Parallel rads vs. serial, quad core block reccomendations

[Rynth]

So I'm designing a two loops for a pair of boxes, both will have CPU, GPU, 2 rads, and a res, plus one MCP655 pump. I hadn't really thought about putting the rads in parallel with Y fittings, but someone suggested it in a thread about someone else's 2-rad setup. What are the benefits and drawbacks of such a system compared to rads in serial?

Also, I've got an Apogee GTX that I'd planned to use on my main loop, but the CPU is a quad core. Are there substantially better quad core blocks out there that would be worth upgrading to, or is the difference small enough to not really be worth the price?

 

[jokers_greg]

difference isn't really that great, but if you're cooling an intel quad core, you'll want a quad-core nozzle. D-Tek has one, basically all it does is redirect the water coming through the input barb to the two sides of the die (since Core 2 quads are basicallly 2 x 2cores). It's useless on a phenom though.

The difference isn't that big, but I'm not even sure if the apogee GTX offers something like that

 

[RedRaider]

Forget the D-TEK w/quad nozzle, either stay with the Apogee GTX or upgrade to the Apogee GTZ (out in a couple of weeks) or EK Supreme...

 

[shachar2]

From no little understanding (no real experience yet) parallel loops are a problem because you might have different restrictions. say if you have a cpu on one loop and two GPU's on the other parallel loop most water will go through the cpu loop since it's the least restrictive. (assuming all three are simple stupid blocks)

I couldn't find any data on the internet or someone who really tried it but I think the restriction problem can be solved by two flow indicators on each parallel loop (to see if the flow is even to both of them) and a clamp to even out the flow in case they're not even

 

[jokers_greg]

I think you're better off setting up a rad after the cpu and one after the gpu. So i guess thats in series (if i remember my circuitry courses well enough... mechanical engineer student here lol). If they're parallel, the water will take the path of least resistance (just like electricity),

therefore whichever load is least restrictive, thats where most water will flow. This results in un-even cooling-- but this could be manipulated into an advantage (dont think it suits your needs though)

 

[CGR]

To reduce restriction with a two loop system, you either want two completely separate loops, or run the rads in parallel.

As long as they are truely parallel then different restrictions after the rads should not matter as its all getting split evenly when it hits the rads.

Personally, if I was even considering going parallel, I would probably just spend a little more for another pump and make two separate loops.

 

[jokers_greg]

To reduce restriction with a two loop system, you either want two completely separate loops, or run the rads in parallel.

As long as they are truely parallel then different restrictions after the rads should not matter as its all getting split evenly when it hits the rads.

Personally, if I was even considering going parallel, I would probably just spend a little more for another pump and make two separate loops.

how is it being split evenly when it hits the rads? if one of the Y fittings was 1/4" (for arguments sake) as opposed to 1/2", More water from the RAD would clearly favor the 1/2" tubing, whereas in series, the water's got nowhere else to go.

 

[CGR]

how is it being split evenly when it hits the rads? if one of the Y fittings was 1/4" (for arguments sake) as opposed to 1/2", More water from the RAD would clearly favor the 1/2" tubing, whereas in series, the water's got nowhere else to go.

Why would you use different fittings? You never do that..

Running them in parallel means all fittings are identical. After the pump the Y sends water to each rad. So each rad has the same restriction going in and going out along with same flow.

Running rads in series causes more restriction as rads in general are restrictive. By running them in parallel, you remove that second restriction, which would be the second rad in series.

 

[jokers_greg]

Why would you use different fittings? You never do that..

Running them in parallel means all fittings are identical. After the pump the Y sends water to each rad. So each rad has the same restriction going in and going out along with same flow.

Running rads in series causes more restriction as rads in general are restrictive. By running them in parallel, you remove that second restriction, which would be the second rad in series.

it was for arguments sake, the 1/4" fittings would be more restrictive (from higher pressure), therefore there would be less water travelling through it.

 

[CGR]

it was for arguments sake, the 1/4" fittings would be more restrictive (from higher pressure), therefore there would be less water travelling through it.

True, however you should never do that in a loop and definitely not a loop with parallel rads, as it defeats the entire purpose.

 

[jokers_greg]

True, however you should never do that in a loop and definitely not a loop with parallel rads, as it defeats the entire purpose.

CGR, for the last time, I KNOW you shouldn't do that. I only gave that example to represent restricition in a loop.

 

[Evilsizer]

the case imo will determine which way to go. if it were me though, since you list the MCP655 i would go series and make a loop like this, RAD->CPU->RAD->GPU->PUMP. although i could say that depending on the rad your using. that a parallel rad setup would be better if sticking with one pump.

im not sure how stuck you are on that MCP655 but i would suggest getting the MCP350/355. still has good "head pressure", slightly cheaper vs MCP655, much smaller vs MCP655. then i would use 2 of them in a series loop that way if one pump were to fail you would have another as a backup.

