[H] Enthusiast Waterblock Testing

FrgMstr

FrgMstr

Just Plain Mean
Staff member
Joined
May 18, 1997
Messages
55,719
OK guys [H]ardOCP is asking for your input. This is your chance to be heard and make a difference!

We have been working on a waterblock shootout and have decided that we are going to improve our format for testing liquid cooling products.

We are going to first create a cooling system that we will use as a baseline. When we review a product be it a waterblock, pump, radiator, or just some fancy new hose we can simply replace the part from our baseline system and record the resulting difference. This format continues the “real world” testing methodology that is the hallmark of a [H]ardOCP review. A ton of charts and graphs are great and can be full of useful information but all that matters in the end is performance.

What do you want to see in our water cooling reviews? What liquid cooling products are you most interested in seeing? What are you totally sick of seeing in reviews, be it ours or others?

Let it out but keep it on topic please.
 
If you are going to setup a baseline I would do more then one. Have like a low end setup, low noise hig end setup, and a high preformance setup.

For complete kits easy of use and whatnot.
 
I would love to see some Euro blocks tested on [H]ard|OCP.

I like your idea of using a test system as the baseline and as long as the same motherboard and CPU are used on each different system. Maybe go with a hella hot quad cord so the baseline can hold some weight for awhile.

Here is what I, as a consumer, really care about:

1. How quite is the entire loop
2. How well can I overclock
3. Can I have multiple blocks in the loop without killing performance?
4. Will the watercooling completely fill my case with clutter
5. Looks, looks, looks
6. How easy is it to get installed
7. Temps because if 1-6 are good, who really cares?
 
I do NOT want to see a bunch of Swiftech kit reviews. In fact, I'm not sure kits are what anyone wants to see reviewed. A bunch of other websites do kits, and if someone's shopping for a kit, they probably aren't [H]ard enough to look at a ton of reviews.

I wanna see a block shootout, with different flow rates; a pump shootout, on a couple of different blocks; a rad shootout, with a couple of different fans. Tubing comparisons, fittings, etc. would be great as well. Please please tell me how much each 90* elbow adds to my temps, etc.

Thanks, this is really needed.
 
If there was a higher precision way of measuring temps, that would be much better. CPU thermistors aren't exactly known for their accuracy.
 
I would like to see the classic Bonneville heatercore (no AC) included as part of any rad comparisons and flow testing. IMO its still the best option for a guy on a budget. (Clean the hell out of it before using it your tests. )

Do you have any thoughts yet as to what size or how you will test configurations with different tubing sizes ?
 
I concurr with the above posters. Please don't make this into a swiftech/danger den love fest.

My ideal test bed (goal would be to use the best of the best hardware wise so overclocks can't be blamed on anything other than how well it cools the CPU):

Computer Hardware
Intel QX6700
Any highend975/680i motherboard
Corsair Domenator RAM (2 x 1gb)
PCPC 1kw Power Supply
Latest and fastest video card

Baseline Cooling
Thermochill PA120.2 Radiator
Petra DDC pump w/ custom top
3/8" or 1/2" tubing


Here are the blocks I would like to see.

1. AquaComputer
2. EK
3. Dtek
4. Innovatech
5. AlphaCool
6. Newest Zalman
7. Koolance

End all be all would be to have charts showing how introducing a GPU block affects the overall performance of the loop.
 
Do a round up of CPU blocks, GPU blocks, pumps, and radiators on a Core 2 Duo or Quad Core test bed and I think I'll marry you guys.

Examples:
DD TDX, Swiftech Apogee/GT, AquaXtreme MP-05 Limited Edition, DD RBX, D-Tek FuZion, Zalman ZM-WB4 (new gold water block), hell, maybe even throw in an AquaComputer Cuplex in there, etc. etc.

For GPU ~ Maze 4, 5, AquaXtreme MP-1, Swiftech MCW-55, MCW-60, Full Cover blocks from EK, and the rest are redundant.

Rads and Pumps: Black Ice Pro and Xtreme 1-3, Swiftech MCR series, Thermochill PA and HE series, Coolrad Ts' and a Dual 120MM Bonneville Heatercore

D5, DDC 1 + 2, Ehiem 1250, D4, blah blah blah

Both of those AquaXtreme blocks really needed to be tested more often (their results are usually really superb) All I see are Maze 4s and MCW-60s nowadays.

