Serial VS Parallel

[Dash Lambda]

I honestly don't understand why one would use serial over parallel for a multi-GPU config.

 

It seems like it would only place more strain on the pump and marginally decrease cooling efficiency on each successive block. Also seems like a fatal obstruction in one of the blocks, assuming it proceeds to damage hardware, would only kill one card in parallel, but the entire array in series.

 

Am I missing something?

[WereCat]

Doesn't affect cooling at all. Also pump won't care. I recommend you to watch JayzTwoCent video on this.

 

[Enderman]

pumps are way  more powerful than necessary, and it doesnt matter for temps

[Dash Lambda]

Just now, WereCat said:

Doesn't affect cooling at all. Also pump won't care. I recommend you to watch JayzTwoCent video on this.

I'll have to watch that, but I disagree with you on the first point.

 

The rate of heat transfer is proportional to the difference of the heat of the object and the environment, and the fluid necessarily gains heat as it travels through the system. In parallel, each block is at the same point in the loop, while in serial each block is further in the loop, meaning that the rate of cooling decreases for each.

-Now, I am not saying that the difference is significant. I expect that it's WELL within even the thinnest margin of error you would encounter with PC cooling. It's just that, by my understanding, there IS a difference.

 

Similar thing with the pump. I don't see how there can NOT be a difference, but I suspect it's only noticeable with particularly expansive loops.

 

Still not sure why anyone would choose series, even if it generally performs the same.

[DildorTheDecent]

Parallel looks much better if you aren't using a terminal block. 

 

For performance, both of my GTX 980s are within a degree or two of each other under load No matter the voltage (up to 1.4v). The CPU also stays nice and chilly at 60C~ with 1.4V. And up to 70C~ load with 1.5V+ (benchmarking)

 

Why anyone would go for serial is beyond me  Maybe for a single card + processor where it would be easier to replace it could help. 

[WereCat]

2 hours ago, Dash Lambda said:

I'll have to watch that, but I disagree with you on the first point.

 

The rate of heat transfer is proportional to the difference of the heat of the object and the environment, and the fluid necessarily gains heat as it travels through the system. In parallel, each block is at the same point in the loop, while in serial each block is further in the loop, meaning that the rate of cooling decreases for each.

-Now, I am not saying that the difference is significant. I expect that it's WELL within even the thinnest margin of error you would encounter with PC cooling. It's just that, by my understanding, there IS a difference.

 

Similar thing with the pump. I don't see how there can NOT be a difference, but I suspect it's only noticeable with particularly expansive loops.

 

Still not sure why anyone would choose series, even if it generally performs the same.

All right. I will agree with you there but I don't think that it makes any noticeable difference in most loops outside of particular usages like laboratory conditions.

[Dash Lambda]

6 hours ago, WereCat said:

All right. I will agree with you there but I don't think that it makes any noticeable difference in most loops outside of particular usages like laboratory conditions.

Which just means that serial will perform indistinguishably from parallel with the only differences being that a single failure will affect the entire loop, and that a particularly large loop will place slightly more strain on the pump.

 

So... I'm still very confused as to why serial exists.

[WereCat]

1 minute ago, Dash Lambda said:

Which just means that serial will perform indistinguishably from parallel with the only differences being that a single failure will affect the entire loop, and that a particularly large loop will place slightly more strain on the pump.

 

So... I'm still very confused as to why serial exists.

For lazy people

[GhostHitWall]

If you want more flow .. and your pump isn't chocking due to the pressure.. That's why you want serial over parallel.

Like you mentioned, it is odd in pc watercooling.

What I can think of is, when you have a weird pump that has very good head but poor flow which maybe can happen when you turn pump speed down for acoustic reason, and by that accepted noise level the pump act like what I assumed.

 

This probably matters for those who aren't using regular pc watercooling pumps.

 

[OddsCrazyStuff]

8 hours ago, GhostHitWall said:

If you want more flow .. and your pump isn't chocking due to the pressure.. That's why you want serial over parallel.

Like you mentioned, it is odd in pc watercooling.

What I can think of is, when you have a weird pump that has very good head but poor flow which maybe can happen when you turn pump speed down for acoustic reason, and by that accepted noise level the pump act like what I assumed.

 

This probably matters for those who aren't using regular pc watercooling pumps.

 

So would something like this pump benefit from serial more then parallel?

http://www.aquatuning.no/vannkjoling/pumper/laing-ddc/ddc-pumper/5080/laing-ddc-pumpe-12v-ddc-1t-plus?c=7751

 

As for parallel vs serial, so would I say loop order matter also, since parallel is less optimal if you have out at top and in at bottom.

[GhostHitWall]

47 minutes ago, OddsCrazyStuff said:

So would something like this pump benefit from serial more then parallel?

http://www.aquatuning.no/vannkjoling/pumper/laing-ddc/ddc-pumper/5080/laing-ddc-pumpe-12v-ddc-1t-plus?c=7751

 

As for parallel vs serial, so would I say loop order matter also, since parallel is less optimal if you have out at top and in at bottom.

welp, Martin's Liquid Lab did this long time ago. if you are interested, look into the PQ curve.
 

I don't know, it could be, it depends on where you sitting on the PQ curve.

Noise perception is subjective as well.

My guess on this is No, this would benefit from serial over parallel.

 

Loop orders does not mater in real-world scenario.  Your pressure of the loop in Serial is accumulated by each block, it doesn't matter which one comes first.

Like several posts above, the contact conductivity of thermal is minimal to make any measurable difference, so loop order really doesn't make much of difference in real application.

 

For gravity, it is pretty much irrelevant in the scale of PC watercooling.  The air/fluid pressure is significantly higher than gravity, but yes, scientifically, gravity does make little to none change to your pump.

 

I'd really like to see any source indicating parallel is less optimal when out-let at top and in-let at bottom.

or if anyone can explain it to me.

 

 

 

 

 

[Dash Lambda]

4 hours ago, OddsCrazyStuff said:

So would something like this pump benefit from serial more then parallel?

http://www.aquatuning.no/vannkjoling/pumper/laing-ddc/ddc-pumper/5080/laing-ddc-pumpe-12v-ddc-1t-plus?c=7751

 

As for parallel vs serial, so would I say loop order matter also, since parallel is less optimal if you have out at top and in at bottom.

The placement of the inlet and outlet doesn't matter.

Or, more precisely, its effect is negligible in a loop that's filled with fluid.

[Mick Naughty]

Just switched from serial to parallel. Didn't change anything. I didn't even pay attention when buying the bridge so I was happy with it regardless. 

[Curufinwe_wins]

On 5/12/2016 at 1:18 PM, Dash Lambda said:

I'll have to watch that, but I disagree with you on the first point.

 

The rate of heat transfer is proportional to the difference of the heat of the object and the environment, and the fluid necessarily gains heat as it travels through the system. In parallel, each block is at the same point in the loop, while in serial each block is further in the loop, meaning that the rate of cooling decreases for each.

-Now, I am not saying that the difference is significant. I expect that it's WELL within even the thinnest margin of error you would encounter with PC cooling. It's just that, by my understanding, there IS a difference.

 

Similar thing with the pump. I don't see how there can NOT be a difference, but I suspect it's only noticeable with particularly expansive loops.

 

Still not sure why anyone would choose series, even if it generally performs the same.

Parallel also has lower flow through each block though, so it depends on what is the limiting factor to cooling...

 

EK has a good article on this, and determining which is better is a non-trivial task esp when you need to standardize by either pump speed or fluid velocity...