Serial vs. Parallel Loop

[anubus45]

So, I'm building my first water-cooled setup (3930K + GTX 670) next week, and I wanted to query for a couple opinions.

I'm building using Swiftech's H20-320 Edge HD kit as the base for my water-cooling and I'm also adding my GPU to the loop. Since the Apogee HD and the MCR-320 are designed with the possibility for a parallel loop in mind, and with that, my plan thus far has been to have a setup similar to the following:

MCR-320===CPU<===MCR-320
===GPU==T==MCR-320
(the "T" is a drain)

I've been thinking about it though, and I'm not sure how much benefit I would gain in temps (or if it would perhaps be detrimental because of slower flow rate for the GPU?). So I want to know what everyone else thinks and if I should run it in the above partial parallel loop, or just have a traditional serial loop of (MCR-320===CPU===GPU==T==MCR-320).

As far as upgrades go, I may, at some point add a second 670, and I'll add that to my loop if I do, but I doubt I'll cool anything beyond that. RAM and HDD blocks I think are too restrictive and provide minimal benefit, and Chipset (Rampage IV Extreme) is very much a maybe, since while it might be a 'little' helpful, I don't think the cost is very justified.

Some additional info about my loop:
Case: Aerocool Strike-X ST
GPU Block: HeatKiller GPU-X3 670
Tubing: Primoflext Advanced LRT 1/2 ID 3/4 OD (Red)
Coolant: Distilled Water and PT Nuke or a KillCoil (Or both?)
Drain: Q Fitting with a temp sensor and a a length of tube with a stop fitting. (KillCoil here?)

Rad is going in the top, same as here http://www.tomshardware.com/reviews/switch-810-cosmos-ii-strike-x-hurrican-2000,3151-3.html, but since the fans have to be on the top, I'm probably going to have them pulling air up to fit with the airflow of the case, unless it's highly recommended to have them pushing the air down? I may also go for a push/pull config to not worry so much about the issue and generally get better performance.

---------

UPDATE:
Computer has been built. Details can be found at my build log here: http://www.tomshardware.com/forum/id-1656048/build-log.html

 

[toolmaker_03]

This is coming from some one who has done that!!!

Unless you are having problems with flow being too slow already, a parallel setup is not needed!!!

Here is my build log have a read for yourself

http://www.tomshardware.com/forum/274855-29-experimental-radiator-build/page-10

 

[4Ryan6]

This is coming from some one who has done that!!!

Unless you are having problems with flow being too slow already, a parallel setup is not needed!!!

Here is my build log have a read for yourself

http://www.tomshardware.com/forum/274855-29-experimental-radiator-build/page-10

Wow!

I have to give you the highest of credit honors for admitting that TM03!

Thank You! Ry

 

[toolmaker_03]

Thanks, but I was kind of hoping that he would read the thread and replay back with his decision on how he planned to build his loop.

 

[anubus45]

Thanks, but I was kind of hoping that he would read the thread and replay back with his decision on how he planned to build his loop.

Heh, I've been away from a computer for most of the weekend. Looking at your log, and the more I consider it, I'm leaning towards just the serial loop setup for my setup as it will currently stand. For future upgrades, I may revisit the possibility of a parallel loop, particularly if I add a chipset block since that will be more restrictive, but we'll see how that goes when the time comes.

 

[toolmaker_03]

Depending On the type of block it is it may not be restrictive at all, a good example of this is the koolance 400 mosfet block for mobo’s, since there is no type of matrix inside of the block, in fact it is totally open on the inside, allowing for very high flow rates. Basically it works like this, the tighter or closer to each other the matrix is on the inside of the block, the more restrictive it will be because the water must pass through it.

 

[4Ryan6]

Thanks, but I was kind of hoping that he would read the thread and replay back with his decision on how he planned to build his loop.

Heh, I've been away from a computer for most of the weekend. Looking at your log, and the more I consider it, I'm leaning towards just the serial loop setup for my setup as it will currently stand. For future upgrades, I may revisit the possibility of a parallel loop, particularly if I add a chipset block since that will be more restrictive, but we'll see how that goes when the time comes.

Chipset along with the VRM water blocks are not necessary, but what is necessary is replacing the needed airflow once the stock air cooler is removed and replaced with any type of water block.

The stock air cooler not only cools the CPU but also the surrounding VRMs and on some motherboards the VRMs and Chipset is heat pipe linked.

So once that stock cooler is removed you have to replace the airflow you removed, because they are air cooled.

If you do not replace the airflow the VRMs will run hotter than they were designed to run and you essentially will be shortening the life span of your motherboard.

