all the "What if I had x pumping power" types of questions. It would be kind of like a c/w vs flow rate, but something better in regards to using pumping power instead of flow rate. Using flow rate will inherently favor the more restrictive blocks, but using pumping water horsepower should fix that problem. Also using delta T (core-water) is something a bit easier to picture than c/w.

Anyhow, that's something I've always wanted to try, but never got around to it.

I swear Cathar is 5 years ahead of his time.

http://forums.procooling.com/vbb/showthread.php?t=10859

[saaya]

2 pumps and 2 tripple rads are a little overpowered for a single cpu single vga rig, i havent seen anybody run that combo... how about a single pump and single rad?
i know you mean well, and i trust you, but its a bit odd to see a manufacturer compare his own product vs the competition... i really recommend you to let somebody else do the reviews...

[shazza]

2 pumps and 2 tripple rads are a little overpowered for a single cpu single vga rig, i havent seen anybody run that combo... how about a single pump and single rad?
i know you mean well, and i trust you, but its a bit odd to see a manufacturer compare his own product vs the competition... i really recommend you to let somebody else do the reviews...

Other people do their own reviews. Seems to me Gabe is just supplying additional test data ... people can choose to use it or not. I find it interesting.

While it would be nice to see a single CPU only test, as well as a CPU + GPU loop with a triple rad, and one with two triple rads, and one with a quad rad ... it will never end because there are so many variations

[gabe]

Glad you like your GTZ.

Of course most realize that this particular test was suited towards the strengths of the GTZ over other blocks, as demonstrated by other tests that show different results. For example if you put greater flow against the 350, and (as stated previously) put it in a single CPU loop, then the results would be quite different. Of course we won't see a manufacturer conducted test that is that unbiased, which is why it's good we have other tests to refer to that are more... applicable to real world implementations.

You could not be more wrong

1/this particular test was simply conducted on the bench I have. it wasn't intended to favor one block or another;
2/ this particular test intention was clearly stated in my introductory paragraph: use an extreme setup similar (which doesn't mean identical) to that of the extreme guys on XS.
3/ the gpu was there because I was also interested in recording GPU temps for unrelated tests, and I opted to keep it in the loop because of 4/
4/ presence of the GPU affects overall flow, and overall temps, but this has very little effect on the differential temps between blocks, and I honestly do not believe that it would affect their ranking.. much more serious issues do that: see below.


The real bomb, I dropped here: http://www.xtremesystems.org/forums/...5&postcount=11, but no one seems to have noticed, or decided to discuss it -at least between members, since I stated that I wouldn't.

[sirheck]

Next cpu water block i will buy will most likely be a GTZ.

[Mech0z]

Well I shouldnt get the GTZ I suppose as I have lapped my Q6600 :&

[Martinm210]

I swear Cathar is 5 years ahead of his time.

http://forums.procooling.com/vbb/showthread.php?t=10859

That's because it is his idea, only thing I would change is to use either water/core delta or the actual core temperature adjusted to some set ambient level...probably best just to use core temp corrected. I like C/W, but I think to convey it in a way that makes sense to a larger audience using something like core temperature is better understood. I think it would also be key to label the specific points that match up with common pumping setups, etc.

Anyhow, just an idea. I've been too lazy to spend the time on it myself, but thought it was a great idea.

Testing with one set pump or pumping setup is great, but now that we've seen some blocks are more flow sensitive than others, I think you really almost need that level of testing to get the full picture.

[Vapor]

I might as well just add the graph to my testing, I have all the ingredients

[Mekrel]

Going to buy a HK block for my new i7 setup, simply because I can use any barbs I would like to use with it. Unlike the the recess on the GTZ.

[Martinm210]

I might as well just add the graph to my testing, I have all the ingredients

I think so, if you have pressure drop curves...I haven't tried it, but it seems like a good way to look at it from a common pumping power perspective. If you labeled each pumping power point it would be pretty easy to understand.

[Vapor]

Yeah, I already have individual graphs for comparison at each pumping power....so this is just a reformation of that.

Getting the X-axis right will be tough (have to know pressure drop curves, which I don't have right now).