1 or 2 loops?

M

Monkeypillow

Gawd
Joined
Aug 30, 2006
Messages
880
I have a Gigabyte GA-EX58-UD5, an overclocked core i7, and two GTX 260s. For a pump I've got an MCP655 with an EK Waterblocks acetal top.

If I were to cool both sets of mosfets, the northbridge and southbridge with a full block, my processor, and both graphics cards, would I need two loops?
 
Last edited:
2 loops doesnt give you any advantage over one. it just looks cooler sometimes, especially if you use 2 different colors. if you add all the water blocks and 4x120mm rad surface area, you should get a very effective cooling solution. depending on how restrictive your blocks are, you may need a second pump. i doubt it though.
 
It's an i7 so you don't have to worry as much about cooling the northbridge/southbridge as a 775 unless you're doing it purely for aesthetics. And even if you're doing it for looks, having waterpipes all over doesn't always look nice IMHO and and the GA-EX58-UD5 has fairly nice looking copper heatsinks.

In fact, that's another reason: the GA-EX58-UD5 which has rather large copper heatsinks and all you need is just a bit of airflow and you'll be fine.

I'd consider dropping the watercooling for the northbridge/southbridge/mosfets as not only is it not needed, you'll actually get better cooling performance without it. And I prefer single loops.
 
meh. if you already have the block you might as well use it unless you think you can return it or make a decent deal back by selling it. i doubt youd actually get cooler temps with stock than the full block :p but yeah its not entirely necessary. make sure you get a good cpu block though as i7s get super hot. the heatkiller cpu 3.0 is getting rave reviews, and the gtz is still pretty good. with a decent mobo you should be able to hit 4GHz really easy on water.
 
I'm still rather new to watercooling, so take my advice with a grain of salt. But given that you've got so many waterblocks, dual loops may be advantegous. My reasoning is that the water in the pipeline is going to get progressively hotter as it travels from one block to the next. Furthermore, your flow rate is going to be impacted by the combined resistance of all of those components. Long story short, your flowrate is going to be slow and the components towards the end of the loop are going to be hotter.

I've read from some that the temperature differential between the component first in the series and the last sometimes only differs by a few degrees. But given that you've got the NB, SB, and Vregs in there, I'm going to guess that it's going to be a large differential than that. So splitting it up may let you dissipate the heat faster.

Get some more opinions though. I haven't had any first hand experience, but that would be my two cents.
 
Where does he say he has the blocks already?

I'll re-state what I said a bit more clearly: it is a "bad" idea to watercool both sets of mosfets, the northbridge and southbridge etc.

Go for it if you're an advanced watercooler and doing it for looks or just creating a challenge for yourself. But otherwise, even if I had the blocks, I'd return them.
 
ekuest said:
2 loops doesnt give you any advantage over one. it just looks cooler sometimes, especially if you use 2 different colors. if you add all the water blocks and 4x120mm rad surface area, you should get a very effective cooling solution. depending on how restrictive your blocks are, you may need a second pump. i doubt it though.
Click to expand...

This isn't exactly true. For example, if I were to split my loop up and run my E3110 on the MCR220, it's temperatures would likely drop, as would my GPU and NB on the other loop. This is because the GPUs put out more heat than anything else and having them share a loop with a cooler GPU and NB raises their respective temperatures. Separating them isn't necessarily cost-efficient, but it would give better cooling performance.
 
i said if he already has the water block. otherwise i said its not worth it.

regarding the 2 loops, youd need a second res, a second pump, more tubing and barbs, and a second rad if you dont already have 2. (you should go with 120.4 if you can, 120.3 would barely be enough for an i7 and 2x260s. this can be in one radiator or 2 smaller ones.) all this would add up to more money, more room taken up in your case, and more of a headache to work with.

as to what ckoz said, take a few grains of salt. the temps level out to be basically the same throughout the loop. the board block would add barely any heat, possibly even suck heat out as echn said because it might be cooler than the water in the loop. :( as for flow rate, you CAN NOT have a different flow rate in one part of the loop than another. simple conservation of mass. if you have a lot of restriction then the entire loop will slow down. you can always add a second pump to the loop instead of adding a whole second loop if restriction becomes a problem.
 
