bluphysted
07-23-2008, 08:02 AM
Fellas..
I'm new to watercooling, I'll just get that out of the way first.
I plan to WC the following gear.
EVGA 790i Ultra (everything I should chipset wise)
2 harddrives (maybe)
Intel E8400
2 x BFG 9800 GTX OCX (SLI)
From what I've been reading, it's generally not recommended to go with some sort of a parallel path manifold approach, and instead, just build a loop with all the devices in series. I've seen a few pics of people who have used Y connectors to split the water prior to the devices as well.
I was thinking, that with the series approach, everything downstream is receiving warmer and warmer water. If you had only 1 loop, say CPU > GPU > GPU, that third device is getting some rather warm water going over it.
I always had the thought of 2 complete, independent loops, each loop with its own pump, rad, rez.. One loop for CPU and mobo, and the other for the two GPUs.
My question is.. If you had two identical GPU blocks, with similar (if not identical) flow restrictions, could you not use a Y connector to split the feed from the pump, so that each GPU receives cool water, right out of the rad?
I understand the issue of the manifold approach, where water is likely to flow more through the path of least resistance, but wouldn’t creating 2 parallel paths to feed the SLI GPUs be a better idea than the series approach. The 2nd GPU would not receive the same cooling as the 1st in a series path.
I guess my biggest concern are the GPUs, as the CPU would be the first in line from loop1, and the rest of the chipset can suffer a little.. But the GPUs are hot, and a series loop just doesn’t seem like the most effective approach..
Anyhoo, just curious.. Any help/insight to this is appreciated.
I'm new to watercooling, I'll just get that out of the way first.
I plan to WC the following gear.
EVGA 790i Ultra (everything I should chipset wise)
2 harddrives (maybe)
Intel E8400
2 x BFG 9800 GTX OCX (SLI)
From what I've been reading, it's generally not recommended to go with some sort of a parallel path manifold approach, and instead, just build a loop with all the devices in series. I've seen a few pics of people who have used Y connectors to split the water prior to the devices as well.
I was thinking, that with the series approach, everything downstream is receiving warmer and warmer water. If you had only 1 loop, say CPU > GPU > GPU, that third device is getting some rather warm water going over it.
I always had the thought of 2 complete, independent loops, each loop with its own pump, rad, rez.. One loop for CPU and mobo, and the other for the two GPUs.
My question is.. If you had two identical GPU blocks, with similar (if not identical) flow restrictions, could you not use a Y connector to split the feed from the pump, so that each GPU receives cool water, right out of the rad?
I understand the issue of the manifold approach, where water is likely to flow more through the path of least resistance, but wouldn’t creating 2 parallel paths to feed the SLI GPUs be a better idea than the series approach. The 2nd GPU would not receive the same cooling as the 1st in a series path.
I guess my biggest concern are the GPUs, as the CPU would be the first in line from loop1, and the rest of the chipset can suffer a little.. But the GPUs are hot, and a series loop just doesn’t seem like the most effective approach..
Anyhoo, just curious.. Any help/insight to this is appreciated.