WC Question

JSimmons · 04-07-2003, 05:42 PM

After a few weeks of research and comparing, and measuring, evaluating, and the re-measuring, I have ordered my Danger Den blocks, and other associated water cooling nik-naks, and all of it should be here by the end of the week. I ordered:

Maze3 CPU block for AMD (all copper)
GPU Block for 9700 Pro (all copper)
Black Ice Extreme radiator
120mm Pabst fan
Via Aqua 1300
BayRes reservoir

Of course, I have a question (probably more than one, but I'll go easy on you guys at first since I'm new here).

I was plannning on the following general sequence of components:

-->Pump-->CPU-->GPU-->Radiator-->Reservoir

My question is this: Wouldn't it make more sense to cool the CPU and GPU in parallel instead of in series? That way, the CPU and GPU each get their own supply of equally cooler water instead of cooling the CPU on the way to cooling the GPU (which would not benefit from the cool water the same way the CPU did).

Remember, I'm new to this water-cooling stuff, so if what I said is just plain stupid or severely out of whack, let me down easy.

Oh yeah, I'm not a "look at all the pretty lights" type, so I don't necessarily care about aesthetics beyond getting it all in the (full tower) case so I can put the sides back on.

(EDIT: Just after leaving this message, I got an email announcing that my danger den stuff is officially "on the truck". - wOOt!)

winewood · 04-07-2003, 06:49 PM

The following is an opinion:
If you run in parallel then you take away the velocity of the water. If you run in series you maintain your greatest velocity at the cpu and then have good flow to your gpu. Remember your gpu doesn't require near as much cooling as your cpu, and your water temps will not be too much hotter after leaving your cpu and maintains a large level of cooling capacity at 2nd stage.
Imagine if you were using a garden hose and wanted to create a spray. If you cut the flow by 1/2 then you won't have the pressure to create as powerful a spray. Besides, if one block is a higher resistance, then the water will seek an alternate path of least resistance this may leave it impossible to calculate efficiency later on.
Thats the way I see it. Feel free to correct me if im wrong.

JSimmons · 04-08-2003, 07:59 AM

I didn't think the split itself would matter much since the "Y" fitting would be within a couple of inches of the pump output fitting. At that point, it seems the velocity of the water wouldn't have been reduced too much, so the effect of splitting the flow in half and running each leg over one chip would actually result in a more effective cooling solution.

Would a beefier pump (say an Aqua 2600 rated at 740 gph) pretty much solve that issue?

I wonder if any of the waterblock manufacturers consider matched resistance an important factor in the design of their water blocks. How do you measure the resistance of a waterblock? I imagine that resistance on a given leg of a parallel system could be tuned to a certain point so that all parallel legs have a reasonably similar value.

I didn't mention it, but I am planning on using 1/2-inch fittings on all of the components.

winewood · 04-08-2003, 08:22 AM

A pump that could run 2 systems by itself would definitely eliminate most concerns

Of course you could use a much cheaper pump in series. BillA measures flow rate and such. You can see if he has measured a block out there, and determine which pump and flow you need for optimal performance.
His link is .. www.thermal-management-testing.com

DigitalChaos · 04-08-2003, 09:01 AM

your final answer will be you trying both configs... and seeing what provides the best results.

but the "pressure" (what your really looking at is velocity/speed) remains constant through your entire loop UNLESS you start splitting the stream.

but think of it this way... if 1" of water goes in at the beginning of the loop, 1" of water must come out at the end of the loop.. and it will come out at the same speed as it is comming in.

when you split the flow into 2 lines.. that 1" going in at the beginning looks like 1/2" to each line (this also means half the velocity of water). of course at the end everything combines back into one line.. and its that 1" of water again.

now the resistance of the split.. the bends, and anything else that slows down your stream... this will effect the ENTIRE LOOP.
If you make a kink in your line, your not just slowing down the velocity after the kink.. the entire loop slows down.

