Some questions regarding WC (Beginner)

[yousif666]

Hi

I am new to water cooling stuff and i am planning on building a new pc, my friend has a water cooling and i asked him for advice so he told me that i should get a 120.4 for the GPU and 120.3 or 120.2 for CPU.

and for the case he told me about caselabs, so i was sondering if a regular full tower will be ok with me and i will get alot of space, the space is what i worry about that is why he suggested the caselabs.

and btw i dont want to buy the kits, i want to build it custom so what i am wondering here is that do i really need 2 radiators for my loop?

here are the specs i will be buying for my new PC:

VGA: EVGA GTX 980 (for now i will get 1 only but later i will be doing SLI so total of 2 VGA)
CPU: Intel i7 5930K

I will water cool the VGA and CPU so i was wondering if i need 2 radiators or just 1 120.4 will be enough for gpu+cpu water cooling regarding if its 2 GPUs in SLI or 1 GPU only.

And also he suggested to install 2 pumps because i will be installing 2 radiators.

so what are your suggestions please.

Thanks.

 

[Lvcoyote]

Do you plan to overclock the CPU and graphics card(s)?

 

[yousif666]

yes, otherwise i wont install water cooling, i faced alot of problems with my old system in overclocking with air cooling and my system is 7 years old so that is why i decided to go with water cooling.

 

[Lvcoyote]

A 120.4 will be fine for a CPU and GPU. The new 980 Series graphics cards are said to run cooler than previous generations, so it might even be good for SLI too. Here is what I would do.
1. Water cool the initial CPU and single GPU with that 120.4
2. Later on if you add another video card, attach it to you existing loop.
3. If you find adding the 2nd GPU makes things to hot for your liking, then you can add another loop then.

 

[yousif666]

A 120.4 will be fine for a CPU and GPU. The new 980 Series graphics cards are said to run cooler than previous generations, so it might even be good for SLI too. Here is what I would do.
1. Water cool the initial CPU and single GPU with that 120.4
2. Later on if you add another video card, attach it to you existing loop.
3. If you find adding the 2nd GPU makes things to hot for your liking, then you can add another loop then.

So even if i added another radiator do i also have to install a second pump? because if i will install another radiator i will install one in the bottom and the other on the top.

Is it important to water cool the RAM also for the overclocking?

 

[Lvcoyote]

RAM doesn't need water, active air in worst case for that. As far as needing another pump goes, that depends on a variety of things such as waterblock flow restriction and the GPH of the pump. Again, the best thing to do is go with the basics until you know what you need, more isn't always better

 

[Daddyjaxx]

Two pumps are good for a backup. If one goes out, you still have the other running. I'm running 4 for two loops. One would definitely be good enough for my CPU/motherboard loop. I'm using two on my GPU loop because it's going through 3 GPUS and 3 rads. One is probably enough, but better safe than sorry for 2500.00 worth of video cards.\

Case Labs cases generally come in two sizes...big and humongous. Just get your wallet out. They come with absolutely nothing but the case and you get to build it. No rad mounts, fans, USB or sound ports...nothing except the hard drive (usually just two), SSD, and front bays.

 

[yousif666]

RAM doesn't need water, active air in worst case for that. As far as needing another pump goes, that depends on a variety of things such as waterblock flow restriction and the GPH of the pump. Again, the best thing to do is go with the basics until you know what you need, more isn't always better

Yeah that's what i thought too.

Two pumps are good for a backup. If one goes out, you still have the other running. I'm running 4 for two loops. One would definitely be good enough for my CPU/motherboard loop. I'm using two on my GPU loop because it's going through 3 GPUS and 3 rads. One is probably enough, but better safe than sorry for 2500.00 worth of video cards.\

Case Labs cases generally come in two sizes...big and humongous. Just get your wallet out. They come with absolutely nothing but the case and you get to build it. No rad mounts, fans, USB or sound ports...nothing except the hard drive, SSD, and front bays.

The pump is not a big deal for me, but yes as you said, a backup is always better, but as a starter the total charges of caselabs are going to be $928 including shipping charges so i started to think of getting a normal Case instead of caselabs, and also getting 1 radiator 480 for the gpu+cpu but my friend told me since the gpu is going to bush out hot water it will go to the cpu then the radiator but i am trying to think of another loop, if you guys can help i would appreciate it.

 

[mgrisoli]

Yeah that's what i thought too.



