Informal test results CPU 350 vs Apogee GTZ

[nikhsub1]

What happens in the real world?
If you take each and us in a commonly used setup like a MP655 pump and a MCR320 rad, which is the better or more effective block?
Thanks.

Much the same as what you are seeing. Obviously the flow rate difference will be less dramatic with a lesser pump like a 655. What I am sure of is that the 1GPM difference in the Swiftech tests accentuate the temp differences between the 2 blocks much more than lesser pumping power.

I don't understand what everyone is complaining about the GPM numbers. From my understanding, those are the results measured in the loop with same pump pressure. Its telling you WHAT impact each block has on the loop.

So putting 350 in your loop reduces your GPM to 1.33

Using a GTZ reduces your GPM to 2.33

What don't you people understand? its showing how restrictive they are. So the GTZ hurts your flow less AND gives you better temp (I'm a little skeptical but we'll see skinee's results, etc.)

I think (hope at least) they are starting to get it.

[gzipper]

Much the same as what you are seeing. Obviously the flow rate difference will be less dramatic with a lesser pump like a 655. What I am sure of is that the 1GPM difference in the Swiftech tests accentuate the temp differences between the 2 blocks much more than lesser pumping power.

I think (hope at least) they are starting to get it.

If you also had the preassure drop over the two blocks (negating tubing/bending etc) you would be able to derivate the individual impedance from each block. Then when you have done that you could use the blocks impedance in a pressure/flow chart for a specific pump to calculate what flow you could expect from that particular pump/block combo.

Gabe could you derivate that?

[skinnee]

Yes, trying to pump the same flow rate and inlet pressure into the block is next to impossible and would do nothing for the test other than to represent a false reality. As gabe (well, Stephan actually) did was start with the same flow rate and pressure (this is an educated guess) at the outlet of the pump.

All tests are good to see, each test report I read I can take at least one thing away from it. That is why each and every test is valuable to the community whether you agree with the results or not.

[voigts]

I don't understand what everyone is complaining about the GPM numbers. From my understanding, those are the results measured in the loop with same pump pressure. Its telling you WHAT impact each block has on the loop.

So putting 350 in your loop reduces your GPM to 1.33

Using a GTZ reduces your GPM to 2.33

What don't you people understand? its showing how restrictive they are. So the GTZ hurts your flow less AND gives you better temp (I'm a little skeptical but we'll see skinee's results, etc.)

Thank you Gabe for sharing your results. It is interesting to know the pressure drop difference.

Why some of you are ranting on and on about your opinions on testing is beyond me when others take their own time to test and publish the results to try to benefit others. I haven't tested anything, and therefore really have no grounds to bash any body else's testing.

Not to say Swiftech is perfect, but for them to intentionally skew results would only be to shoot themselves in the foot. They haven't become the quality company they have by screwing their customers and trying to deceive about their parts performance numbers.

[hellcamino]

What happens in the real world?
If you take each and us in a commonly used setup like a MP655 pump and a MCR320 rad, which is the better or more effective block?
Thanks.

+1 These are the only type of tests that actually matter.

[warriorpoet]

+1 These are the only type of tests that actually matter.

No, they are not, and saying they are is not only inaccurate, it is counterproductive.

[Kaldor]

The problem with real world tests is they have too many variables. Unless everyone uses the same test gear, same pump/rad/tube/blahblahblah, to test with, the results can always be different. I think gabe hit it on the head here. Use something industrial, that can output the same results, flow and temp, time after time, and thats a true test.

Id like Swiftech to get their hands on every CPU block on the market, test them all the same, removing all variables. That would be a true test. But you can hardly expect gabe to collect all the different manufacturers blocks and test them for us now can we?

[NaeKuh]

*passing popcorn to skinnee*

*sigh* evertime theres a comparison thread, drama always seems to follow...




Gabe tell stephan thanks for his time!

I for one apreciate it a lot.

[andersson.j]

Thank you Gabe for sharing and keep up the good work.

