How dumb is this?

[faster3200]

Alright, I haven't eaten yet today and I am on meds so I am kinda loopy right now, so I thought I would ask you guys this since most of you are probably clear minded.

How dumb is this?


Basically, going from pump/res>gpu>NB>back through the same gpu

I was going to test this but my second pump died today .

[millertime359]

Your saying that you want to run through the GPU twice?

Why?

[MaByBoI]

wheres the rad?

[faster3200]

Originally Posted by millertime359

Your saying that you want to run through the GPU twice?

Why?

Super short loop. I think it might cause too much turbulence which would be a flow killer, but GPUs don't need much flow. Then I start thinking about how turbulence sounds similar to ambulance and it goes "woo-woo" and makes all the lights it drives by change colors and how I wish I could have that power since then I could play God of the lights and... you get the idea.

Originally Posted by MaByBoI

wheres the rad?

Underneath the couch, look harder. Yeah, I forgot it but there will be one in there, not in parallel.

[MaByBoI]

how will putting it through the gpu again work? wouldnt water just go right through the fittings instead of going through the block? one end being closed is what forces the water through the block... if there isnt a block... the water would just go right through would it not?

[millertime359]

Originally Posted by faster3200

Super short loop. I think it might cause too much turbulence which would be a flow killer, but GPUs don't need much flow. Then I start thinking about how turbulence sounds similar to ambulance and it goes "woo-woo" and makes all the lights it drives by change colors and how I wish I could have that power since then I could play God of the lights and... you get the idea.


Underneath the couch, look harder. Yeah, I forgot it but there will be one in there, not in parallel.

Yea, I was wondering bout that rad too.

It probably would cause alot of turbulence, but hey, give it a try and see what happens.

[Elusid]

Well some water would pass through the GPU wouldn't it? It depends on how restrictive the NB block is because isn't that how parallel GPU blocks work? But seeing as the NB block probably has little restriction there wouldn't be much movement in the GPU block... just go in a series IMO

[Caoder]

well as with all parallel loops it depends on balanced resistance.. so what block are u using for the nb block?

[coolmiester]

Might be worth having something to eat and getting some better medication

[SNiiPE_DoGG]

its a normal parallel loop yes, and I think it will work fine even if there is unbalanced restriction. fullcover GPU blocks work well with low flow and the NB on anything needs only minimal cooling power.

[NaeKuh]

Originally Posted by faster3200

Alright, I haven't eaten yet today and I am on meds so I am kinda loopy right now, so I thought I would ask you guys this since most of you are probably clear minded.

How dumb is this?


Basically, going from pump/res>gpu>NB>back through the same gpu

I was going to test this but my second pump died today .

Im trying real hard to visual this..

So your gonna use your GPU blocks as a passthough? feed your NB and have it come back down as a pass though?

I dont think it will work. :X

Originally Posted by Elusid

Well some water would pass through the GPU wouldn't it? It depends on how restrictive the NB block is because isn't that how parallel GPU blocks work? But seeing as the NB block probably has little restriction there wouldn't be much movement in the GPU block... just go in a series IMO

i dont think it will work, because its a straight pass though. There is no place for the water to turn inside the gpu block.
Its gonna go inside the bottom come out top then goto NB, then from NB go back down, straight though the hole on the gpu block.

[Caoder]

it'll only work if the northbridge is rather restictive.... no?...or balanced restriction :/....

[MpG]

Originally Posted by NaeKuh

Im trying real hard to visual this..

So your gonna use your GPU blocks as a passthough? feed your NB and have it come back down as a pass though?

I dont think it will work. :X



i dont think it will work, because its a straight pass though. There is no place for the water to turn inside the gpu block.
Its gonna go inside the bottom come out top then goto NB, then from NB go back down, straight though the hole on the gpu block.

Maybe, maybe not. It's basically the same concept as running GPU's in parallel. As mentioned, the whole question is whether the chipset block is going to be restrictive enough to divert some of the flow through the GPU block. But I expect that a typical flat-bottom chipset block may not be restrictive enough to do this, and the vast majority of the flow will miss the GPU block entirely. It'll probably still work, because GPU's have a lot of headroom under water, but I can't see it being an ideal situation.

@OP: Really, you're not going to have 'too much turbulence' if you just run stuff in series. You may not see much benefit past a certain flowrate, and whatever pump you've chosen, it may be over-powered and dump a couple extra watts into your loop, but it's not going to hurt anything.

[eth0s]

"Too much turbulence"? No. You will not have too much turbulence. Remember: you WANT turbulence in the GPU and NB blocks, because turbulence aids in the removal of heat from the copper blocks. The opposite of turbulent flow is laminar flow, which means smooth flow. Laminar flow is BAD for cooling because it rushes past the copper block and only a small outer layer of water captures heat. Whereas with turbulent flow in the blocks you will have mixing of water, which will allow a greater amount of water to absorb a greater amount of heat from the block. Turbulence is also very good for radiators too, as turbulence is what exchanges the heat from the water to the air inside the radiator.

