Overclock.net banner
Cancel
Create thread New Forums

CPU and GPU in parallel?

Jump to Latest
Status
Not open for further replies.
21 - 40 of 211 Posts
#21 ·
#23 ·
Quote:
Originally Posted by billbartuska View Post

But in two laps both cars spend the exact same amount of time between the lines because the second car gets there twice as often.
Click to expand...
Sorry I'm jumping back to this, but it makes no sense, while everything else does. If the 120 mph car goes two laps, it will spend 1 second between the lanes, 0.5 sec for each lap. The 60 mph car will spend 2 seconds between the lines, 1 sec per lap. Maybe I'm just not understanding it?
 
#24 ·
Quote:
Originally Posted by starliner View Post

Quote:
Originally Posted by billbartuska View Post

But in two laps both cars spend the exact same amount of time between the lines because the second car gets there twice as often.
Click to expand...
Sorry I'm jumping back to this, but it makes no sense, while everything else does. If the 120 mph car goes two laps, it will spend 1 second between the lanes, 0.5 sec for each lap. The 60 mph car will spend 2 seconds between the lines, 1 sec per lap. Maybe I'm just not understanding it?
Click to expand...
WOOPS! Sorry, the original post should have said

Consider a race track that is a one mile oval and has two lines painted on it 88 feet apart.
A car going 60 mph around the track spends one second between the lines each lap.
A car going 120 mph around the track spends one half second between the lines each lap.
But in two laps minutes both cars spend the exact same amount of time between the lines because the second car gets there twice as often.
 
Save
Rep+ Rep+
#26 ·
Ha! Just goes to show how many people actually read and comprehended what I wrote!

Thanks for the heads up.
 
Save
Rep+ Rep+
Discussion starter · #27 ·
Quote:
Originally Posted by billbartuska View Post

Ha! Just goes to show how many people actually read and comprehended what I wrote!

Thanks for the heads up.
Click to expand...
nah, I knew what you were going for.

I am curious as to why you have to have your rads in parallel though. If you had them run in series but Rad 1 to Rad 2 and then to pump I don't see what the problem would be.
 
#28 ·
Quote:
Originally Posted by madbrayniak View Post

nah, I knew what you were going for.

I am curious as to why you have to have your rads in parallel though. If you had them run in series but Rad 1 to Rad 2 and then to pump I don't see what the problem would be.
Click to expand...
the radiators if placed in series with each other would restrict the total flow rate of the water loop.

lets go with worst case scenario here lets say the radiators being used are restrictive radiators with a top flow rate 4.0LPM.

that would mean that with two components in parallel with each other the flow rate trough each would be only 2.0LPM.

I see this type of scenario all the time in water cooling loops, where all of the loop is in series, but the video cards are in parallel with each other.
redface.gif
 
#29 ·
Quote:
Originally Posted by toolmaker03 View Post

Quote:
Originally Posted by madbrayniak View Post

nah, I knew what you were going for.

I am curious as to why you have to have your rads in parallel though. If you had them run in series but Rad 1 to Rad 2 and then to pump I don't see what the problem would be.
Click to expand...
the radiators if placed in series with each other would restrict the total flow rate of the water loop.

lets go with worst case scenario here lets say the radiators being used are restrictive radiators with a top flow rate 4.0LPM.

that would mean that with two components in parallel with each other the flow rate trough each would be only 2.0LPM.

I see this type of scenario all the time in water cooling loops, where all of the loop is in series, but the video cards are in parallel with each other.
redface.gif
Click to expand...
The above drawing shows rads in series/parallel as an example of what can be done. It's not a recommendation to do either.

But, as I and toolmaker03 mentioned above, parallel reduces flow rate through a component.
 
Save
Rep+ Rep+
Discussion starter · #30 ·
So you would say that any sort of parallel loop is a bad idea because of flow rate.

Just want to make sure I am reading this all correctly.

But my question is how much flow rate then is "enough" to dissipate the heat? If the water is running slower through a block wouldn't it just have a higher heat transfer?

Would this risk burning up the pump sooner?
 
#31 ·
Quote:
Originally Posted by madbrayniak View Post

So you would say that any sort of parallel loop is a bad idea because of flow rate.

Just want to make sure I am reading this all correctly.

But my question is how much flow rate then is "enough" to dissipate the heat? If the water is running slower through a block wouldn't it just have a higher heat transfer?