One thing that I don't think anyone has mentioned yet is that in a real loop flow rate affects radiator performance. Since a chiller was used in this test the water temperature was the same with both blocks. In a real loop the less restrictive GTZ would leave more flow to the radiator which in turn would perform better. I don't know how much this would increase GTZ lead though?

[hellcamino]

No, they are not, and saying they are is not only inaccurate, it is counterproductive.

The reason I said that is simple, the average person watercooling their pc uses a single pump, radiator, res/T-line and cpu block. Swiftech radiators are very prevalent, DDC3.2 and D5 pumps are also very prevalent. A simple demonstration of results achieved with a combination of these parts would be great information! If you use a common configuration such as either of these parts and a MCR220 or MCR320 in Gabe's case changing out only the cpu block during your testing you can definitively show which block performs better with each pump, the radiators go without saying. If all that this test was intended to accomplish was which block was more restrictive I would say it did so admirably but it tells nothing about actual capabilities of the blocks using common pumps and radiators.

[Zehnsucht]

Thank you Gabe for testing!
Is it possible we could see some pictures of your test equipment? It's sounds mighty impressive and controlled

[Webby]

The reason I said that is simple, the average person watercooling their pc uses a single pump, radiator, res/T-line and cpu block. Swiftech radiators are very prevalent, DDC3.2 and D5 pumps are also very prevalent. A simple demonstration of results achieved with a combination of these parts would be great information! If you use a common configuration such as either of these parts and a MCR220 or MCR320 in Gabe's case changing out only the cpu block during your testing you can definitively show which block performs better with each pump, the radiators go without saying. If all that this test was intended to accomplish was which block was more restrictive I would say it did so admirably but it tells nothing about actual capabilities of the blocks using common pumps and radiators.

I think you are missing the point of the test, the test is not see which block works best in an arbitary system made of common watercooling components it is which CPU block performs the best.

If you want to test a CPU block you need to eliminate all other variables the way in which Swiftech have achieved this is by using a pump set to a particular flowrate via control valves which is then left constant for all tests and they have used a chiller to eliminate any variations a radiator could introduce (like for example changes in ambient temperature).

Yes you can not say if I bought a blah blah and a blah I would get this temperature but then you couldn't say that anyway as every block, radiator, pump, fan and not to mention CPU will be different.

What do you expect to see if a DDC 3.2 and a 360mm rad are used? The flowrates may be closer together but the GTZ will still be less restrictive which will help with heat dissipation in the radiator the is no logical way in which changing to a different pump and a rad rather than a chiller will switch the result it might get closer but it shouldn't switch.

Oh and thanks for the testing Gabe

[smee]

*passing popcorn to skinnee*

*sigh* evertime theres a comparison thread, drama always seems to follow...




Gabe tell stephan thanks for his time!

I for one apreciate it a lot.

I agree 100% Naek!

Gabe, Stephan, thank you.

[santiagodraco]

No, they are not, and saying they are is not only inaccurate, it is counterproductive.

Yes They Are.

Anyone here who says that a manufacturer who does 1) their own tests and 2) does it using lab equipment and not real world equipment is ok... is either a hypocrite or is sitting smack dab on one side of the fence.

Do you see legitimate reviewers doing this? No.

And again, I refer you to the lynch mob and Koolance's radiator tests.

[santiagodraco]

The problem with real world tests is they have too many variables. Unless everyone uses the same test gear, same pump/rad/tube/blahblahblah, to test with, the results can always be different. I think gabe hit it on the head here. Use something industrial, that can output the same results, flow and temp, time after time, and thats a true test.

Id like Swiftech to get their hands on every CPU block on the market, test them all the same, removing all variables. That would be a true test. But you can hardly expect gabe to collect all the different manufacturers blocks and test them for us now can we?

Translation. Let the vendor run the tests with a single set of variables that may or may not benefit said vendor over another said vendor. Gotcha.

And I can't believe you, or anyone, would think that having a setup comprised of components readily available to consumers (such as say the two most popular pumps and the two most popular radiators) would be introducing unreasonable variables? What variables exactly? How tight the compression fittings are?