I know this is a pipe dream, but, oh well.
 
It's no pipe dream. The whole reason we are asking is to get info just like that ^
A lot of what you guys are saying confirms what I thought already about what you want to see. You guys are also giving me some great article ideas and general input/feedback. Loving it so far. On topic even! :eek:
Keep it up! :D
 
I've yet to see a Fuzion reviewed on a Kentsfield. The Fuzion's design is OBVIOUSLY based on focusing efforts right in the center, but Kentsfield (and Presler before it) have no die in the center. I want to know if the Apogee GT does better in this situation. For that matter, maybe an RBX in reverse offers the best performance! :eek:

Disclaimer:
I know the RBX wouldn't do sh!t in that situation, because it loses the impingement design. I was trying to illustrate a point.
 
Idea: Adjustable peltier rig :D turn the peltier up, and see how many watts it takes to get to a set temperature :p (50 or 60? not sure what that is in F ) thats would be great to see how blocks will do in the future, and would be a one and only [H] thing :D
 
Firstly there is a ton of hardcore testing done at Xtremesystems, and I think you guys should really strive to separate yourselves from their individual members to accomplish these reviews. What they do not really do at XS is break stuff up (for noobs and such) like this;

Give a complete list of all things that will be needed for WCing. I think that you should divide the reviews into say three categories:
low budget, midrange budget, and high end budget. (this is common ground, but list the block choices and then show the test results and such

Lets say for example:

High budget: Storm beats out XXX block.
Mid range: Dtek Fuzion/Apogee GT
Low: TDX is still a good block, beats out XXX block.

(sorry if those 'budgets' are not accurate, those are just relative prices than I encountered)

I agree with those ^ who said to skip the kit reviews. Usually the enthusiast will want to mix and match the best parts, and kits dont have all of those bundled.
 
What not to do:

(1) IMO, you have gotten quite lazy relying on EXOS and Nautilus pre-built systems for CPU and GPU testing. I seriously doubt an EXOS or Nautilus could handle any serious Kentsfield O/C.

(2) The BFG 8800 review was another bad example. You completely ignored the piss-poor quality of the DD 'block interior machining, despite posting interior block shots. And the Corsair unit is in no way intended to be used for GPU cooling. You should have slammed the BFG card [H]ard for being an overpriced POS. You could have done much better qualitatively and economically buying the regular BFG (or any other) 8800 GTX card, and an EK 8800 GTX Water block.

Next what to do:

(3) OK, for the W/C noobs out there, do complete W/C loops build at a 2 price points. Just like any system build. Do a detailed analysis of what you bought and why based on the hardware it will be cooling. Example a quad core needs a different 'block than a single or dual core.

(4) Show how to properly build a loop build from scratch, leak testing, etc Tons of pics. Show why a custom loop is better than a pre-built system (Koolance / Corsair).

(5) As far as test beds, get two cases (conventional ATX), not reverse-ATX (Lian Li PC-2000 series). Get a mid tower case that can take at least a 2 x 120mm fan rad, and one large tower case big enough to hold a 3 x 120 mm fan rad. Replace with a smaller rad as needed for testing.

(6) Explain why there is no need for cooling things like hard drives or PSUs.

(7) Show why ¼” based systems are inferior to 3/8” or ½” based systems.
 
Make sure to when doing results to use a Single Die Proccesser(C2D, but not the quads), and use/post Dual Die(Kentsfield, etc) results seprately, because for example, an apooge/ apooge gt will beat a storm on a dual die cpu, but the result will be reversed on a single die cpu, ie the storm will beat the apooge/ apooge gt.
 
A base setup that i would like to see in the review:

Thermochill PA 120.2 , 120.3, Swiftech MCR220 , or MCR320
Swiftech MCP655 (reliable for 24/7 uses and etc) (why not MCP355? because there are many people experiences suddens death and need to go through RMA).
1/2 inch or 7/16inch tubing


and please report some accurate results.
 
