DavidNJ

Here is my calculation: it takes 260 or so watt-minutes to raise a gallon of water 1 degree. In a water cooling system, the temperature in the loop varies then by 1-2 degrees, depending on the heat transferred and the rate of flow. So, even if there is a 250w CPU followed by two 200w graphic cards, in a system appropriate for that load (say 2.5 gal/min) the temperature of the water would still rise only a degree or so.

It shouldn't make much difference to the temperature of the component whether the water is 32C or 33C...at the beginning of the loop or at the end. It should vary more than than with normal changes in ambient temperature.

Therefore, why is it beneficial to have multiple loops?

starlon

It's good to have restrictive impingement waterblocks in their own loop.

By-Tor

As starlon said, many use restrictive blocks in our loops.

I'm using a 5.5 nozzle in my D-tek fuzion and really didn't want to hurt the flow through my GPU blocks. So I have 2 loops in my rig now.

CPU Loop: D-Tek Fuzion, Swiftech MCP355 w/petra's top, BIX 240 rad.

GPU Loop: Swiftech MCW60 x2, Swiftech MCP350 pump w/petra's top, BIX 360 rad.

__________________
3D Mark 06= 21557

Intel Rig:
DFI DK x38-T2R MB, Intel e8500 E0 @ 4.5 ghz, 2 gigs Crucial Ballistix Tracers, 2 Visiontek 4870's xfired, WD Raptor, Thermaltake DuOrb VGA cooler, D-Tek Fuzion V2 CPU block, MCP350 pump w/Petra's top, Micro-res., BIX240 rad. w/151 CFM delta fans, 610 watt PSU

http://www.overclock.net/case-mod-wo...rk-spider.html

http://www.overclock.net/computer-ca...h-station.html

dralb

Well, I could be very wrong here, but I would say because there is very little fluid. Much, much less than a gallon. Even if the temp increase is measured in gals, that needs to be divided to reflect the amount of liquid in a loop. I have less than a 1/2 liter in mine (if I remember correctly, around 400cc)

Also, as has been stated, flow. Besides, every C counts



edit: When I filled my WC loop for the first time, I was shocked at how little fluid was needed.

__________________
Antec 1000 Case Mod Worklog

3358 Windsor 3800


"[Vietnam] only made billionaires out of millionaires. Today's war is making trillionaires out of billionaires. Now I call that progress." -Kurt Vonnegut

CD in response to whether folding is a waste:

DavidNJ

The typical system, according to Martin's spreadsheet, flows 1-3 gallons per minute.

No matter how restrictive your CPU block, you need to pump fluid through it! If the pump from the second loop was in series in the first, it would move the same fluid through both the CPU and GPU blocks.

dralb

Yeah, but flow rate is not quantity. I can move the same three oz. at 50 gal/min. It would still heat up more as there is less volume to hold/transfer/remove the heat. Also, try adding blocks to his estimater and see what happens to flow rate.


edit: Are you also implying running 2 pumps in one loop?


edit2: Also, since GFX card run much hotter than CPU's, even if the delta were the same, it would inhibit the cooling of the CPU.

__________________
Antec 1000 Case Mod Worklog

3358 Windsor 3800


"[Vietnam] only made billionaires out of millionaires. Today's war is making trillionaires out of billionaires. Now I call that progress." -Kurt Vonnegut

CD in response to whether folding is a waste:

DavidNJ

Not so said the frog. You would have less capacitance, but you lose the heat in radiator. It only cares about the temperature differentials between its surfaces and the two fluids (air and coolant) and its resistance to transfering heat between the two. If volume counted, you would have a 2 gallon reservoir.

Ira?

TonyL222

In general, I agree with you - I don't see how there can be much cooling difference between a single and dual loop, given that flow rate in either case is greater than 1 GPM (I'm counting on Martin's testing here). The only way I can see a difference is if the heat generation/dissipation capacity is not separated evenly in the dual loop. For example if the rads were a 220 and a 360, breaking them into two loops might give more dissipation capacity to the 360 loop. In the case of a cpu loop and a gpu loop, the component in one loop might run cooler and the one in the other loop might run warmer than each did in a single loop.

That's how I see it. But then again, I don't see where loop order can have any significant impact on component cooling, either. as I said in an earlier thread, the fluid in my system is moving at almost 5'/sec. That means the line between "coolest" and "warmest" water is blurred ( the water temp across the loop tends to equalize).

DavidNJ

I don't think the temps equalize. There clearly is hotter water going into the radiator the coming out. However, the difference is only a degree or so. The faster moving the water the lower the difference.

In the cases where there is a high restriction water block for the CPU, having the units in a single loop should help. Using the second pump in series instead of its own loop would increase the pressure and flow rate in the high restriction CPU water block, increasing its effectiveness.

That also begs the question: should GPUs be cooled with high restriction jetted water blocks instead of the full coverage blocks often used?

TonyL222

The key word was tend to equalize. I don't doubt that if you measure the temp at a point just leaving the rad, that the temp will be cooler than that at a point just entering the rad. If not, the rad is useless. But the water is not static. In about a second it has traveled almost 5' (in my rig with my pumps), which is probably about the size of my entire loop (same with the "warmest" water in the loop). All of it is in constant motion at a relatively high velocity. I agree with the your last statement. Any temp difference across the loop is probably 1-2C at worst. If it's more than that, then there may be a problem with your flow rates.


A second pump in series would "help" only if flow rates with one pump is less than 1 GPM. I am relying on Martin's testing that found that flow rates of 1 GPM to 1.5 GPM was optimal and that flow rates over that don't significantly impact the cooling.

I think that some have a false impression of full coverage blocks. In some (if not all) cases, the heat from the memory and mosfets is NOT added to the loop. Rather the the wide part of the full cover blocks are just passive heat sinks and don't rely on or touch the water. In the case of the Swiftech Stealth, Swiftech tested a loop with a Stealth and then with a MCW60. The temp of the cpu in that loop was the same in either case:

Stealth (V1) Vs. MCW60: comparative impact on CPU temps:
http://www.swiftech.com/forums/viewtopic.php?t=79