Does pump speed matter?
Last response: in Overclocking
The title says it all...I have a decent amount of money set aside and I can get a uber nice, extremely fast pump (1000+ lph) or I can go with a slower pump. My intuition says faster would be better. For example, the faster air travels through a radiator, the better cooling you get, so it seems the faster water travels through a CPU/GPU block, the better cooling you would get. However, most of the guides I read say pump speed really isn't an issue, and to focus on getting a powerful enough pump to push water through all your components.
The only thing I can think of that's relevant is that higher-speed pumps would seem to impart more heat to the water. This would, of course, be bad. My setup is going to be a CPU and two xfire GPUs on an RX 240 (that's a double-thick radiator, FYI) and an RX 120 (again, double thick). The case I have doesn't allow for a 360mm rad, so I had to split them up. Regardless, I'm debating whether to run them on a single loop or a dual loop (single loop performance vs. dual loop awesome factor), but I was just wondering if pump speed really mattered in either setup.
I guess my basic question is this: why doesn't pump speed matter for cooling efficiency?
The only thing I can think of that's relevant is that higher-speed pumps would seem to impart more heat to the water. This would, of course, be bad. My setup is going to be a CPU and two xfire GPUs on an RX 240 (that's a double-thick radiator, FYI) and an RX 120 (again, double thick). The case I have doesn't allow for a 360mm rad, so I had to split them up. Regardless, I'm debating whether to run them on a single loop or a dual loop (single loop performance vs. dual loop awesome factor), but I was just wondering if pump speed really mattered in either setup.
I guess my basic question is this: why doesn't pump speed matter for cooling efficiency?
More about : pump speed matter
It doesn't until the power required to run the pump creates more heat dump and a diminished return on performance increase from the increased flowrate and heat generated. In most instances, this is very, very, very rarely an issue.
Single loops using the same components as dual loops perform better- the reason? In a single loop, every component has the ability to be cooled by all radiators in the entire loop. In dual loops, you are limited to the cooling potential of only the radiator in that loop, as well as the addition of the handful of heat watts dumped by the pump.
Depending on which GPUs you are running in X-Fire (simply from a total heat load standpoint), I'd say you are onto a good start.
Single loops using the same components as dual loops perform better- the reason? In a single loop, every component has the ability to be cooled by all radiators in the entire loop. In dual loops, you are limited to the cooling potential of only the radiator in that loop, as well as the addition of the handful of heat watts dumped by the pump.
Depending on which GPUs you are running in X-Fire (simply from a total heat load standpoint), I'd say you are onto a good start.
rubix_1011 said:
It doesn't until the power required to run the pump creates more heat dump and a diminished return on performance increase from the increased flowrate and heat generated. In most instances, this is very, very, very rarely an issue.Sorry, Rubix, I guess I'm not quite getting it...you seemed to say that pump speed did not matter, but then mentioned that at a certain point the heat from high-performance pumps could add enough heat to offset the performance gain from increased flow rate. Is there a performance gain from higher flow rate? Is it enough of a gain to matter?
I understand your point that there can be diminishing returns at a certain point, but my question was more if pump speed mattered at all. If it does, how much does it affect performance?
Yes, it does matter- you don't really want a flow rate lower than .5 gpm, but that's not to say it's not possible to have good temps that way. Most of us shoot for 1.0+ gpm.
Sorry- I thought you meant to ask if pump speed made a difference in performance based on higher speed. I guess it's still a bit relevant, but you really aren't going to run into this issue until you run a very high-end pump.
In short- you want to shoot for 1.0+ gpm and you'd be just fine. Pump speed does matter, so I'm not sure why you stated that it doesn't in your original post.
Sorry- I thought you meant to ask if pump speed made a difference in performance based on higher speed. I guess it's still a bit relevant, but you really aren't going to run into this issue until you run a very high-end pump.
In short- you want to shoot for 1.0+ gpm and you'd be just fine. Pump speed does matter, so I'm not sure why you stated that it doesn't in your original post.
Related ressources
- Does Speed Matter ? - Forum
- Air in pump when water heats up - Forum
- Which header to plug pump into? - Forum
- FPS, does it really matter ? - Forum
- Water pump too loud, how to reduce voltage? - Forum
rubix_1011 said:
Yes, it does matter- you don't really want a flow rate lower than .5 gpm, but that's not to say it's not possible to have good temps that way. Most of us shoot for 1.0+ gpm.Sorry- I thought you meant to ask if pump speed made a difference in performance based on higher speed. I guess it's still a bit relevant, but you really aren't going to run into this issue until you run a very high-end pump.
In short- you want to shoot for 1.0+ gpm and you'd be just fine. Pump speed does matter, so I'm not sure why you stated that it doesn't in your original post.
I've Just had people tell me it doesn't, is all. It seemed incorrect, and I just wanted to check to make sure. Also, people seemed to think that above a certain speed it does not matter, which I didn't quite understand. I guess I'm trying to get an appropriate sense the the scale, here...I understand that the difference between 200rpm and 400rpm on a fan is pretty small, and that the ceiling is around 2800rpm, for example. With pump speed, I don't really have the same understanding. For example,I don't quite understand why 3.0gpm doesn't result in waaaay better temperatures than 1.0gpm. 3gpm still seems significantly under the heat threshold, while being triple the speed of 1gpm. It seems to me, intuitively, that it would be similar to a difference between 900rpm and 2800rpm, which is a huge and important performance difference, but people don't really care if a pump is 1.0gpm or 3.0gpm, it seems.