What we really need to know to help you is what rads and blocks are you going with. other then cpu WB, plus i would keep the GTX block.

 

[CGR]

CGR, for the last time, I KNOW you shouldn't do that. I only gave that example to represent restricition in a loop.

I just dont understand why you would use that example

 

[saxile]

splitting them wouldn't work that great unless you are running the exact same thing on each side of the Y. One of those blocks has more restriction than the other so the more restrictive of the two will suffer from that. Rad's in general are not that restrictive, in fact the quad split nozzle is more restrictive than a PA 120.3.

 

[CGR]

splitting them wouldn't work that great unless you are running the exact same thing on each side of the Y. One of those blocks has more restriction than the other so the more restrictive of the two will suffer from that. Rad's in general are not that restrictive, in fact the quad split nozzle is more restrictive than a PA 120.3.

Any restriction before and after the rads doesnt matter really. The point to running rads in parallel is to reduce the restriction of running them in series thats all..

 

[Albuquerque]

I agree with CGR on this: running the radiators in parallel would be a good idea; running most of a whole loop in parallel isn't.

This loop:

              /-> Radiator -\
Pump -> CPU -<               >-> GPU block -> Back to pump
              \-> Radiator -/

... is far less restrictive than this loop:

Pump -> CPU -> Radiator -> GPU block -> Radiator -> Back to Pump

And while you might think you get better cooling with the latter, the reality is that water will saturate to a single temperate -- the temperate gradient (ie, localized cooling effect) of having radiators placed like this is essentially zero as the water will heat-soak after only a few minutes.

As for "irregular resistance" causing the water flow to act wierd in the parallel rad example above -- why would it? You buy two of the same radiators and use the same plumbing to each. We could sit here and make up stories about how one rad will have a dead mouse in it, or some d-bag soldered on the wrong fittings and you somehow didn't realize it -- but that's NOT what will happen in 99.98% of the attempts.

And if you still think it's worth mentioning the 0.02% 'what if' factor, then watercooling is FAR too unsafe for anyone that picky to realistically use. They'd be too concerned about the 0.02% chance that their quad-hoseclamped hoses would somehow spring a leak due to the tension of the hoseclamp eating the wall of the hose.

 

[saxile]

correct on the rad part, but from if he goes from rad to something else before the next Y it won't work. But in truth that Y is very restrictive.

 

[Rynth]

It looks like I wasn't clear enough in my initial post.

I'm going to have 2 loops, each comprised of CPU block, GPU block, res, pump, and 2 radiators. Due to space restrictions, the pump, res, and 2 rads will all be near one another, and the CPU/GPU blocks will be connected. Basically, I'm cooling 2 rack mounted boxes, and there will be a combined 'cooling bay' at the bottom of the rack that the rads, res's, and pumps will be in.

What I'm wondering is whether it'd be better to run the rads in paralell or serial, but in both cases there wouldn't be anything "between" them.

EDIT:

https://flic.kr/p/4AkgPL

So for a better idea of what I'm dealing with, space wise, here's my setup. The two big boxes in the left rack are getting cooled, the bay at the bottom will house 4 rads, 2 pumps, 2 res's and a dedicated power supply for the water cooling portion.

My plan for the loops was: Pump -> CPU -> GPU -> Rad -> Rad -> Res -> Return to pump.

 

[HiProfile]

Rynth if the loops are going to be the same, you should have just refered to them as one, and apply advice to both, rather than ask about both. With today's 2-sentance posts people make, its best to make it as easy to understand as possible.


Anyways I'd probably suggest serial, depending on what you have. High restriction blocks, small lines, and BIP-type radiators will favor parallel though. You want to get decent turbulance in the radiators, but not if it causes a good deal of restriction in an already restrictive setup.

 

[jokers_greg]

the case imo will determine which way to go. if it were me though, since you list the MCP655 i would go series and make a loop like this, RAD->CPU->RAD->GPU->PUMP. although i could say that depending on the rad your using. that a parallel rad setup would be better if sticking with one pump.

im not sure how stuck you are on that MCP655 but i would suggest getting the MCP350/355. still has good "head pressure", slightly cheaper vs MCP655, much smaller vs MCP655. then i would use 2 of them in a series loop that way if one pump were to fail you would have another as a backup.

What we really need to know to help you is what rads and blocks are you going with. other then cpu WB, plus i would keep the GTX block.

that's the exact setup I'm going to be going with . I only got one 355 though, with alphacool top I think.