I know that "removable" IHSs are on their way out.... so alot depends on WHEN you guys start your WC reviews, but at this point in CPU history, many of us are still using S939 based systems and usually, for those who Watercool, the IHS comes off. I'd like to see test results (for whatever component) for CPUs with and WITHOUT IHSes..... please. :D
 
I agree a lot with what is being said.

I would like to see some comparisons on non-conductive fluids. Which prevent build up and
which do not? What fluids look better in clear tubes? Do some glow better than others? Do
some fluids cool better than others when all else is the same?

I was going to use Fluid XP, but I read some bad things and now I am thinking about using
PrimoChill ICE instead.

It would be nice to know what the best loop order is as well. I am sure there is one.

Pump>CPU>Radiator>Resevoir>Pump
or
Pump>CPU>Resevoir>Radiator>Pump
 
qdemn7 said:
What not to do:

(1) IMO, you have gotten quite lazy relying on EXOS and Nautilus pre-built systems for CPU and GPU testing. I seriously doubt an EXOS or Nautilus could handle any serious Kentsfield O/C.

(2) The BFG 8800 review was another bad example. You completely ignored the piss-poor quality of the DD 'block interior machining, despite posting interior block shots. And the Corsair unit is in no way intended to be used for GPU cooling. You should have slammed the BFG card [H]ard for being an overpriced POS. You could have done much better qualitatively and economically buying the regular BFG (or any other) 8800 GTX card, and an EK 8800 GTX Water block.
Click to expand...

Well you know we are damned if we do, damned if we don't. But do know this, I am not going to start constructing entire systems for CPU, GPU, or any other computer component testing. We don't have the time, resources, and it is hard to get anything repeatable out to any other editors. I wish it was all done here on one table and we did not have to worry about anything else in the world but making the persons that are worried about how water flow is impacted 1% in their block, but that just ain't gonna happen.

If you will sit here and tell me that YOU WILL BUY the next BFGTech video cards with blocks you approve of, I will make sure to pay close attention to these minute points in the future.......but the bottom line is that product is NOT made for you as you are not going to pay a premium for it. So keep in mind, that while you are not happy with the format of that particular comment, it was in NO WAY SHAPE OR FORM DIRECTED AT YOU.

Frankly I would suggest you just don't "get it" and had you read the article fully you could easily grasp that it was not written for the hardcore DIY water cooler. Why would you even begin to think that it was, so why would you hold that article to a different level?

When you seen a video card waterblock comparison done (not a VC buy with a preinstalled and warranted block), then maybe worry about the machineing....which you give your opinion on but I would suggest you have never once put a mechanical gauge on to back up your opinion.

As for your suggestions about the Corsair unit in no way being meant for that duty, I would suggest you have little experience with the new hardware out there as these video cards setups are sometimes pulling many more watts than a Prescott OC setup ever did. I can easily see systems today where the VC needs to be water cooled before the CPU. Look outside of the box bro..

As for Kentsfield OCing, I have had both the Koolance and Nautilus units work fine. Let me ask you this, since you seem to know so much about what I am doing, please share your own personal data to back up the opinions above.

And no reading someone else's uninformed opinion in a public forum hardly makes you an expert on the subject. I rely on personal experience first, then make sure I read others' personal experienes to compare my own to. Your opinions above are rude and uninformed and do nothing buy underline your ignorance about the components in question.

And also, you did not follow instructions as were given to no going off-topic. If you have anything else to say on this, please start another thread or email me personally.
 
I'm a NEWB at posting but I've read tons of useful information her over the years. There are ALOT of bright minds here whose knowledge has helped me go from buying off the shelf PC's to building my own. I say we use this wealth of knowledge and start a poll. Which CPU block, which Radiator, which Pump etc. for the baseline system. Then I would love to see you test everything you could get your hands on in that system. I'm sorry if I got off topic but thats just my 2 cents.
 
R1ckCa1n said:
End all be all would be to have charts showing how introducing a GPU block affects the overall performance of the loop.
Click to expand...
I agree with this. I plan on just cooling my CPU for now, but once I move to a DX10 card, I am thinking about adding the GPU to the loop. Very interested in something like this.

