3/8" or 1/4" cpu waterblock?

[p0intblank]

This is a post for a friend since he can't register:

I have a question about how to route my tubing for a water cooling setup I'll be doing. I'll be getting the Koolance Exos-2 and I'll be cooling my GPU, mobo, and CPU. Now, the cpu waterblock can be bought with 3/8" fittings or 1/4", and I'm wondering which setup would provide better cooling. With the 3/8" fitting I would have it go:

rad->cpu block->opening splitter->one to gpu, other to mobo->closing splitter->back to rad

Then with the cpu block that has 1/4" fittings I would have it go:

rad->opening splitter->one to cpu then to mobo, and the other just to the gpu->closing splitter->back to rad

I'm wondering which size fitting I should go with since the size of the cpu block fitting will change the setup, or if theres another better setup I didn't mention. Thanks.


If you want to talk to him directly, his AIM screen name is DevilTrigger529

 

[Erasmus354]

You dont have to get Koolance waterblocks...

Personally I would pair with the Exos2 a Swiftech 6000 (cheap and very good performer), DD Maze4 with 3/8" , and then an appropriate 3/8" nb cooler.

 

[p0intblank]

Ok, this is who wanted to know about the tubing, so I can reply back myself, my friend is letting me use the account instead of having to have him post it for me.

So anyway, now that you stated that, I'm actually going to go with 3/8" throughout the system, since im getting a pretty big case size isnt and issue. First thing, price isnt a factor at all, so what blocks do you recommend for the cpu, gpu, and mobo that would perform the best with the exos-2?

Asus A8N-E

XFX 7800GT

And the CPU is going to be a AMD X2 4200+

 

[Erasmus354]

Swiftech 6000 or the Swiftech STORM for the CPU block. (Storm is better but more expensive)

DD Maze4 or Swiftech MCW50 for the GPU

DD Maze4 Chipset block

 

[thewhiteguy]

You might want a pump somewhere in the loop

 

[arisythila]

A Pump, can you just stick another pump in the loop?

Sorry for hijacking.

 

[Stugots]

I'd go with the Koolance 3/8" block.

 

[Erasmus354]

I'd go with the Koolance 3/8" block.

Koolance waterblocks are a complete and utter ripoff. The Swiftech 6000 performs as well as or better than the Koolance blocks and costs 20$ less than the "old" model. Dont get me started on their new 90$ waterblocks, which dont seem to have changed anything other than the color of the top. For that money you are better off getting the STORM which is the proven king of waterblocks.

As for the pump, there are two already in the Exos 2

 

[p0intblank]

I was looking at the Storm waterblock but I wasnt sure if the exos-2 system would have a good enough flowrate for that to perform properly. But Im guessing that isnt true and to go with the Storm?

 

[ashmedai]

I'd seen the Koolance CHC-A05 given as one of the very few non-custom northbridge blocks that would fit on the DFI mobos...if you know an alternative, post it.

 

[Erasmus354]

I'd seen the Koolance CHC-A05 given as one of the very few non-custom northbridge blocks that would fit on the DFI mobos...if you know an alternative, post it.

Except for the fact that chipset block is a P4 chipset block and not A64

If you custom build your own bracket then yes it might work.

That being said what does that have to do with the discussion? The OP has said he is cooling the ASUS board, which has ample room for most any chipset cooler....

 

[ashmedai]

Except for the fact that chipset block is a P4 chipset block and not A64

I'm told it fits. I haven't exactly tried myself. I was attempting to illustrate that sometimes Koolance is a good choice and that other factors in the system might need to be taken into account.

 

[p0intblank]

So if I use a Storm cpu waterblock it will still work fine with the pump rate of the Exos-2? And then for the gpu memory just get ramsinks for each? I liked the idea of having the gpu block cool the core and memory, but just ramsinks is fine on the gpu memory?

 

[ashmedai]

Ramsinks are okay, but I like the idea too...

