How much WC is enough???

[Clovett]

First stats on my system, your mileage may vary.

Coolance all copper core and fin dual 140 double row radiator
Swiftech Apogee HD using a secondary bypass for the GTX 760
Swiftech MCW82 universal GPU block
Koolance D5 with tach attached to CPU fan header
4770K processor
EVGA OC GTX 760 video card
ambient temp between 21C and 22C

Running four 120mm intake fans and two 120mm and two 140mm exhaust fans. All fans running on low with my total sound so minimal I can hear my hard drives seek.

Under the stock fan I got these temps

CPU at 3.5Ghz idle at 30C and after about 5 minutes of Prime95 around 83C
GTX 760 idle at about 35C and under load after about 20 min about 57C

With my watercooling setup

CPU at 4.1Ghz idle at 27C (all power saving disabled) and after about 5 minutes of Prime95 torture around 57C

GTX 760 idle at about 23C and under load after about 20 min about 43C
After 3 1/2 hours of playing Ghosts on high video settings the 760 idled at 26C

Why is the GTX 760 idle speed important? Because I'm running the card off a secondary port from the Apogee HD. The video card temperature at idle is a great indicator of how warm the water is in my radiator. Ghosts gives a real workout for both the processor and the video card. It's more real world usage than anything else I have.

This setup can easily cool another GTX 760 in SLI mode. So why is everyone recommending huge radiators and even dual radiator setups for real world usage?

 

[yamaharacer19]

You already met the thermal limits of the radiator with the GPU alone. If you add another 760 to the loop, you might as well take out your processor and throw it into a fire.

We recommend at minimum 120.2 for gpu and 120.1 for cpu. If you add another GPU then add another 120.1. That'll keep you in the safe zone.

 

[Clovett]

You already met the thermal limits of the radiator with the GPU alone. If you add another 760 to the loop, you might as well take out your processor and throw it into a fire.

We recommend at minimum 120.2 for gpu and 120.1 for cpu. If you add another GPU then add another 120.1. That'll keep you in the safe zone.

I am sorry, but the temperatures I am seeing are not equaling what you are saying. You did see the temps I posted? If I am at the limit of this setup, which is what everyone seems to think, why are my radiator temps only raising a few degrees C after hours of constant gaming with minimal fans?

Plus this being a double row radiator and I am using push/pull with fans wouldn't that be the equivalent of a single row 140.4?

 

[yamaharacer19]

Push pull will only net you a maximum 20% temp decrease. Where as having 140.4 rad will roughly be 2-4 times better than your push/pull setup because of more surface area to disperse heat.

 

[ATMINSIDE]

CPU is getting too hot for 4.1GHz. Definitely needs more radiator.

 

[Clovett]

Push pull will only net you a maximum 20% temp decrease. Where as having 140.4 rad will roughly be 2-4 times better than your push/pull setup because of more surface area to disperse heat.

So you are saying that a xual row 140.2 is the same as a single row 140.2 even though it has twice, or close to twice, the fin surface area?

 

[ATMINSIDE]

So you are saying that a xual row 140.2 is the same as a single row 140.2 even though it has twice, or close to twice, the fin surface area?

In that case, what fans are on the radiator?

I bet you're not getting enough air through it at all.

 

[Waza]

5 minutes prime ain't going to tell you if your loop can handle the heat. run Furmark and Prime for at least 20 minutes for any kind of trustworthy result

 

[yamaharacer19]

So you are saying that a xual row 140.2 is the same as a single row 140.2 even though it has twice, or close to twice, the fin surface area?

No I'm not. a single row 140.2 would have the inlet port on one side of the rad and the outlet port on the opposite side. When did you build this loop? If you just recently built it I would highly recommend getting another 140.1 rad so you'd have enough surface area to disperse the heat.

 

[m0r7if3r]

OP, have you calculated your heatload? If so, what have you calculated.

 

[ATMINSIDE]

OP, have you calculated your heatload? If so, what have you calculated.

TDP is 84+170 at STOCK. So, 254W, plus OC going to be more like 300W. Maybe more.

 

[m0r7if3r]

TDP is 84+170 at STOCK. So, 254W, plus OC going to be more like 300W. Maybe more.

While I appreciate the help, I can pull the numbers and do the math - it's relatively trivial for me. I'm trying to determine how much research the OP has done and how much OP understands the mechanics of watercooling. Based on how this thread is going, I believe that OP does not have the knowledge that OP should to make well informed decisions about the loop.

 

[ATMINSIDE]

While I appreciate the help, I can pull the numbers and do the math - it's relatively trivial for me. I'm trying to determine how much research the OP has done and how much OP understands the mechanics of watercooling. Based on how this thread is going, I believe that OP does not have the knowledge that OP should to make well informed decisions about the loop.