-bZj
 
Looking around on the web, I don't see a lot of reviews that show a very loaded wc system and the effects rads and pumps have on them. Most hook up either a kit as posted before or just show one or two blocks in the loop with a single rad, or focus a lot on flowrate. Past a certain point, flowrate doesn't matter as much.

I would like to see a very high heat rig, like a quad core cpu, dual 8800 gtx vid cards, and maybe even a chipset block in one loop, then see how changing the radiator or pump can help. A lot of times changing those can help a lot more than getting better blocks.

So something along of the lines of seeing the difference between single, dual, and tri 120mm rads and also seeing the effects of higher flowrate (typically AC pumps) and higher pressure (DC) pumps would be very useful.
 
This is just asking for opinions about testing blocks for the next shootout, so a single loop should be fine. As R1ck suggested, a PA 120.2/3, DDC with a Petra's top, and 1/2" tubing should be fine.

In general, more precise readouts would be appreciated. Ease of use for installation and changing barbs/components (impingment plates in TDX/RBX, if applicable) should also be factored in somehow.

For the actual testing, performance is still what gives us numbers to easily compare. Even though quad cores are the latest and greatest, if you can afford one, you could probably go phase as well. It would nice to have two different processors used (not just OC'd/increased voltage, but a different number of cores as well). Basically, use the kind of processors we would likely be using with these blocks.

It would be great to have as wide of a selection as possible, but a quick list should include:
TDX and 6002 (and they have been reviewed before)
Storm
Apogee/Apogee GT
Fuzion
European blocks are fine, but that might involve using 1/2" to 3/8" adapters or restricting barbs to match the thread size.

As for R1ck's other comments, those pertain more to the entire loop, but cosmetics could be included as well. If you wanted to measure the flow/restictiveness of each block, that would be wonderful as well. Including GPU blocks would add another variable to the testing, but as noted, would be greatly appreciated if you guys could do it.
 
What I would like to see is a graphical plot of cooling additive performance, and the percentage added to water. Even compare off-the-shelf non-mixing coolant to homebrew coolant.

For instance, what would be the temperature of a cpu when 5%, 10%, 15% etc (up to whatever) of additive is added to distilled water? People recommend so many differeing percentages of additive on this and other forums. Okay, many people add as much to give them a nice colour, but it would be very interesting to have the facts. If the difference between 5% and 15% was small, you could add the extra for protection (and the colour ;) ).

Another idea you could do is a fan shootout for use on radiators. Use the exact same system (rad, loop etc) and only change the fans. Plot component temperature with noise at a given voltage. 'What fan(s) should I use on my rad' is a common question with no validated winner (that I know of).
 
One thing I would like to see in Radiator testing is performace both in a case and free air test, a lot of people pull warm case air thru the rad which should impact the performance quite a bit compared to less popular mounting methods that provide cooler ambient room air to the rad.
 
My #1 want from [H]ard|OCP waterblock testing would be this :

Create a test bench that uses the Intel TTV spec for measuring the core temperature. Namely you machine a shallow groove into the IHS and put a highly accurate thermal probe there. Then fill in the groove with a special hardening thermal paste. This would take out the gray area of typical thermistor inaccuracies. I don't care about the controlled water temps because you want "real world" testing....but I really really want to avoid built in thermal sensors because they are utter shit.


Secondly, I want a good record of what the ambient temperature was during all of the tests. Instead of showing an absolute temperature on the results, you should instead show the delta between ambient temps and CPU temps.


Now about the actual test-bench. I think it should definitely be a DIY system. For the default test bench you should use a Swiftech Apogee or DD TDX as the CPU block, mainly because these are the most common. For the pump it should be a Laing D5, or DDC. For the radiator a Swiftech MCR-220QP or Black Ice (Pro or GTS) 2x120. If you include a GPU in the baseline then a Swiftech MCW60 or DD Maze 4 should be the block.

Waterblock testing would just switch out the waterblock. When testing a GPU show the difference between with GPU and without, as well as that GPU block against others. Pump testing should ideally use the longest loop you have tested.


One special series of tests I would like to see is a radiator stress test. The focus of this should be on radiator size, and how they perform in various loops. There should be three loops tested with the radiators. Loop 1: CPU only, Loop 2: CPU + 1 GPU, Loop 3: Everything you can do, CPU, SLI/CF GPU's, Chipset. Each of these three loops should be tested with a 120mm, 2x120mm and 3x120mm radiator. This would give people a really good idea of how large a radiator they need to get to achieve the performance they want.
 