 

[jhego]

Sorry, instead of starting another thread i'll just post here. I'm thinking of getting the Exos 2 myself. This is what i'm planning to purchase.
First off, i have a:
Asus A8V Deluxe
AMD FX-53
6800 Ultra

Koolance Exos-2 $349.99
OR
Koolance INX-720SL Cooling System $274.99

Swiftech MCW6000-64™ for AMD® K8, Athlon™ 64 & Opteron™ processor $42.50

Danger Den NV-68 for Ultimate Nvidia Cooling (cools the rams as well) $124.95
OR
Acetal Version MAZE4GPU $44.95

No idea what to get for the NB chipset block unless the Danger Den ASUS A8N MAZE4 Chipset Block will also fit the Asus A8V?

Is this a good setup? Can you recommend something better? I'm looking for superior cooling and best price. Thanks.

 

[Top Nurse]

Except for the fact that chipset block is a P4 chipset block and not A64

If you custom build your own bracket then yes it might work.

Just so you know, as I have that chipset cooler in my junkbox, it would have no problem fitting just about any NB configuration in existence. As I have said many times before, the Koolance method of attaching waterblocks is some of the best I have seen.

 

[Top Nurse]

Sorry, instead of starting another thread i'll just post here. I'm thinking of getting the Exos 2 myself. This is what i'm planning to purchase.
First off, i have a:
Asus A8V Deluxe
AMD FX-53
6800 Ultra

Koolance Exos-2 $349.99
OR
Koolance INX-720SL Cooling System $274.99

Swiftech MCW6000-64™ for AMD® K8, Athlon™ 64 & Opteron™ processor $42.50

Danger Den NV-68 for Ultimate Nvidia Cooling (cools the rams as well) $124.95
OR
Acetal Version MAZE4GPU $44.95

No idea what to get for the NB chipset block unless the Danger Den ASUS A8N MAZE4 Chipset Block will also fit the Asus A8V?

Is this a good setup? Can you recommend something better? I'm looking for superior cooling and best price. Thanks.

For that kind of money you can have an Aqua Computer system. Do yourself a favor and get an Aqua Computer AquagraFX 6800 rev 2 cooler as the DD NV-68 has some well known problems with mismatching between the ram and the GPU that the AC unit does not have. The AC unit also has native 3/8" ID capability if you wanted to match it up to the Koolance unit. BTW, the EXOS II is the better unit.

 

[jhego]

For that kind of money you can have an Aqua Computer system. Do yourself a favor and get an Aqua Computer AquagraFX 6800 rev 2 cooler as the DD NV-68 has some well known problems with mismatching between the ram and the GPU that the AC unit does not have. The AC unit also has native 3/8" ID capability if you wanted to match it up to the Koolance unit. BTW, the EXOS II is the better unit.

cool, thanks for the info, even though it's only 20$ difference it's still a saving. How is the Exos II better than the internal one btw? I thought the only difference is the internal external thing.

 

[p0intblank]

For that kind of money you can have an Aqua Computer system. Do yourself a favor and get an Aqua Computer AquagraFX 6800 rev 2 cooler as the DD NV-68 has some well known problems with mismatching between the ram and the GPU that the AC unit does not have. The AC unit also has native 3/8" ID capability if you wanted to match it up to the Koolance unit. BTW, the EXOS II is the better unit.

Would that AC gpu cooler work with the 7800 GT?

 

[Erasmus354]

Would that AC gpu cooler work with the 7800 GT?

i dont believe ANY combined gpu/ram cooler works.

 

[p0intblank]

i dont believe ANY combined gpu/ram cooler works.

So I'm going to have to go with a gpu core waterblock and get ramsinks for the memory?

And another thing, I was reading about the Storm performance and it says on Frozencpu.com that it should be in a system with .75 - 1 gpm flow rate. The Exos-2 only has about .59 gpm flow rate. Would that matter or is getting the Storm still my best bet for the cpu?

 

[ashmedai]

I know Danger Den is supposed to be releasing a 7800 block early next week, although I obviously can't say anything about what it looks like or how well it will work.