The pump is not a big deal for me, but yes as you said, a backup is always better, but as a starter the total charges of caselabs are going to be $928 including shipping charges so i started to think of getting a normal Case instead of caselabs, and also getting 1 radiator 480 for the gpu+cpu but my friend told me since the gpu is going to bush out hot water it will go to the cpu then the radiator but i am trying to think of another loop, if you guys can help i would appreciate it.

I would have one pump with reasonable GPH and a splitter to not pump hot water on the GPU you create two loops, one for CPU, one for GPU and you merge them back before going in the radiator.
The backup pump you can save in your drawer in case of any incident.

 

[yousif666]

I would have one pump with reasonable GPH and a splitter to not pump hot water on the GPU you create two loops, one for CPU, one for GPU and you merge them back before going in the radiator.
The backup pump you can save in your drawer in case of any incident.

I guess i will be installing 2 rads, 360 for the CPU and 480 for the GPU so that way i dont need to split, all i do care about is the temp because i am living in a very hot weather and the AC isn't always on in the room so that setup would satisfy me better.

 

[billb]

No, never run components in parallel. It reduces the flow rate to each component that's in parallel, and any test report on any component will show that a lower flow rate equals lower performance. Components in serial only.

 

[yousif666]

No, never run components in parallel. It reduces the flow rate to each component that's in parallel, and any test report on any component will show that a lower flow rate equals lower performance. Components in serial only.

I will run the GPU in series when i get my other card but for now i only have the cpu and on gpu to include in the loop.

So if i am installing 2 rads how will i make the loop if the 360 is for CPU and the 480 for the GPU and using 1 res and 1 pump?

 

[mgrisoli]

No, never run components in parallel. It reduces the flow rate to each component that's in parallel, and any test report on any component will show that a lower flow rate equals lower performance. Components in serial only.

We may agree to disagree on this one.

There are several factors involved, if you split you may have 1/2 the GPH which can still be pretty decent depending on your pump capacity, on the other hand you will be pumping cool water on both components instead of trying to refresh your GPU with hot water that just left the CPU.

There are two ways of looking at watercooling from this perspective you can have a good amount of water running slower or a smaller amount running faster both will be efficient and can be proportional in cooling capacity.

I don't mean to state which one is the best solution for this case I am just calling out that there are two options because the word "never" called me into the debate

 

[yousif666]

We may agree to disagree on this one.

There are several factors involved, if you split you may have 1/2 the GPH which can still be pretty decent depending on your pump capacity, on the other hand you will be pumping cool water on both components instead of trying to refresh your GPU with hot water that just left the CPU.

There are two ways of looking at watercooling from this perspective you can have a good amount of water running slower or a smaller amount running faster both will be efficient and can be proportional in cooling capacity.

I don't mean to state which one is the best solution for this case I am just calling out that there are two options because the word "never" called me into the debate

How will i will do the loop if i have 2 rads and dont want hot water from cpu goes to GPU since i will be adding 2 rads 1 for cpu and 1 for gpu.

 

[Supertrucker]

There is no "hot" or "cool" water in a water cooling loop with decent flow. Under a one degree difference from hottest to coldest point. You can disagree with this if you want but you will be wrong and basing your opinion on a hunch instead of science.

Parallel is also only a viable option when both lines have similar flow restriction (i.e. matching blocks). If you have dissimilar flow restriction the less restrictive block will get the majority of the flow.

 

[mgrisoli]

How will i will do the loop if i have 2 rads and dont want hot water from cpu goes to GPU since i will be adding 2 rads 1 for cpu and 1 for gpu.

Reservoir -> Pump -> Splitter -> cpu -> rad1 -> splitter - reservoir
splitter\> gpu -> rad2 /splitter>

Benefits are: you just need 1 pump and 1 reservoir, what means less tubing than the option with 2 independent loops.
For some reason the idea seems to be controversial in this forum but I like to keep my minds to different ideas.

 

[Supertrucker]

It's not controversial. It's pointless. Running multiple gpus in parallel is fine, I run my 290s in parallel. Gpu blocks are less sensitive to lower flow than modern cpu blocks. But if your gpu block is less restrictive than your cpu block the majority of your flow will go to the gpu, hurting your cpu temps far more than the half a degree or so you would get by doing serial gpu->cpu.

http://martinsliquidlab.org/2012/04/08/swiftech-mcr320-xp-radiator/4/

Pick through the data. That page has measurements for water temp going into the radiator(hottest point) and water coming out of the radiator (coldest point). With a 100w heatload the variance is less than a quarter of a degree. 200w load it's less than half a degree. 300w load it's less than 3/4ths a degree. Flip through a few other radiator tests and see the same thing.