What does everyone think of my idea above? i personally would love to see how certain waterblocks scale with an increase in thermal power...
 
0mega said:
What does everyone think of my idea above? i personally would love to see how certain waterblocks scale with an increase in thermal power...
Click to expand...

Its ok, but it doesn't necessarily jive with the philosophy (as I understand it) of [H]ard|OCP testing. What you are suggesting is essentially what the artificial test beds are made out of. They use a copper die connected to a heat source (sometimes adjustable sometimes not). [H] wants to perform real world testing, which means using real processors. Personally I don't like that because of the inability to accurately measure the temperature, which is why I suggested they go through the effort of machining the groove into the IHS for the thermal probe, it would retain the real world element while giving them a much increased accuracy in their testing. I mean, they dont do video card reviews by eyeballing the frames per second.

The other problem is that processors aren't necessarily going to keep increasing in heat output. Right now about 100-130W is as hot as they can get and still work well on air cooling. Smaller processes, and recent breakthroughs in High-K materials will lend to lower power consumption. In fact the problem will likely be an increased thermal density as opposed to overall heat output. I mean right now you have a 100W bulb condensed into an area about the size of your thumbnail.
 
OK guys i have a little info for you about the testbed thus far. First of all we are changing to an overclocked C2D E6300 CPU on a Gigabyte DQ6 v2.0 with the same Corsair XMS2 RAM from the AM2 testbed. I have not OC'd the chip yet as I just posted the machine but I am pretty optomistic. I think this hardware is a bit more representative of what most of what you guys are running now vs. the X2 4200+ rig.
On the watercooling side of things we are going to use a D5 so that we can test at multiple settings without changing out pumps and a Coolrad 22T dual 120mm Radiator. For right now it has a pair of 69CFM fans on it but if we decide to do a rad shootout we will be sure to include multiple fan options.

Here is the part where I really need you all to focus and stay on topic.
No flamebait or trolling OK?

Temp recording...
TAT? Coretemp? I know speedfan wont cut it with the C2D. what is the best/most accurate software option?

Load testing...
TAT again? Orthos? SP2004? Prime95 still?

For right now lets try to concentrate on the waterblock shootout. We will deal with testing other components after this it worked out. I think the first paragraph says it all though, keep talking we are listening.
 
Justin_Diduch said:
Temp recording...
TAT? Coretemp? I know speedfan wont cut it with the C2D. what is the best/most accurate software option?

Load testing...
TAT again? Orthos? SP2004? Prime95 still?
Click to expand...
Only two proper choices, IMO.

Core Temp for temp measurement.

Orthos for Load Testing.

One thing you might want to take a look at is the updated version of Motherboard Monitor 5 with the Core Temp Plugin. See http://www.thetechrepository.com/showthread.php?t=74

Note that some of voltages will not work with all boards. But the fact that Alex has uopdated MBM after quiite a while is hopeful.
 
Justin_Diduch said:
OK guys i have a little info for you about the testbed thus far. First of all we are changing to an overclocked C2D E6300 CPU on a Gigabyte DQ6 v2.0 with the same Corsair XMS2 RAM from the AM2 testbed. I have not OC'd the chip yet as I just posted the machine but I am pretty optomistic. I think this hardware is a bit more representative of what most of what you guys are running now vs. the X2 4200+ rig.
On the watercooling side of things we are going to use a D5 so that we can test at multiple settings without changing out pumps and a Coolrad 22T dual 120mm Radiator. For right now it has a pair of 69CFM fans on it but if we decide to do a rad shootout we will be sure to include multiple fan options.

Here is the part where I really need you all to focus and stay on topic.
No flamebait or trolling OK?

Temp recording...
TAT? Coretemp? I know speedfan wont cut it with the C2D. what is the best/most accurate software option?

Load testing...
TAT again? Orthos? SP2004? Prime95 still?
Click to expand...
The hardware/pump/rad sound fine for an average loop/system that is still affordable for most.