Would this risk burning up the pump sooner?
Click to expand...
"Enough" flow rate, for a rad, is the flow rate you need with whatever fans you use at the fan speed you want and be able to remove the amount of heat you need removed. There is no "right" answer., but, generally, people recommend at least 1 gpm (3.8 lpm).

Even considering flow rate only, faster flow removes more heat because there is a larger (longer?) Delta T between the walls and the water.



Don't forget, the higher flow rate water comes back through more often too.
 
Save
Rep+ Rep+
#32 ·
Quote:
Originally Posted by madbrayniak View Post

So you would say that any sort of parallel loop is a bad idea because of flow rate.

Just want to make sure I am reading this all correctly.

But my question is how much flow rate then is "enough" to dissipate the heat? If the water is running slower through a block wouldn't it just have a higher heat transfer?

Would this risk burning up the pump sooner?
Click to expand...
well my D5 water pumps last between 7 to 10 years running 24/7. so a parallel loop does not harm the pump, if anything it would be easier on the pump, as there is less pressure in the system.
thumb.gif


as for the "enough" to dissipate heat question?

this is why I suggested making small simple test loops for the component water blocks so that you would know what the top flow rate through each of the water blocks is.

but I have tested a lot of water blocks so I know they flow from 3.0LPM to 7.0LPM depending on the water block.

a XSPC raystorm CPU water block has a flow rate of 4.6LPM
a EK supremacy CPU water block has a flow rate of 5.0LPM

a set of corsair memory water blocks have a flow rate of 7.0LPM
a universal south bridge water block has a flow rate of 3.0LPM

a swiftech GPU water block has a flow rate of 3.4LPM
a EK GPU water block has a flow rate of 4.7LPM

so I hope this helps with your understandings about component flow rates.

radiators from 20mm thick to 45mm thick will have flow rates from 4.0LPM to 6.0LPM
radiators from 48mm thick to 80mm thick will have flow rates from 7.0LPM to 10.0LPM

so enough flow rate is what the component is capable of handling.

chose the water blocks you want, test them for there flow rates, add up the total flow rates, than chose radiators that can support that total flow rate.
thumb.gif


here are the numbers, on how I calculate my potential flow rates through my parallel setup.
radiator = 5.9LPM 5.9 + 5.9 = 11.8LPM is the total flow rate possible on this parallel configuration
radiator = 5.9LPM

CPU = 4.6LPM 4.6 + 3.4 + 3.4 = 11.4LPM is the total flow rate possible on this parallel configuration
GPU = 3.4LPM
GPU = 3.4LPM

I achieved 10.5LPM on that parallel setup, so that is .9LPM less than the total flow rate possible.
the total flow rate possible is 11.4LPM or the total flow rate of the most restrictive side of the build.
wink.gif


the radiators in parallel are just a little faster than the components in parallel this is the ideal situation for this type of build.

you want the radiators side of the build to be a little less restrictive than the component side of the system, so that the components can get all the flow rates that they desire.
 
#33 ·
Water blocks just like any other component don't have 'flow rates'. They have a resistance or pressure drop relative to the flow rate they receive. For any particular component a small pump will provide a low flow rate and a larger pump will provide a higher flow rate. Doubling pump pressure will roughly increase flow rate by 50%.

A single D5 pump can produce enough pressure to push over 8LPM through a modern CPU block like the EK Supremacy.
 
#34 ·
Quote:
Originally Posted by Ashcroft View Post

Water blocks just like any other component don't have 'flow rates'. They have a resistance or pressure drop relative to the flow rate they receive. For any particular component a small pump will provide a low flow rate and a larger pump will provide a higher flow rate. Doubling pump pressure will roughly increase flow rate by 50%.

A single D5 pump can produce enough pressure to push over 8LPM through a modern CPU block like the EK Supremacy.
Click to expand...
wrong!!!! wrong!!!! wrong!!!!

nothing about this is right.

a single D5 pump will never be able to get a flow rate of 8LPM through a CPU water block.

if you would like to actually test your theory, you will see for your self that it is wrong!!! wrong!!! wrong!!

a larger pump will not increase the flow rate through any component by 50%, this will never happen.

again if you would like to test your theory you will see for yourself that it is wrong.
wink.gif
 
#35 ·





it's not like I don't know what I am talking about, as I have actually done this type of testing to know better.
biggrin.gif
 
#36 ·
Quote:
Originally Posted by toolmaker03 View Post

wrong!!!! wrong!!!! wrong!!!!

nothing about this is right.

a single D5 pump will never be able to get a flow rate of 8LPM through a CPU water block.

if you would like to actually test your theory, you will see for your self that it is wrong!!! wrong!!! wrong!!

a larger pump will not increase the flow rate through any component by 50%, this will never happen.

again if you would like to test your theory you will see for yourself that it is wrong.
wink.gif
Click to expand...
Calm down champ

Sorry but you are mistaken. Either there is something wrong with your testing gear or method.