Radiators aren't magic. Water cooling loops don't break the laws of physics or thermodynamics. If you do a parallel flow split between the cpu block and the gpu block I cannot stress enough that you will need to make sure the flow restriction for both is as close to identical as possible, because if they aren't a perfect match, you are doing more harm than good.

 

[GTXJackBauer]

to OCFs.

I'd suggest in reading up the water cooling stickies and checking out Martinsliquidlab if you want to learn the basics, differences and sciences involved.

If you don't want a caselabs, a Corsair 900D at a much less price can do the job. Has more than enough room for 120.9+ of heat surface.

If you truly stick with the GTX 980s in SLI for years to come, you wouldn't need much. Honestly, in a 900D, you'd install a 120.4 and a 120.2 (PSU side) in the bottom compartment leaving you tons of room everywhere else in the case with great temps at low noise (low RPM).

Since CPUs are more sensative to heat (coming from the GPUs) I'd have the flow going from the CPU to the GPUs. Save you a few degrees. The loop will reach a equilibrium. As already stated, there's really no cold or hot spot.

You want a dual pump setup? Sure why not if you want the added redundancy but it isn't needed because a MCP-35x or a D5, for example, will surely handle all that no problem. The sweet spot (efficiency) for the flow is between 1.0-1.5 GPM.

Two loops aren't really needed anymore in today's day in age of technology. The old days, yes. Now things have moved along to a point it is not needed ONLY if you're looking for that .5c-2c difference or feel like spending a ridiculously amount for w/e reason.

In the end, if you're patient and do the research and hard work involved in creating this build, you will have a power house of a rig with complete silence. That is the goal.

 

[RGE]

How will i will do the loop if i have 2 rads and dont want hot water from cpu goes to GPU since i will be adding 2 rads 1 for cpu and 1 for gpu.

It takes 400W to heat water by 1C at 1.5gpm. But math aside, in testing different waterblocks and setups, I have used 10 calibrated dallas 1 wire temperature probes, that in same reservoir will read within 0.1C of each other, have a pic somewhere with several showing such. And in past 10 years of watercooling, have used these probes to test water temps after different components in varying setups.

My loop a few years ago was: reservoir>>mcp355 pump>> 2 gpus oced (380W total)>>cpu oced (200W)>>360 rad>>280 rad then back to reservoir. The water temp before and after each component running following programs:

Prime 95: reservoir (temp 29.2C) >> mcp355 pump >> 2 gpus idle so only 90W total (water temp after both gpus 29.5C) >> cpu (200W) water temp leaving cpu 30.3C >> 360 rad >>280 rad and back to 29.2C in reservoir. POINT is the warmed water from gpus was 0.3C, which is roughly what math works out to be as well, so your cpu temp is 0.3C warmer by coming after gpus.

Now 3d benching/looping, I did this since it is closer to full gpu load than gaming. water in reservoir (after rads) 30.3C, after gpus (estimated 300W load) 31.0C, >> cpu 120W max (3d much less load than prime) 31.3C water after cpu, then back to rads then reservoir and again 30.3C. POINT is even max 3d load with 2 gpus, cpu temps were only warmed 0.7C. So in my loop, water to cpu or gpu first, made less than 1C difference to either component. Gaming, the difference is so low and variable, cant even measure the difference.

Now, look at your loop, if you run the water to cpu first then gpus, the max your cpu can heat the water would be ~0.8C at full prime 28.5 load (never normal use, but via 300W if overclocked to 4.6 with 1.4V running prime). So your warmer water from cpu to gpu would be 0.7C warmer, ie your water cooled gpu would max at say 46C instead of 45.2C. In normal gaming and other use, however it will be less than 0.5C warmer water after cpu. So all in serial 1 component will on average get 0.5C worse temps than the other. IF you did it mgrisoli way, and you managed to get equal flow, then both components will get about .5-1C worse temps since both get 1/2 flow. So more complicated and both temps slightly suffer because the lower flow affects components slightly more than higher temp of even second component in serial , many have tested it over 15 years, no one with any significant testing experience does it that way for a reason. But it is even worse than that, you wont get even flow, what will happen is you will get worse flow in both, but in one you may get very low flow since restriction will be unequal. Parallel is fine on just gpus, as it is a near wash with temps and flow is equal. But parallel on differing components is adding complication for worse temps all around and potentially problematic flow.......makes absolutely zero sense, mathematically or experimentally.