For temperature monitoring, as Erasmus said, a physical measurement will be much more accurate than the onboard probe. However, since most people aren't drilling temp probe holes in their IHSs, Core Temp is widely used and could give us an idea of what temperatures we would see, as compared to what others have seen thus far.
 
Justin_Diduch said:
OK guys i have a little info for you about the testbed thus far. First of all we are changing to an overclocked C2D E6300 CPU on a Gigabyte DQ6 v2.0 with the same Corsair XMS2 RAM from the AM2 testbed. I have not OC'd the chip yet as I just posted the machine but I am pretty optomistic. I think this hardware is a bit more representative of what most of what you guys are running now vs. the X2 4200+ rig.
On the watercooling side of things we are going to use a D5 so that we can test at multiple settings without changing out pumps and a Coolrad 22T dual 120mm Radiator. For right now it has a pair of 69CFM fans on it but if we decide to do a rad shootout we will be sure to include multiple fan options.
Click to expand...

No flamebait but I think [H] really needs to think about using a quad core processor in order to allow for some longevity to the testing.

Justin_Diduch said:
Temp recording...
TAT? Coretemp? I know speedfan wont cut it with the C2D. what is the best/most accurate software option?

Load testing...
TAT again? Orthos? SP2004? Prime95 still?

For right now lets try to concentrate on the waterblock shootout. We will deal with testing other components after this it worked out. I think the first paragraph says it all though, keep talking we are listening.
Click to expand...

TAT would be a good baseline since it seems to cook a CPU moreso than anything else; whatever the means of measuring temps and producing loads needs to stay consistant.

As my poll from last week shows most people who read this forum are interested in quite operation while allowing for overclocking AND not max overclocking so the use of onboard sensors and such should suffice most. I am a fim believer that the waterblock producing the lowest temp on a thermal die isn't always the best block while mounted on a CPU so I like your direction.
 
My only response to the temp monitoring program is what I said earlier. Machine a groove and use a proper thermal probe. I mean if MaxxxRacer over at XS, a single person, can get it done the [H] should have no problem doing it.

For applying a load to the processor I say use Orthos.
 
Kyle and Justin,

I am glad to see that the forums audience is being included in your brainstorming.

Even though I have a zallman resorator and a thermaltake setup here that was given to me, I have stayed away from watercooling for two major reasons:

1. Portability, and

2. Reliability.

There have been some setups that have enticed me in the past, but I decided a while ago if I cannot get a quiet, cool system that I can easily move around then watercooling is not for me.

Since I don't feel like any opinion I can offer is expert, my only request would be to include in this article a top of the line air cooler, like the tuniq tower to base watercooling results on.

Can we also get a price/performance/value graph?
 
Erasmus354 said:
Create a test bench that uses the Intel TTV spec for measuring the core temperature.
Click to expand...
Oh, you mean something like this? :rolleyes:
E6400.jpg
E6400Side.jpg

rigBig.jpg
rigClose.jpg


My little "project" aside, it seems to me that people's ideas of 'real world' testing are utterly ridiculous and provide very little in terms of useful data. By proceeding down the path that you've started on, all you will actually end up doing is collecting a bunch of useless CPU temperature information that applies only to your specific test setup. Stop for a moment and think about this, step back and look at the bigger picture--are you doing this solely for yourselves or are you doing this to really help the water cooling community by providing useful information? If your answer was the latter, then I'll tell you what the water cooling community has needed for a long time...be forewarned, though, it's nothing short of a monumental undertaking (which is why nobody has done it).

What we need is for a single, well thought out, well implemented, well operated, controlled test bench to fully evaluate most (if not all) of the commercially available CPU and GPU waterblocks, radiators, and pumps. CPU and GPU blocks would need to be tested on varying die sizes (i.e. either various sizes of die sim or various TTV-modded CPUs with differing die size(s) and location(s)) and tested for pressure drop vs. flowrate and thermal resistance vs. flowrate (figuring the approx. amount of heat being applied to the DUT with live CPUs would be tricky, at best, though). Radiators would need thermal resistance data gathered vs. coolant flowrate with an assortment of fans at different speeds, along with the usual pressure drop vs. flowrate information. Pumps would need head pressure vs. flowrate curves plotted (PQ) that include power consumption (hydraulic pumping power could then be calculated and plotted). Now, if you put all of this data together, it would be possible to create an interactive water cooling system 'predictor' of sorts... a user would be able to input a given ambient air temperature, component heat load, die size/configuration, choose blocks, pump(s), and fans and, from that input, it could provide a rough prediction of flowrate through the cooling system, power consumption of the system, and cooling effectiveness (i.e. temperature estimates). What could possibly be more "real world" and useful than that?