Best solution
There is a difference in performance as flow rate increases, but as you get above 1 gpm, you really don't need to worry much. I understand what you are saying, and I'll try to explain.
The thermal capacity of water is so much higher than air, that with flowrates, as long as you are getting flows over 0.5 gpm, you are likely to be fine. The biggest difference in performance lies with radiator surface area, but more importantly, delta-T. This is a calculation of a flow rate typically of 1.0-1.5gpm, heat in watts produced and transferred into the loop, radiators used and their heat dissipation ability, fan types/speeds and ambient room air temperature.
Delta is affected by flow rates, but most pumps will likely get you less than 2.0gpm in a normal loop, usually 0.75-1.5gpm. You might see 3.0gpm in some very unrestricted flow tests, but these aren't typical in real world application.
One thing to consider is that the more powerful a pump is, usually the more power it demands and thus, more heat it produces. Watercooling pumps rely on the water flow to cool them, so you might see as little as 10-15 watts of heat dump, up to 30-50 watts if you have a very high end pump. As mentioned before, when you get to this point, there becomes a point of diminishing return on high flow vs. cooling ability.
Simply put- your pump is producing more heat at a higher speed to net you better temps via the higher flow, but due to the increased heat being produced, you actually are losing cooling ability for that extra 0.2-0.5gpm because of the pump heat dump offset.
The sweet spot for almost every watercooler is 1.0-1.5 gpm. Less is ok, more is ok. Getting a giant pond pump simply to have insane flow isn't what you want, if that's where this is headed.
Here is a thread that discusses pump choice based on your needs:
http://www.overclock.net/t/1108918/what-can-my-pump-handle-a-guide
The thermal capacity of water is so much higher than air, that with flowrates, as long as you are getting flows over 0.5 gpm, you are likely to be fine. The biggest difference in performance lies with radiator surface area, but more importantly, delta-T. This is a calculation of a flow rate typically of 1.0-1.5gpm, heat in watts produced and transferred into the loop, radiators used and their heat dissipation ability, fan types/speeds and ambient room air temperature.
Delta is affected by flow rates, but most pumps will likely get you less than 2.0gpm in a normal loop, usually 0.75-1.5gpm. You might see 3.0gpm in some very unrestricted flow tests, but these aren't typical in real world application.
One thing to consider is that the more powerful a pump is, usually the more power it demands and thus, more heat it produces. Watercooling pumps rely on the water flow to cool them, so you might see as little as 10-15 watts of heat dump, up to 30-50 watts if you have a very high end pump. As mentioned before, when you get to this point, there becomes a point of diminishing return on high flow vs. cooling ability.
Simply put- your pump is producing more heat at a higher speed to net you better temps via the higher flow, but due to the increased heat being produced, you actually are losing cooling ability for that extra 0.2-0.5gpm because of the pump heat dump offset.
The sweet spot for almost every watercooler is 1.0-1.5 gpm. Less is ok, more is ok. Getting a giant pond pump simply to have insane flow isn't what you want, if that's where this is headed.
Here is a thread that discusses pump choice based on your needs:
http://www.overclock.net/t/1108918/what-can-my-pump-handle-a-guide
It is pretty easy to understand that if the flow rate is too low, then heat might not be carried away fast enough. I have had people tell me that if the flow rate is too high then it would not be as efficient either, but I have not been able to understand that. For instance I am an engineer and my father is a mechanic in business for over 50 years. He said it is common knowledge that if the water flow through a radiator and car engine is too fast, it will not cool as well. I took his word for it because I could not rationalize it for myself.
It does work that way to a point, but in watercooling you don't really see it because you aren't getting high enough heat production to substantially overcome the cooling ability of the loop. It takes something like 250 watts to raise the water temp of your loop by 1C, or so. Even if you have a poor delta, you should be able to sustain a decent amount of dissipation to make up for flow. Regardless, very low flow is something you want to avoid and very high flow really doesn't exist in most water loops due to pressure drop and flow reduction through various components.
cadder said:
It is pretty easy to understand that if the flow rate is too low, then heat might not be carried away fast enough. I have had people tell me that if the flow rate is too high then it would not be as efficient either, but I have not been able to understand that. For instance I am an engineer and my father is a mechanic in business for over 50 years. He said it is common knowledge that if the water flow through a radiator and car engine is too fast, it will not cool as well. I took his word for it because I could not rationalize it for myself.Cars have thermostats. Thermostats block the flow until the temp in the engine is high enough to expand and open the thermostat. This allows the (too small) radiator enough time to do its job - cool the water. Take the thermostat out of the system and more than likely, your car will overheat.
In PC cooling systems, we don't have thermostats. As mentioned somewhere in this thread, as long as we get the hot water away from the block fast enough is all you normally need.
Related ressources:
- ForumInstalling swiftech pump
- ForumSwiftech Drive pump is LOUD, any sugestions?
- ForumPump not working properly...?
- ForumAMD in Dell - does it matter ?
- ForumRAM Speed : Does it Matter ?
- ForumPowering WC pump ?
- Forumhow do you overvolt a pump ?
- ForumDoes stock ram speed matter ?
- ForumDoes it matter how you get your CPU speed ?
- ForumDoes memory speed matter for e8000 series?
- ForumCPU fan clicking?
- ForumCPU:Memory ratio speed
- ForumDoes CPU draw volts?
- Forum[Solved] Can't overclock/overclock does not have effect
- ForumCPU Fan does not reduce RPM anymore
- More resources
Read discussions in other Overclocking categories
!