Doubling pump pressure does indeed increase flow rate by about 50% and all the testing done by other people agrees that 8LPM is fine for a D5. There is lots of people doing it now you know.
I hope this isn't going to be another episode of the toolmaker is right and everybody else is wrong show. Remember how sure you were about how flow meters work.

You have been promoting these funny ideas about flow rates all over the place and its all a bit off.
 
#37 ·
Quote:
Originally Posted by Ashcroft View Post

Calm down champ

Sorry but you are mistaken. Either there is something wrong with your testing gear or method.

Doubling pump pressure does indeed increase flow rate by about 50% and all the testing done by other people agrees that 8LPM is fine for a D5. There is lots of people doing it now you know.
I hope this isn't going to be another episode of the toolmaker is right and everybody else is wrong show. Remember how sure you were about how flow meters work.

You have been promoting these funny ideas about flow rates all over the place and its all a bit off.
Click to expand...
and here is the difference between the two of us.

I actually tested my theory and found that I was wrong about how those flow meters worked, and showed that fact.

same here, I have actually tested a lot of different water blocks with one and two D5 water pumps attached to them. so if you would like to prove that your right. go right ahead and try, I would love to see you get a flow rate of 8.0LPM through a single EK supremacy CPU water block. in fact, I will let you use two D5 pumps, to try and get that flow rate, good luck.
thumb.gif
 
#38 ·
Quote:
Originally Posted by toolmaker03 View Post

and here is the difference between the two of us.

I actually tested my theory and found that I was wrong about how those flow meters worked, and showed that fact.

same here, I have actually tested a lot of different water blocks with one and two D5 water pumps attached to them. so if you would like to prove that your right. go right ahead and try, I would love to see you get a flow rate of 8.0LPM through a single EK supremacy CPU water block. in fact, I will let you use two D5 pumps, to try and get that flow rate, good luck.
thumb.gif
Click to expand...
Oh Dear.

If Necessary I will because despite your assumptions I have tested it, and it agreed with all the testing done by others like Martin, Stren and Fast Fate. All you need to do is look at their charts

But; here is Fast Fates EK Supreme HF test and he got 2.84GPM with 2 D5 in a Bitspower dual top that is one of the worst performing tops available. That's Nearly 11 LPM with a bad top and the restriction of the flow meter itself included.

http://www.overclock.net/t/1501978/ocn-community-water-cooling-test-thread/20#post_22587635
 
#39 ·
Even a fire hydrant wouldn't do 8 LPM.........................
 
Save
Rep+ Rep+
#40 ·
Quote:
Originally Posted by Ashcroft View Post

Oh Dear.

If Necessary I will because despite your assumptions I have tested it, and it agreed with all the testing done by others like Martin, Stren and Fast Fate. All you need to do is look at their charts

But; here is Fast Fates EK Supreme HF test and he got 2.84GPM with 2 D5 in a Bitspower dual top that is one of the worst performing tops available. That's Nearly 11 LPM with a bad top and the restriction of the flow meter itself included.

http://www.overclock.net/t/1501978/ocn-community-water-cooling-test-thread/20#post_22587635
Click to expand...
well the only pic he has of the flow rate, only shows 1.2GPM, and that matches what I have seen with this block at 5.0LPM.

so please, build us a simple loop with the EK CPU water block, and show me that it can flow 8.0LPM, with is a lot lower than 11.0LPM. so this should be a synch win for you.
thumb.gif
 
21 - 40 of 211 Posts
Status
Not open for further replies.
You have insufficient privileges to reply here.
About this Discussion
210 Replies
17 Participants
Last post:
Klue22
Overclock.net
posts
28M
members
562K
Since
2004
The premier forum for overclocking experts and enthusiasts. Discuss hardware optimization, custom builds, benchmarking, cooling solutions, and pushing the boundaries of computing performance. From beginner guides to extreme overclocking, join our technical community to master system tuning.
Full Forum Listing
Explore Our Forums
Intel CPUs Hardware News NVIDIA PC Gaming Water Cooling