Now I realize that I'm not really going as in-depth as I could about this, and that the above described tool would be providing people with ideal results (which wouldn't be that much of a problem, assuming a certain level of user competency)... but this is what really needs to be done.

Justin, the "testing" method that you seem to be suggesting in your post above is, frankly, not one which belongs on any credible tech site as it is far too uncontrolled--especially considering the very small performance differences you're attempting to measure. There is a reason why only two tech sites, that I know of, have ever even attempted to properly evaluate waterblock performance...

Please keep in mind that none of this is intended to be inflammatory--just blunt.

Anyway, to answer the two questions that you posed, CoreTemp would be best for temperature acquisition (within your proposal)... but you're still drawing information from a very small point of measurement. The C2D's onboard thermal transistor can't tell you what's going on across the whole die and it's not something that you can readily calibrate against a thermocouple or RTD. As such, the data it gives you still presents itself as a pretty big variable that's being left unaddressed. As for loading the CPU, TAT will likely give you higher heat output than Orthos, Prime, or a Pi calc program.
 
Yes Petra, that is exactly what I mean ;)

As for your criticisms of the testing methodology, while I do share your point of view, I understand that is not the philosophy of [H] overall. [H] is interested in real world testing, for video cards, CPU's and apparently cooling.

Knowing this is the case, instead of trying to get them to adopt a test bench with pressure gauges, flow meters, thermally controlled chillers keeping the water at a constant 25C etc...I tried to suggest something which would increase the accuracy and credibility of the testing while maintaining their core values.

Using the TTV spec and a proper thermal probe would exponential reduce their error in reading core temps. It would also do something to set them apart from all the other tech sites doing meaningless mashups of cooling reviews. It also isn't necessarily very hard to do, just costs a bit of money and you have to find a machinist. I am sure if they would ask you, you could tell them the measurements (or where to get them) and how much it costs to machine.


Overall if they could improve the temperature reading accuracy I could live with the testing. In fact it might be really good, because they will likely present the information in a much easier to understand manner than the typical scientific waterblock review. This would mean more of the general public that is interested in watercooling would understand what to buy and what to steer clear of. However using the internal temp probe is amateurish at best in my opinion.
 
Erasmus354 said:
As for your criticisms of the testing methodology, while I do share your point of view, I understand that is not the philosophy of [H] overall. [H] is interested in real world testing, for video cards, CPU's and apparently cooling.
Click to expand...
I agree with Alex. And the notion of "real world testing" is all well and good, but we didn't reach the current level of quality of waterblocks, radiators and pumps simply by "real world testing".
 
Erasmus354 said:
Overall if they could improve the temperature reading accuracy I could live with the testing. In fact it might be really good, because they will likely present the information in a much easier to understand manner than the typical scientific waterblock review. This would mean more of the general public that is interested in watercooling would understand what to buy and what to steer clear of.
Click to expand...
See, that's just it, though... it wouldn't make things any more clear-cut than they are now. All that would do is add to the confusion, uncertainty, and FUD that's floating around. Collect enough scientific data and slap an interface onto it and no lay person ever has to see the raw data... all they would get is their questions answered. :rolleyes:

All of those "what would happen if I used XX pump instead" or "what if I add this block to the loop" questions would be come very easily answerable--it would just take a lot of work to get to that point.
 
Petra said:
All of those "what would happen if I used XX pump instead" or "what if I add this block to the loop" questions would be come very easily answerable--it would just take a lot of work to get to that point.
Click to expand...

Except how well it will work on a motherboard and real CPU.........

The Fuzion is a great example of a block that would be so so on a simulated system but in real world kicks major ass.

What has been lost in this world is how well a block actually cools the CPU.
 
You must log in or register to reply here.
Back
Top