It boils down to one thing and one thing only. While Gabe's tests may be valid, or not, until it's conducted in a real world environment it's going to be questioned.... just as most of you have done in the past, very vocally I might add

I'm sure Gabe is used to the way things work around here. I am not saying Gabe is dishonest, but as the vendor here he shares a much greater burden to be as completely unbiased and to demonstrate the greatest amount of flexibility in order to have his results treated as meaningful. Otherwise it's just another test showing that one block is more restrictive than the other which is NOT necessarily indicative of poorer performance, hell it might be indicative of better performance under specific conditions such as running a higher powered pump over a lower powered pump.

[Archmage]

The testing circumstances seem valid.

I'd like to see a test on an overclocked and overvolted i7 920. 4Ghz would be good.

You could actually reduce the flow-rate to about 2GPM on the GTZ (and whatever that yields on the 350) to add another test, but I don't see this as necessary.

The other thing I might suggest is coolant temperatures at 25C for another data point. I don't know why this should matter, but it'd be easy enough to adjust quickly.

A final point: there may be temperature differences when the pump is placed directly before the block (more restrictive blocks may improve?). This could be tested, or at least we should know the flow-order of the current setup.

That about covers it. I can't really argue with this test, but I can remain skeptical because it contradicts other tests.

[SiGfever]

Just because Gabe has a controlled environment and a professional testing setup by no means invalidates his testing. He has always been known for quality products and is a valued member of this forum.

And I still do not understand why so many cry "We need real world testing" when that is not how controlled scientific testing is done. You have a controlled environment, regulated heat and or cooling sources and professional meters and gauges. Being able to closely control every aspect of the testing is what validates it.

[NaeKuh]

Funny thing is half these people dont even know how to read gabe's data.




That is why im ROFLing.

[billb]

Thank you Gabe for sharing and keep up the good work.

[Pellepel]

Much the same as what you are seeing. Obviously the flow rate difference will be less dramatic with a lesser pump like a 655. What I am sure of is that the 1GPM difference in the Swiftech tests accentuate the temp differences between the 2 blocks much more than lesser pumping power.

I think (hope at least) they are starting to get it.

Yup I agree with the methodology, but how much temp difference would it be between the two blocks say if the industrial pump were regulated to pumping power of a stock D5 or DDC3.2 with top?

[ranker]

Yes They Are.

Anyone here who says that a manufacturer who does 1) their own tests and 2) does it using lab equipment and not real world equipment is ok... is either a hypocrite or is sitting smack dab on one side of the fence.

Do you see legitimate reviewers doing this? No.

And again, I refer you to the lynch mob and Koolance's radiator tests.

Because not only did Koolance outright lie about their product's performance, but they also ran tests where not even .000001% of the population can claim to have comparable equipment.

However, doing a review of your past posts, I can tell you're a Koolance fanboy or Koolance shill. Any test that would put Koolance behind any other product would be a faulty test in your eyes.

Read the data. I'm appalled that you can't even interpret the data.

[Webby]

Yup I agree with the methodology, but how much temp difference would it be between the two blocks say if the industrial pump were regulated to pumping power of a stock D5 or DDC3.2 with top?

Without running the test nobody can say for certain but I would expect the difference in flowrate to become smaller and with it the temperature difference although the use of the chiller eliminates the greatest advantage of the GTZ which is higher flowrates through the radiators especially if the flows were reduced.

[SNiiPE_DoGG]

you guys are idiots, you realize that if the gtz gets 1.86 from the ddc3.2 then the 350 is gonna be down to ~.86 gpm and its just gonna perform worse

[coolmiester]

Like most other individuals have said, the info is without doubt appreciated but now Martin has gone on to do other things we need a new bench marks.


I know skinnee is working on it but until the benchmark is "benched" there is little or no point in bringing numbers to the table that don't mean too much against others imo so lets get some ground rules and work off them.

[Pedalmonkey]

i love how the guys that i have seen nothing of, with very low post count, and one with koolance branded stuff all over his sig trying to push around the true everlasting knowitalls of the liquid cooling world. Gabes data, if his test was done to what he said ( i belive it was), is infact less biased than any test you could possibly produce.