It's true that it'd cost more and be more complicated, but you said that it would offer no advantage, which isn't true. I don't think it's cost-efficient, but it certainly gives better temperatures, therefore, an advantage.
 
how does it give better temperatures? from a fluid dynamics and thermodynamics standpoint i disagree with this statement. the flow will be better yes, but then thats only cuz you have a second pump. the exact same components in a second loop shouldnt be any better than all in one loop. if im wrong please enlighten me. ive just always figured theres no advantage from what i know of fluid- and thermodynamics.
 
It does give better temperatures based on what I've read from other people's builds. By separating components that don't put out as much heat from those that do, the cooler ones don't have to share the heat of the hotter ones in the same loop, so they can run cooler. I don't understand how that statement violates fluid or thermodynamics or how that even factors into this.
 
If you guys want to discuss theory, overall, you will actually have (marginally) more heat in a dual loop system verses a single loop as you now have two pumps. (The heat doesn't somehow disappear)

And as for splitting the cooler components from the hotter ones, well LoL. Sure the cooler components will certainly be cooler. And of course, the hotter components will be hotter. Now guess which components are the hottest? This is a really bad idea as who cares if you get a cooler NB/SB/mosfet at the cost of higher CPU and GPU temps?

No. What people often do when they go dual loop is they split up the GPU's and CPU's (or isolate the CPU) and put more cooling on the CPU. So the GPU's in this example will run hotter (yes hotter) than a single loop, but the CPU will be cooler. And this "trade-off" allows you to get a higher CPU overclock, but at a cost. This is the advantage of a well designed dual loop.
 
ckozlowski said:
I'm still rather new to watercooling, so take my advice with a grain of salt. But given that you've got so many waterblocks, dual loops may be advantegous. My reasoning is that the water in the pipeline is going to get progressively hotter as it travels from one block to the next. Furthermore, your flow rate is going to be impacted by the combined resistance of all of those components. Long story short, your flowrate is going to be slow
Click to expand...
True. This can impact temps and this is indeed a valid argument in favour of dual loops.

In fact if I were to do a build like the OP and insisted on cooling everything, I'd probably do dual loop for this reason (as well as isolating my CPU to keep it cooler).

However, this is also why sometimes people don't have so many waterblocks so they avoid this problem altogether and can therefore stick to a single loop.

and the components towards the end of the loop are going to be hotter.
I've read from some that the temperature differential between the component first in the series and the last sometimes only differs by a few degrees. But given that you've got the NB, SB, and Vregs in there, I'm going to guess that it's going to be a large differential than that. So splitting it up may let you dissipate the heat faster.
Click to expand...
No. From a practical perspective, the temps are going to be fairly similar. That's why no one really cares that much about loop order any more and recommend the shortest most efficient loop in almost all cases.
 
echn111 said:
It's an i7 so you don't have to worry as much about cooling the northbridge/southbridge as a 775 unless you're doing it purely for aesthetics. And even if you're doing it for looks, having waterpipes all over doesn't always look nice IMHO and and the GA-EX58-UD5 has fairly nice looking copper heatsinks.

In fact, that's another reason: the GA-EX58-UD5 which has rather large copper heatsinks and all you need is just a bit of airflow and you'll be fine.

I'd consider dropping the watercooling for the northbridge/southbridge/mosfets as not only is it not needed, you'll actually get better cooling performance without it. And I prefer single loops.
Click to expand...

Thanks for the suggestion. I was originally considering the excessive cooling because my previous EX58-UD5 ran extremely hot, but I had to RMA it because the NICs were letting out a terrible screech. The new one seems fine, so it shouldn't be necessary. It would allow me to use three graphics cards (2 GTX 260s plus a 9800GTX for PhysX) and a soundcard.

If my current motherboard develops a problem, I'll likely move to something else.
 
ekuest said:
meh. if you already have the block you might as well use it unless you think you can return it or make a decent deal back by selling it. i doubt youd actually get cooler temps with stock than the full block :p but yeah its not entirely necessary. make sure you get a good cpu block though as i7s get super hot. the heatkiller cpu 3.0 is getting rave reviews, and the gtz is still pretty good. with a decent mobo you should be able to hit 4GHz really easy on water.
Click to expand...

I was able to get to 3.7GHz on the board I returned, but I couldn't get it to 4GHz. I doubt it's the chip, as I didn't read up on anything before I started overclocking the i7, just raised the voltage a bit, upped the base clock, and up it went. I haven't had time to play with the new board yet, as I'm still waiting for RMAs, new parts, etc.
 
You must log in or register to reply here.
Back
Top