04-07-2003, 05:42 PM	#1
JSimmons Cooling Savant Join Date: Apr 2003 Location: 4-sided room with an exit going east, and an exit going south Posts: 392	WC Question After a few weeks of research and comparing, and measuring, evaluating, and the re-measuring, I have ordered my Danger Den blocks, and other associated water cooling nik-naks, and all of it should be here by the end of the week. I ordered: Maze3 CPU block for AMD (all copper) GPU Block for 9700 Pro (all copper) Black Ice Extreme radiator 120mm Pabst fan Via Aqua 1300 BayRes reservoir Of course, I have a question (probably more than one, but I'll go easy on you guys at first since I'm new here). I was plannning on the following general sequence of components: -->Pump-->CPU-->GPU-->Radiator-->Reservoir My question is this: Wouldn't it make more sense to cool the CPU and GPU in parallel instead of in series? That way, the CPU and GPU each get their own supply of equally cooler water instead of cooling the CPU on the way to cooling the GPU (which would not benefit from the cool water the same way the CPU did). Remember, I'm new to this water-cooling stuff, so if what I said is just plain stupid or severely out of whack, let me down easy. Oh yeah, I'm not a "look at all the pretty lights" type, so I don't necessarily care about aesthetics beyond getting it all in the (full tower) case so I can put the sides back on. (EDIT: Just after leaving this message, I got an email announcing that my danger den stuff is officially "on the truck". - wOOt!)

04-07-2003, 06:49 PM	#2
winewood Cooling Savant Join Date: Nov 2002 Location: in my chair Posts: 574	The following is an opinion: If you run in parallel then you take away the velocity of the water. If you run in series you maintain your greatest velocity at the cpu and then have good flow to your gpu. Remember your gpu doesn't require near as much cooling as your cpu, and your water temps will not be too much hotter after leaving your cpu and maintains a large level of cooling capacity at 2nd stage. Imagine if you were using a garden hose and wanted to create a spray. If you cut the flow by 1/2 then you won't have the pressure to create as powerful a spray. Besides, if one block is a higher resistance, then the water will seek an alternate path of least resistance this may leave it impossible to calculate efficiency later on. Thats the way I see it. Feel free to correct me if im wrong. __________________ -winewood-

04-08-2003, 08:22 AM	#4
winewood Cooling Savant Join Date: Nov 2002 Location: in my chair Posts: 574	A pump that could run 2 systems by itself would definitely eliminate most concerns Of course you could use a much cheaper pump in series. BillA measures flow rate and such. You can see if he has measured a block out there, and determine which pump and flow you need for optimal performance. His link is .. www.thermal-management-testing.com __________________ -winewood-

04-08-2003, 09:01 AM	#5
DigitalChaos Cooling Savant Join Date: Jan 2002 Location: Waukesha, Wi Posts: 698	your final answer will be you trying both configs... and seeing what provides the best results. but the "pressure" (what your really looking at is velocity/speed) remains constant through your entire loop UNLESS you start splitting the stream. but think of it this way... if 1" of water goes in at the beginning of the loop, 1" of water must come out at the end of the loop.. and it will come out at the same speed as it is comming in. when you split the flow into 2 lines.. that 1" going in at the beginning looks like 1/2" to each line (this also means half the velocity of water). of course at the end everything combines back into one line.. and its that 1" of water again. now the resistance of the split.. the bends, and anything else that slows down your stream... this will effect the ENTIRE LOOP. If you make a kink in your line, your not just slowing down the velocity after the kink.. the entire loop slows down. __________________ http://www.digitalsimplicity.com/eatme/

04-08-2003, 07:59 AM	#3
JSimmons Cooling Savant Join Date: Apr 2003 Location: 4-sided room with an exit going east, and an exit going south Posts: 392	I didn't think the split itself would matter much since the "Y" fitting would be within a couple of inches of the pump output fitting. At that point, it seems the velocity of the water wouldn't have been reduced too much, so the effect of splitting the flow in half and running each leg over one chip would actually result in a more effective cooling solution. Would a beefier pump (say an Aqua 2600 rated at 740 gph) pretty much solve that issue? I wonder if any of the waterblock manufacturers consider matched resistance an important factor in the design of their water blocks. How do you measure the resistance of a waterblock? I imagine that resistance on a given leg of a parallel system could be tuned to a certain point so that all parallel legs have a reasonably similar value. I didn't mention it, but I am planning on using 1/2-inch fittings on all of the components.

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)