 

[p0intblank]

Oh ok then, I'll wait out to see what comes out for the 7800 line. But what about what I stated in the previous post about the Storm cpu block? Would it still perform well with the Exos-2?

 

[Top Nurse]

Oh ok then, I'll wait out to see what comes out for the 7800 line. But what about what I stated in the previous post about the Storm cpu block? Would it still perform well with the Exos-2?

At that flow rate there are other better cooling blocks around. Alphacool NexXxos XP?, Aqua Computer Cuplex XT, and some other German based waterblock companies.

 

[Top Nurse]

i dont believe ANY combined gpu/ram cooler works.

Actually there are a couple of Aqua Computer blocks that do fit with some minor modifications. The AquagraFX 6800 rev 2 and the AquagraFX 6800 LE. However, I believe Aqua Computer will be delivering their AquagraFX 7800 block in the next week or two. The 6800 and 7800 Aqua Computer blocks have a floating GPU heatsink that is totally unique to them. It eliminates any problems with mismatches between the height of the GPU to the ram.

 

[jhego]

So I'm going to have to go with a gpu core waterblock and get ramsinks for the memory?

And another thing, I was reading about the Storm performance and it says on Frozencpu.com that it should be in a system with .75 - 1 gpm flow rate. The Exos-2 only has about .59 gpm flow rate. Would that matter or is getting the Storm still my best bet for the cpu?

Can someone confirm this flow rate?

 

[Erasmus354]

At that flow rate there are other better cooling blocks around. Alphacool NexXxos XP?, Aqua Computer Cuplex XT, and some other German based waterblock companies.

And now how exactly do you reason that? The Storm G4, which is worse than the Swiftech Storm, outperforms the NexXxos XP with an Eheim 1046, 1048, 1250, mcp600, and mcp650. Basically the Storm outperforms the NexXxos XP in EVERY situation. Souce

If you look at some of the european waterblock reviews they show that the NexXxos XP is near the top, if not the top european waterblock. So therefore how do you reason that the NexXxos XP, Cuplex XT, or any other waterblock for that matter would perform better?

I should remind you that a C/W vs Flow Curve is only half the story. The NexXxos XP is a VERY restrictive block. Meaning in any given system you get a much lower flow rate than you would with the Storm, which is why the Storm outperforms the NexXxos XP once you compare apples to apples.


Also, if you have to modify something to make it fit, then it doesn't fit

 

[J-Pepper]

And now how exactly do you reason that? The Storm G4, which is worse than the Swiftech Storm, outperforms the NexXxos XP with an Eheim 1046, 1048, 1250, mcp600, and mcp650. Basically the Storm outperforms the NexXxos XP in EVERY situation. Souce

If you look at some of the european waterblock reviews they show that the NexXxos XP is near the top, if not the top european waterblock. So therefore how do you reason that the NexXxos XP, Cuplex XT, or any other waterblock for that matter would perform better?

I should remind you that a C/W vs Flow Curve is only half the story. The NexXxos XP is a VERY restrictive block. Meaning in any given system you get a much lower flow rate than you would with the Storm, which is why the Storm outperforms the NexXxos XP once you compare apples to apples.


Also, if you have to modify something to make it fit, then it doesn't fit

Erm.. actually not quite right.

At low flo (1046/8) the XP out performs the G4... i think cathar even said so himself iirc in these very forums as the G4 is designed to be able to be used in a variety of setups, but at a certain low flow rate it ceases to be 'king' so to speak.

If you go to the interactive graph on procooling and plot only the XP and the Storm G4... you'll find that the XP has a lower delta at lower flow rates but as the flow rates increase, the G4 wins.

The XP is a good block... the shame thing about it is that it is branded as a 'lo-flo' block when infact is shines very well at high flows... I would personally consider this a viable alternative to the G4 (for use europeans anyway) as a much cheaper option at only 29.99 Euros (£20 or $37).

And until I see some standardised procooling tests between the Swifty Storm G4 and Cathars G4.. i'm not going to comment on how much better one is over the other as any difference is speculation.