 

[anubus45]

Thanks, but I was kind of hoping that he would read the thread and replay back with his decision on how he planned to build his loop.

Heh, I've been away from a computer for most of the weekend. Looking at your log, and the more I consider it, I'm leaning towards just the serial loop setup for my setup as it will currently stand. For future upgrades, I may revisit the possibility of a parallel loop, particularly if I add a chipset block since that will be more restrictive, but we'll see how that goes when the time comes.

Chipset along with the VRM water blocks are not necessary, but what is necessary is replacing the needed airflow once the stock air cooler is removed and replaced with any type of water block.

The stock air cooler not only cools the CPU but also the surrounding VRMs and on some motherboards the VRMs and Chipset is heat pipe linked.

So once that stock cooler is removed you have to replace the airflow you removed, because they are air cooled.

If you do not replace the airflow the VRMs will run hotter than they were designed to run and you essentially will be shortening the life span of your motherboard.

Yeah, I generally agree with you there. The only reason I'm even considering VRM/Chipset blocks is because VRM's can 'sometimes' use cooling with heavy OCing, but the practical side of me keeps telling me it's not necessary.

As far as airflow, the case has a spot for a 120mm fan behind the CPU, which will help cool the rear of the board, and I'm also going to try to finagle the 200mm fan from the top as an intake on the side panel blowing onto the board. So I'm hoping between both of those, it will keep things cool even with the loss of a fan on the CPU.

Also, for airflow
This is the airflow setup for the case as recommended by Aerocool:

This is the airflow as recommended by Swiftech

Obviously, contradictory. I'm leaning towards having the airflow go UP and pulling air through the rad and exhausting it outside of the case, since otherwise I'm trying to fight the natural flow of air since heat rises, but what do you guys think?

 

[hdeezie80]

No that's actually the best way to set up your fans for top mounted radiators since the actual water temps can only get as cold as what ever air is being pushed through them if your pushing the hot air from the ram/vrms than your water temps will be hotter than the other way around. Probably the only situation where you'll really benefit from push pull since the fans facing the motherboard will be actively cooling it and the ram as well as moving air through the radiator. It's obviously more optimal to mount the radiator in the bottom of the case, but there's not a whole lot of cases that really give you that option. You should look into a standalone pump/radiator though.

 

[toolmaker_03]

If the flow direction is out of the rad into the video cards then it might??? be fine, but if not and it is going through the processor first, it will have issues, because the flow will drop to a level that is too slow for the loop, not to mention the rad itself is rather restrictive. Is that going to be a “monstra” rad on this??? They have real good flow through them, in fact almost all of the thick rads have a good flow rate through them. Making it easy to achieve good flow through the loop, what video blocks will be in this loop?? And I think it will need a better motor to push all of the water through all of the blocks. It might have been able to handle it with one card added(might have handled it!!!) but not two. At this point I would reconsider the "base" as it was stated and build a custom loop for this system.

 

[4Ryan6]

If the flow direction is out of the rad into the video cards then it might??? be fine, but if not and it is going through the processor first, it will have issues, because the flow will drop to a level that is too slow for the loop, not to mention the rad itself is rather restrictive. Is that going to be a “monstra” rad on this??? They have real good flow through them, in fact almost all of the thick rads have a good flow rate through them. Making it easy to achieve good flow through the loop, what video blocks will be in this loop?? And I think it will need a better motor to push all of the water through all of the blocks. It might have been able to handle it with one card added(might have handled it!!!) but not two. At this point I would reconsider the "base" as it was stated and build a custom loop for this system.

If the flow direction is out of the rad into the video cards, then he better not have any overclocking the CPU intentions at all, because he'll be preheating the water going to it!

@anubus45

Your bottom picture shows 2 water cooled GPUs in a single loop, in that setup scenario you need to run 2 loops, one for the CPU and the other for the GPUs.

Two full coverage water blocked GPUs produce a lot of heat.

 

[Shadow Guardian]

OK, I really have no idea what is up with this forum, but if I'm using my normal account (I'm also Anubu45), it will absolutely not let me reply or post in this thread at the moment.

Anyways, to clear things up.

@hdeezie80
One of my main reasons for not wanting to use the Rad fans as an intake is because that leaves me with a disproportionate amount of intakes to exhausts. 200mm front fan, 120mm side fan, 200mm side fan, and the 3/6 120mm top/rad fans as intakes, with only a single 140mm exhaust. But it may be something to re-examine if using the rad fans as an exhaust is generally agreed to be worse?