I believe that also. Just trying to help the OP.

 

[Clovett]

CPU is getting too hot for 4.1GHz. Definitely needs more radiator.

My temps are well below what some people experience on air cooling at stock 3.5Ghz and I am running 4.1Ghz. I don't see the issue.

In that case, what fans are on the radiator?
I bet you're not getting enough air through it at all.

I am running two 120mm fans and two 140mm fans in a push pull config for the radiator. I run them at ultra low speeds and prefer it that way for the minimal sound factor.

5 minutes prime ain't going to tell you if your loop can handle the heat. run Furmark and Prime for at least 20 minutes for any kind of trustworthy result

Result for what exactly? I run Prime95 torture test to see if the system will take being maxed out for 5 minutes. Running the test for 20 minutes is irrelevant when compared to real world usage.

No I'm not. a single row 140.2 would have the inlet port on one side of the rad and the outlet port on the opposite side. When did you build this loop? If you just recently built it I would highly recommend getting another 140.1 rad so you'd have enough surface area to disperse the heat.

I have dual inlets and dual outlets for this radiator. I use one inlet for the CPU and one for the GPU.

If I were thinking of bitcoin mining or something similar I would agree. But there are no real world instances where myself or any other user is going to push both video and cpu in a constant that will heat the PC to that degree.

OP, have you calculated your heatload? If so, what have you calculated.

It's actually impossible to calculate unless you are building the system to specifically run both your CPU and GPU constantly for hours on end. So yes you can calculate it for a program that maxes out your CPU and GPU for testing purposes. That's the point I am trying to make here. The fact is that your CPU and GPU are never loaded together in real world useage. It's always variable. I use my system for work which barely taxes it. I use it for gaming and that pushes it but with my CPU maxing ag 57C that means it's running well below that. And my GPU hits like 47C tops mostly running well under that. In real world usage I still have overhead to spare but if I maxed it for hours on end I don't, but there are no situations in my usage, nor the majority of all users, that would do that.

That said running a benchmarking program that maxed out my processor at stock speed with the GTX 760 maxed for an extended period would probably exceed the capacity of my radiator. I don't plan on doing that. The CPU would pull about 160 watts and the GPU would pull about 170 for a total of 330 watts.

 

[Supertrucker]

At 4.5ghz 1.24v my 4670k is 60c over water temp inside of 20 seconds of launching prime95 small fft, and it will climb a few degrees from there as the water gets heat saturated.

You are running a very mild overclock, and you dont load your cpu and gpu at the same time, which is fine, but it isnt uncommon for people to max out both the cpu and gpu at the same time. Just because real world usage to you is a mild overclock and cod ghosts doesnt mean that is my real world usage, or anyone elses. There is an entire subforum dedicated to cryptocurrency, which by its very nature fully maxes out the cpu and gpu depending on the currency you are mining.

Above and beyond that, the smarter people on this forum dont recommend an amount of radiator without knowing usage and heat load. Myself, for a computer on the quieter side, i recommend 120.1 per 100 watts, which your radiator is pretty close to since a 140.2 is almost the same surface area as a 120.3, and your heat load is pretty close to 300 watts at full usage.

With your usage pattern and overclock you could easily run on a 120.1 until the day you tried to encode a movie while playing a game, or anything of that nature, and then your cpu would overheat and throttle.

 

[Clovett]

At 4.5ghz 1.24v my 4670k is 60c over water temp inside of 20 seconds of launching prime95 small fft, and it will climb a few degrees from there as the water gets heat saturated.

Exactly! But mine doesn't. My water temp raises 5C after 5 minutes of running prime95 on torture with 8 instances. My radiator and the 250ml resevoir are both located outside my case, could that affect it? Also I was incorrect when I thought this radiator had 2 rows of pipe it's actually three rows of pipe all copper with all copper fins 60mm thick and the pipe is 13mm ID and 2mm thick walls. 7FPI fins

You are running a very mild overclock, and you dont load your cpu and gpu at the same time, which is fine, but it isnt uncommon for people to max out both the cpu and gpu at the same time. Just because real world usage to you is a mild overclock and cod ghosts doesnt mean that is my real world usage, or anyone elses. There is an entire subforum dedicated to cryptocurrency, which by its very nature fully maxes out the cpu and gpu depending on the currency you are mining.

I addressed this in a previous post on this thread that the majority of people don't mine for currency and would not have this type of usage and that those that do would probably need more cooling.

Above and beyond that, the smarter people on this forum dont recommend an amount of radiator without knowing usage and heat load. Myself, for a computer on the quieter side, i recommend 120.1 per 100 watts, which your radiator is pretty close to since a 140.2 is almost the same surface area as a 120.3, and your heat load is pretty close to 300 watts at full usage.

