Serial loop Vs. Parallel loop

What happens to your flowrate if you remove that inline filter? You shouldn't ever need to run something like that in a loop, plus it looks like it's very restrictive.

The links on your hardware are broken. You need to use the[*url] and [/*url] tags for links (remove the *).

Can you show a diagram of the loop order?

[Edit] when I zoom in on your photo of the loop, it shows your GPUs in serial.

Looks like you're seeing firsthand how restrictive full cover GPU blocks in serial vs. parallel can be.

That is definitely one of the more interesting parallel loops I've seen; not very common.

Interesting that you run the CPU in parallel with the GPUs rather than the CPU in serial, then parallel the GPUs and serial the radiators. Radiators are usually low pressure drop and like higher flow. Also, the MCR series is pretty low restriction compared to other rads, so running them in series isn't very detrimental. I ran 2x in series for several years; still have them.

Also wondering what difference would be if you left GPUs and CPU as-is and ran the radiators in serial, like first loop. Thinking the difference would be negligible - your restriction is your blocks.

When you say 'limited', in what way are they limited?

So, you are saying the entire loop is running faster in that parallel setup than in the single rad in that test loop?

That wouldn't make sense - 1 single radiator has less restriction and pressure drop than your entire loop and components being run in it, even in parallel.

You're still bound in a parallel loop by the most restrictive point, though, although those are fewer in nature due to the dual paths being configured.

And yes, GPU blocks are typically fairly restrictive, which is why those SLI/X-Fire bridges exist to parallel. Full cover blocks are also more restrictive (by nature) than universal blocks, so you can get away with serial Uni blocks more than you can with serial FC blocks.

Can you imagine how restrictive RAM and other MB blocks are now based on your testing?

In your RAM/MB blocks or all blocks in general? I was meaning 'all of those blocks in general'...many are restrictive.

Right, it was a generalized statement, not about your specific blocks. I actually forgot you had them. Historically, MB and RAM blocks have been pretty restrictive, especially those MB blocks that snake from NB/SB/MOSFETs.

I see the point you are attempting to make series vs parallel, but I don't see the additional expense to the water cooling being a necessary need.

We have enough problems at Toms just getting users to read the stickies and help themselves,
can't you see that you are so over complicating the issue to the point you'll literally scare some completely away from water cooling?

I have a killer below ambient water cooling system, but I do not stick it in every ones face with links in every thread I enter,
because it is specialized cooling and if anyone is interested they usually are searching for peltier cooling information,
and then if they're searching for that type of information I'll link to it.

Only rich people with more money than sense IMO, would even run a 4 way SLI, those of us that would avoid a setup like that is not because of the cooling,
but because even today after all this time of SLI or Crossfire setups even 2 way, there are driver bugs and issues and some games do not run SLI smooth at all.

Just when you think all the SLI issues are resolved new drivers come out with new problems it is a never ending cycle,
I bought my GTX Titan to get away from SLI bugs running my 2 580GTX, the same cards you have,
some SLI issues I didn't even realize were happening until they disappeared with the single card.

So I understand you are attempting to prove a point about your parallel vs series and flow rate,
but try to understand the more you complicate the issue the more that would be interested you loose.

Additionally when you post your pictures, please clean up those nasty radiators.

Now to the reason I quoted your sketch:

First of all I would never do what you're proposing in your sketch,
It takes a high dollar motherboard to run 4 way SLI, 4 GPU water blocks at a minimum of 100 bucks each and the 4 GPUs as well,
I do not think it is worth it but it is your money.

Iooking at your sketch what if you have a partial blockage in one of the GPUs?

I know you think your inline screens will trap any problems but what if the problem is after the screen when you start up the loop?

It's not like you live in a sterile dust free environment your radiators prove that?

Instead of the center radiator splitting the return to both reservoirs, why not just make one large reservoir?

Why have the CPU at the end of the parallel loop, to me it makes more sense to have it at the beginning?

Because the CPU is not in parallel it is in series because you cannot run a 2 port (Inlet/Outlet), CPU in true parallel.

You may pull it off with Swiftechs 4 port water block?, but not the traditional flow CPU water block.

But that may cool even worse!

Though series vs parallel alone even with a simple setup, is an ongoing argument of whether the cooling performance rewards outweigh the expense,
keeping in mind your setup should be all about performance because it is not about looks and convenience.

And with my setup having an additional case housing all the cooling, neither is mine!

But please don't scare away our first time water coolers over complicating the issue!

http://www.tomshardware.com/answers/id-2786480/good-water-cooling-parts.html

It would be best if you only post your links when someone is curious regarding series parallel,
the same as I only post mine when someone is inquiring regarding peltier cooling.

Thanks for reading TM_03!

Ryan

Any time you add anything to a loop, you introduce restriction: inches of tubing, fittings, radiators, blocks, etc. You aren't decreasing restriction by adding these, so I don't know how you can say this is possible. Any time water turbulence within the tubing diameter flows back into itself, it causes turbulence, which causes restriction. This means for each inch of tubing, you cause this or each time tubing intersects with a fitting, the lip of the fitting causes turbulence.