@ TM3 and Ryan
As I originally noted in my first post, and as you guys swayed me even more towards; My loop order is going to be:
MCR-320===CPU===GPU===Drain===MCR-320

The diagrams I posted were only for the sake of showing the recommended airflow patterns as recommended by Aerocool and Swiftech respectively. For the time being, I'm only cooling a 3930K and a single GTX 670 (Heatkiller GPU-X3 670 Waterblock).

An upgrade to SLI is a potential future upgrade, though I believe that if I do so, I'd probably be looking at adding a second rad and pump to go with it.

 

[4Ryan6]

OK, I really have no idea what is up with this forum, but if I'm using my normal account (I'm also Anubu45), it will absolutely not let me reply or post in this thread at the moment.

Anyways, to clear things up.

@hdeezie80
One of my main reasons for not wanting to use the Rad fans as an intake is because that leaves me with a disproportionate amount of intakes to exhausts. 200mm front fan, 120mm side fan, 200mm side fan, and the 3/6 120mm top/rad fans as intakes, with only a single 140mm exhaust. But it may be something to re-examine if using the rad fans as an exhaust is generally agreed to be worse?

@ TM3 and Ryan
As I originally noted in my first post, and as you guys swayed me even more towards; My loop order is going to be:
MCR-320===CPU===GPU===Drain===MCR-320

The diagrams I posted were only for the sake of showing the recommended airflow patterns as recommended by Aerocool and Swiftech respectively. For the time being, I'm only cooling a 3930K and a single GTX 670 (Heatkiller GPU-X3 670 Waterblock).

An upgrade to SLI is a potential future upgrade, though I believe that if I do so, I'd probably be looking at adding a second rad and pump to go with it.

Affirmative!

 

[hdeezie80]

Honestly you should stray away from that kit all together, I just looked at the pump specs max flow rate is 17.5 LPH compared to a d5 with a max flow rate of 1200 LPH, or a mcp 350 with a max flow rate of 440LPH combined with the . If you go with a good pump like a d5 from the beginning you wont need to worry about flow rate on a regular setup, if your serious about overclocking you should focus more on rad space an MCR-320 is not going to perform exceptionally well on a system like this. This kit would be a much better value @ $30 more, it comes with a d5 and full sized 360 -> RX360 kit. I would recommend another 240 mm x 60 mm radiator on top of that. I currently have 2 x 240 AX240 (40 mm thick) radiators with 9 additional case fans. They perform well cooling wise but are way louder than I would like them to be so i wouldn't suggest it if your going for a quiet loop, plus they are getting replaced with an RX 480 and an RX 360 as soon as my new case comes in.

 

[rubix_1011]

Most of the Swiftech kits use a MCP35x pump or variation of the DDC MCP355.

 

[Shadow Guardian]

Honestly you should stray away from that kit all together, I just looked at the pump specs max flow rate is 17.5 LPH compared to a d5 with a max flow rate of 1200 LPH, or a mcp 350 with a max flow rate of 440LPH combined with the . If you go with a good pump like a d5 from the beginning you wont need to worry about flow rate on a regular setup, if your serious about overclocking you should focus more on rad space an MCR-320 is not going to perform exceptionally well on a system like this. This kit would be a much better value @ $30 more, it comes with a d5 and full sized 360 -> RX360 kit. I would recommend another 240 mm x 60 mm radiator on top of that. I currently have 2 x 240 AX240 (40 mm thick) radiators with 9 additional case fans. They perform well cooling wise but are way louder than I would like them to be so i wouldn't suggest it if your going for a quiet loop, plus they are getting replaced with an RX 480 and an RX 360 as soon as my new case comes in.

Heh, wishing no offense, but part recommendations are pretty moot at this point, since I bought all my water cooling stuff weeks ago, I just didn't have the rest of my stuff until this week, so at this point I'm just asking minor design questions (ie. Serial vs. Parallel and airflow). I do appreciate the advice though. Out of curiosity though, where did you get the pump specs? As rubix pointed out, it's the MCP35x pump, which by all accounts that I've found, is a very good pump. Max flow rate isn't 17.5 LPH, it's ~17.5 LPM, which is about 1050 LPH.
A second rad is something I'll definitely be considering in future. I'll have to see how well I can manage a 240.2 in front of the hard drive cage, and that will most likely come if I upgrade to SLI.

 

[hdeezie80]

Most of the Swiftech kits use a MCP35x pump or variation of the DDC MCP355.

Ah realized it was LPM on the swiftech website lol, so the 17.5 * 60 equates to 1050 LPH, totally fine for a GPU/CPU loop in my opinion, but one tip for the x79 platform would be to make sure your VRM's are getting enough airflow, I had to strap an 80 mm fan to mine to hit 4.6 GHz.