Now that is where we disagree. The more I have thought about it the more I question what is being thrown about as "heat load". I based the wattage for the GTX760 on competent reviews that it pulls about 170 watts when it's maxed. But I thought about it and the CARD is pulling 170 watts with the VRM disipating quite a bit of that load. I'm not using a full coverage block and am only cooling the chip itself. With that said whatever the temp of the chip itself is I am also not pulling that either. Same goes for the CPU. If I leave 57C on the chip I am only disipating part of that load. I'm not disipating the entire load for either case.

 

[Waza]

Warming up fluid in loop isn't exactly as simple as 5 minutes of stress, If your radiator handles say 95% of the heat with high delta it will slowly start increase in temps. Could be that it takes an hour or two even to reach maximum temps with varying load. Running the prime on the processor for 5 min and not stressing the gpu isn't very real world. Ambient temps can be higher and you might play some other game say crysis3 or bf4 that really use the processor much better that badly ported CoD.If you ever decide to multitask more heavily you will run into problems.

The recommendation is generally one that guarantees that you won't run into problems and the one recommending doesn't necessarily know exactly how much stress your system will endure.
If you are mindful and remember that using your PCs whole potential will make it crash and burn, then sure run it with any kind of cooling that suits you.
Just remember that what applies for your system might be different for someone else.

 

[Clovett]

Warming up fluid in loop isn't exactly as simple as 5 minutes of stress, If your radiator handles say 95% of the heat with high delta it will slowly start increase in temps. Could be that it takes an hour or two even to reach maximum temps with varying load. Running the prime on the processor for 5 min and not stressing the gpu isn't very real world. Ambient temps can be higher and you might play some other game say crysis3 or bf4 that really use the processor much better that badly ported CoD.If you ever decide to multitask more heavily you will run into problems.

The recommendation is generally one that guarantees that you won't run into problems and the one recommending doesn't necessarily know exactly how much stress your system will endure.
If you are mindful and remember that using your PCs whole potential will make it crash and burn, then sure run it with any kind of cooling that suits you.
Just remember that what applies for your system might be different for someone else.

I agree for the most part and thanks for a reasonable reply. However the point I am trying to get people to understand is I can use my PC's whole potential for quite a bit of the time under my current water cooling, I just can't do it constantly for hours on end which I would never do. There are very few instances that someone would unless they dedicate to something like coin mining. A fair estimate would be that this represents well over 95% of all PC usage.

So why are we trying to scare people that are looking to water cool their PC's away from this forum? To my knowledge there is no sticky that explains any of this in a form people can understand. It's more like if your system can't run balls to the wall for 24 hours it's not ready for prime time and that is about as far from reality as it gets.

If I bought an off the shelf system with a decent video card and a 4770 I7 running on air and ran a benchmark program that maxes both CPU and GPU how many hours do you think it would run before it crashed and burned? And that system is quite suitable for the above average user for years.

 

[Waza]

If I bought an off the shelf system with a decent video card and a 4770 I7 running on air and ran a benchmark program that maxes both CPU and GPU how many hours do you think it would run before it crashed and burned? And that system is quite suitable for the above average user for years.

Assuming that air cooled system has decent circulation and case design that system could run fine for at least 3 years and more even without maintenance.
But if you would decide to overclock or add one more gfx card and the case Isn't designed to handle that you would run into overheating issues or at least shorten the lifespan of critical components.
The idea of water cooling is to either increase your overclock headroom or go for virtually silent operation or both. As a bonus you probably get a good year or two more out of your water cooled components since their operating temperature is lowered typically 10-20c or even more.
Building air cooled system is fine but usually is louder than liquid cooled system. Almost all aftermarket cooling components are aimed for overclockers and to those who prefer silent operation.
I myself like to run my processor under 55c load and as quietly as possible, at the same time I like to run benchmarks and push the system just for the kicks of it. It's just my preference.

If people would give a better idea of what they were after then the recommendations could be more suited. Most just want maximum performance and that's usually why recommendations are what they are.
For first timers that look for that little performance boost for games over air cooling solution there are pre optimized WC kits available that will do their job really good.

 

[GTXJackBauer]

So why are we trying to scare people that are looking to water cool their PC's away from this forum?

Please explain. Giving advise may differ a bit from others but generally meet at the same junction.

To my knowledge there is no sticky that explains any of this in a form people can understand.

There's a lot of great information and links to sites on our watercooling stickies. To state that comment is misinformed because I push new comers to read those stickies all the time. This is a hobby and watercooling eletronic components is serious business. Its not riding a pony or anything. I want people to take it serious but have fun with it when all is said and done.