 

[p0intblank]

Can someone confirm this flow rate?

Unless my math is wrong, that should be correct. The review on hardocp for the Exos-2 says the flow rate is rated at 2 liters per minute. I used convert-me.com to see how many gallons were in a liter to see the relation with the gpm that was on Frozencpu.com for the Storm. And it turns out that there for 2 liters its about .59 gallons. Correct me if Im wrong but I think I'm right.

 

[SpoogeMonkey]

For a low flow setup like the Exos 2, you might want to look at the Swiftech mcw6000 block, as they perform quite well in both low/high flow loops. And since it's American made, you don't have to pay a premium for shipping.

 

[p0intblank]

For a low flow setup like the Exos 2, you might want to look at the Swiftech mcw6000 block, as they perform quite well in both low/high flow loops. And since it's American made, you don't have to pay a premium for shipping.

Alright cool, I guess I'll go with the mcw6000 block then. Thanks guys.

 

[Erasmus354]

Erm.. actually not quite right.

At low flo (1046/8) the XP out performs the G4... i think cathar even said so himself iirc in these very forums as the G4 is designed to be able to be used in a variety of setups, but at a certain low flow rate it ceases to be 'king' so to speak.

If you go to the interactive graph on procooling and plot only the XP and the Storm G4... you'll find that the XP has a lower delta at lower flow rates but as the flow rates increase, the G4 wins.

The XP is a good block... the shame thing about it is that it is branded as a 'lo-flo' block when infact is shines very well at high flows... I would personally consider this a viable alternative to the G4 (for use europeans anyway) as a much cheaper option at only 29.99 Euros (£20 or $37).

And until I see some standardised procooling tests between the Swifty Storm G4 and Cathars G4.. i'm not going to comment on how much better one is over the other as any difference is speculation.

*sigh* apparently you failed to read what I said. The C/W vs flow is not the whole story. If you put the NexXxos XP in the same system as a STORM block, the Nexxxos loop would get a much lower flow rate. This is because the block is much more restrictive. The loss of flow makes the nexxxos xp perform worse than the storm in identical loops with the aforementioned pumps. Just so you dont get confused this time I will snatch the graph and host it and put it here.....lemme go do that.

EDIT
"java.lang.NullPointerException" Ok well hosting the picture wont work so I will hyperlink directly to the picture...

Real World Comparison

Also read this part of the review please :

However, one can see that contrary to conventional wisdom the Alphacool is NOT OPTIMIZED for loops with low flow rates (such as the Eheim 1046 can provide). In such a loop the Swiftech MCW6000-A would actually be a better choice. As flow rates increase with larger pumps, however, the NexXxoS XP becomes extremely competitive. In essence what this graph is showing is that the decreased flow rates caused by the Alphacool block being extremely restrictive (and has lower flow rates) more or less negates the fact it performs much better as a function of flow rate than other blocks I have tested. Hopefully this provides a more clear picture of how cooling performance, flow rate, and flow resistance are related in a real system.

Now please, next time actually READ what I post before you go about trying to prove me wrong.



NOW : @ the OP : I think you made a nice choice with the Swiftech mcw6000. If you actually look at the graph I linked to in this post you can see that at low flow rates the mcw6000 actually performs very close to the Little River Storm G4. You should be very pleased with that very efficient and cost effective waterblock.

 

[Top Nurse]

And now how exactly do you reason that? The Storm G4, which is worse than the Swiftech Storm, outperforms the NexXxos XP with an Eheim 1046, 1048, 1250, mcp600, and mcp650. Basically the Storm outperforms the NexXxos XP in EVERY situation. Souce

If you look at some of the european waterblock reviews they show that the NexXxos XP is near the top, if not the top european waterblock. So therefore how do you reason that the NexXxos XP, Cuplex XT, or any other waterblock for that matter would perform better?