However, turbulence is BAD for pumps. You do not want turbulence entering your pump, because that can cause a drop in pressure and/or cause cavitation, which is where the surface pressure of the water drops to the point where the water "boils" in the pump. This is bad for the pump. BUT, you have a rad/res combo, so you don't have to worry about turbulence entering your pump. Furthermore, the shorter and straighter your tubing runs are, the more laminar the flow will be in them. Think of a rifle barrel straightening out the trajectory of a bullet, and you will get the idea of what straight tubing runs will do for turbulent flow.

Okay now that we discussed turbulence, let's turn to your proposed idea next. I think what will happen is that you are basically running the GPU and the NB blocks in parallel. You will be splitting the flow between the GPU and NB blocks. To see what I mean, imagine that you have two "Y" splitters before and after the GPU block, splitting the flow into parallel between the GPU and the NB blocks, however in your system, there will be uneven lengths of tubing, and I think the GPU might actually be starved for pressure and flow, and the NB block will get more than it's share. As you have it diagrammed, most of the flow will bypass the GPU and go into the NB block, and them come back down and go straight to the pump, I think. There will be some water bleeding off into the GPU block, but it will not have much pressure or flow. Anyway that is my guess, based upon what I have pictured in my head. However, that does not mean your setup will not work, as I think it will. It will not be the most efficient use of your pump, but I think it will cool everything. Once the system reaches steady-state equilibrium, you will adequately cool the GPU and the NB block. However, I think a more conventional approach will yield better results. But try it out to see what happens. Check your GPU temps, and see if they are acceptable, then keep it. Have some fun in life.

[iandh]

This would work fine, but you'll to place a flow restrictor in the tubing on the inlet of one of the two blocks, unless they somehow happen to be optimally matched by chance.

If you PM me I'd be happy to make up a couple different size ID restrictor inserts from delrin for free, it will take me about 5min to do on the lathe at lunchbreak.

[Adamantine]

You could just use a 1/2 inlet on the GPU and 3/8 on the outlet along with smaller tubing to the NB in order to restrict flow.

[Caoder]

or just clamp part of the tubing on the nb block

[{GOD}Raptor22]

Brawndo! Its what the plants crave!!!

A cookie to anyone who gets that reference...

[HDCHOPPER]

LOL Raptor !! it's got eletrolites

But ya OP just put a pill jet to & fro the gpu , don't listen to me I'm on meds but your sound better besides I'm a motor man

I agree with iandh

[Noisy]

Ok, let look at it this way, water loop is very much like a close circuit electrical loop.

water flow is current = I; block is like resistor = R; and the pump is potential = V

for a parallel loop, potential(pump) is the same for across each resistor (block), so unless one of the resistor is a wire, meaning no restriction, then all the current (water) will go through that branch.

So at here each branch (gpu and nb block) does have restriction, it will work. It just you have no idea how much current (water flow) will go through that particular block, thereby the effect of enough heat is being carry away.

[{GOD}Raptor22]

Originally Posted by HDCHOPPER

LOL Raptor !! it's got eletrolites

But ya OP just put a pill jet to & fro the gpu , don't listen to me I'm on meds but your sound better besides I'm a motor man

I agree with iandh

Hahaha.

I love the narrator, he says probably the funniest crap in that movie.

After 6 hours of trying to persuade them with logic and reason, Joe gave up and just told them he could talk to plants....

And right near the beginning:

Get your hands off mah junk!

"I'm actually supposed to be getting out of jail right now"
*smack*
"You're in the wrong line dumbass!"

Oh, and how Fuddruckers turns into Butt*uckers....

[meanmoe]

Your going to starve you gpu. Go with you instincts (indicated by thread title).

[eXa]

Done right, this shouldnt be a problem at all. the nb hardly needs much flow so if you use a restrictor in the tube or something right before the nb block most of the water will be forced through the gpu block. i would use 7/16 from pump to gpu and from gpu to rad. And 3/8 from gpu to nb and back. and place a restrictor so its only like a 2-3mm hole the water can pass through for the nb.

The more i think of this the more the more brilliant it gets.... as longs as its done right. I would really jump on Iandhs offer, get some with 2-3-4-5mm hole in it. check nb temp without restrictor, test nb temp with 5mm restrictor and then with 2. i bet you wont even se a temp difference on the nb with the 5mm and can probably easy go with something smaller, maybe even just 2mm.

Edit: man... almost bothers me that i havent thought of this myself
Edit2: Just to make myself even clearer: To make this work properly the key is to balance restriction properly. The more restrictive the nb is the more of the water will pass through the gpu. so make the nb as restrictive as possible without loosing too much performance.
Edit3: Dont listen to all the nay sayers, if you understand how restriction will affect flow through different components in a paralell loop, you will understand why this will work.

[meanmoe]

I'm not a naysayer. swiffy did it with the vga ram blocks a while back - and weren't those a hit. If you want the flow to be balanced just make sure that the tubes to the nb have half the cross-sectional area of the smallest cross-sectional area within the gpu block. Restriction is a misnomer for minimal wetted cross-sectional area. Why not run serial, worried about airflow with the additional tubing?

edit: If Iandh is offering... That would work to reduce the area in the place of using a smaller tube to the nb.

[demonkevy666]

I'm interested in the results.
water is cohesive, and limianir flow isn't really possible because there is still some turbalce caused by the gap into the gpu block.
btw riles have rifled barrels for spin. although I agree with you don't want turbulence in the pump it also would cause air bubbles.