I should remind you that a C/W vs Flow Curve is only half the story. The NexXxos XP is a VERY restrictive block. Meaning in any given system you get a much lower flow rate than you would with the Storm, which is why the Storm outperforms the NexXxos XP once you compare apples to apples.


Also, if you have to modify something to make it fit, then it doesn't fit

My source is Cathar himself. He has stated that the G4/5 doesn't perform well at the lower flow rates. Sorry, but I can't remember what the flow was he said, but I think the Exos II qualifies in this respect. Also since there is no impartial testing comparing a G4 made by Little River and a Swiftech Storm we can assume from past occurences of Cathar designed blocks manufactured by third parties that there may be significant differences.

If you modify anything to custom fit it then it fits. Please remember we are discussing watercooling here and not plopping in a video card.

 

[ikellensbro]

I think this the post by Cathar that you're talking about: http://www.hardforum.com/showpost.php?p=1027841269&postcount=13

He has stated before, and it has been tested, (as shown in the graph linked in the post above) that the 6000 performs better at very low flow rates such as 6 or 8mm loops with several/many 90* mitre elbows and/or restrictive blocks. This can be easily seen and predicted if you look at ProCooling's real world performance graph, where you can easily see that with a weaker pump than a 1046 or a more restrictive loop, the 6000 would outperform the Storm. Still, the Storm performs very well even at 0.5gpm, just a fraction of a degree below the 6000 (Interactive Waterblock Test Results).

As for the Swiftech Storm peforming different compared to the LR G4, ProCooling's tests found Dtek's WhiteWater performed just slightly worse than Cathar's WW (0.5* difference @ 2.45gpm) as seen here. I haven't seen any distinct differences between Cathar's G4 and the Storm, other than the Storm's plates look slightly thicker to me. I doubt that the performance difference between the two is even half that of the difference between the Dtek WW and Cathar's WW. Please TN, back up what you're claiming with a link or two. And what (I believe) Erasmus is saying is that if it you have to modify something to fit, that means it doesn't fit out of the box (anything can fit if you spend enough time and use enough adhesive ).

 

[Top Nurse]

Please TN, back up what you're claiming with a link or two.

Also since there is no impartial testing comparing a G4 made by Little River and a Swiftech Storm we can assume from past occurences of Cathar designed blocks manufactured by third parties that there may be significant differences.

No links needed as it is common knowledge that the Dtek WW was not the same as the Little River WW.

 

[J-Pepper]

*sigh* apparently you failed to read what I said. The C/W vs flow is not the whole story. If you put the NexXxos XP in the same system as a STORM block, the Nexxxos loop would get a much lower flow rate. This is because the block is much more restrictive. The loss of flow makes the nexxxos xp perform worse than the storm in identical loops with the aforementioned pumps. Just so you dont get confused this time I will snatch the graph and host it and put it here.....lemme go do that.

EDIT
"java.lang.NullPointerException" Ok well hosting the picture wont work so I will hyperlink directly to the picture...

Real World Comparison

Also read this part of the review please :



Now please, next time actually READ what I post before you go about trying to prove me wrong.



NOW : @ the OP : I think you made a nice choice with the Swiftech mcw6000. If you actually look at the graph I linked to in this post you can see that at low flow rates the mcw6000 actually performs very close to the Little River Storm G4. You should be very pleased with that very efficient and cost effective waterblock.

ok.. misunderstandings are misunderstandings but what i'm saying is that you can't label the G4 being better than the XP in EVERY respect (as you put it previously) when the link you refer to is THEORECTICAL.

as the link you gave refers to a theorectical (i.e. a hypothesis, not tested, an idea) then yes you are correct... but that is speculation on the part of the test designers at the minute as it is has not been experimentally tested... but the fact remains is that the XP block can compete competively and beat the Storm G4 at low flow rates.

I'm not a XP fan boy, and you may not be a G4 either (but not from i've read so far) and I for your information use a Storm in my rig at the moment... but from pure experimental evidence... the XP can compete with the G4 at low flow rates...

The XP according to experimental data performs better than the Storm at lower flow rates.. so in theory, the lower flow rates exhibited in the same loop wouldn't necessarily equal less of a performance as the XP actually performs better than the G4 at those low flows rates... that's kinda what i'm getting it... unless some standardised experimentation is done, it's up in the air.
The G4 may most probably be better in all respects, but until experimentation (yes, i'm going to use that word a lot!) is done.. I will not blindly state Theorectical hypothesises as true statements, which you are doing.


that's all i'm saying... no concrete evidence (i.e. standardised experimentation)... no chance i'm going to trust them. That's the way it is... theoretical research can be great for hypothesis making and predictions... but in the real world, a scientist would never trust these theories and would always do experimentation to back them up as it is just speculation at the end of the day.

 

[Erasmus354]

*snip* Taken to PM, lets try to get the thread back on topic. *snip*

 

[ikellensbro]

The tests done with ProCooling's real world loop has been discussed plenty, but as of now seems to be a very good representation a block's actual performance in a loop. Granted, not as many people use Thermochill radiators or Papst fans in terms of market compared to all other commerical offereings combined, but this test does isolate the block's performance based on a given pump, as the loop's other variables are controlled.

Does the XP perform better than the G4 from 0.5-1.25gpm? Yes, but you are gonna need one heck of a pump to get you there. To quote Pro's review:

[font=Arial,Helvetica,Geneva,Sans-serif,sans-serif][size=-1]The maximum flow rate I could achieve was 1.42 GPM with the Alphacool waterblock, which makes it the most restrictive waterblock I have tested.[/size][/font]

Notice that in Pro's Interactive Waterblock Test Results, the XP's results stop after 1.42gpm, while the G4 is able to achieve 1.7gpm. At 1.38gpm, the G4 is outperforming the XP, and 1.38gpm is an actual achieveable flow rate that is realistically possible to achieve with the Storm.

Going back to their real world testing, you can see that, using the Delta T vs. flow rate graph in the previous link, that the G4 would be able to achieve higher flow rates with the same pump (yes, the water inlet temperature isn't being controlled in their real world loop, but if the XP performed from 0.4*C to 0.1*C better from 0.5-1gpm, then the Storm is getting much higher flow rates in the real world loop, as it's outperforming the XP by about 0.5*C, except with very high flow coming from the MCP600, and even then the G4 is cooling about 0.4*C better).

... but the fact remains is that the XP block can compete competively and beat the Storm G4 at low flow rates.

Yes, that's true, but the G4 is able to achieve significantly higher flow rates than the XP with the same pump, and that's in a very unrestrictive loop. In a loop with extremely low hydraulic power, it looks like the G4 and XP are about even, but consider that this graph is theoretical, that a 1046 can produce about 0.5W of hydraulic power, and this graph is talking about pure hrdraulic power vs. performance, not actual flow rate. In the real world loop graph, you can see that the G4 has it's biggest performance margin over the XP with a 1046, and that the gap is less with a 1048, so in a restrictive loop with a 1046, it loks like the G4 might have an even larger performance margin.

As for how Dtek's WW performed compared to Cathar's, I guess this is a major performance loss after being mass produced:

From 0.75 (about what a 1046 could get in a WW loop) to 2gpm, the difference is from >0.1-0.3*C.

 

[Erasmus354]

Why are people even questioning the difference between the blocks. Cathar and Swiftech have both said that the Swiftech STORM is an IMPROVED version of the Little River block. Go read the initial thread about it....

So in using the LR Storm G4 for comparisons we are taking an inferior product and comparing it. By treating it as the same performance as the Swiftech Storm in our comparisons we are helping competing products look better.

 

[jhego]

Why are people even questioning the difference between the blocks. Cathar and Swiftech have both said that the Swiftech STORM is an IMPROVED version of the Little River block. Go read the initial thread about it....

So in using the LR Storm G4 for comparisons we are taking an inferior product and comparing it. By treating it as the same performance as the Swiftech Storm in our comparisons we are helping competing products look better.

Swiftech it is