SOLVED Parallel or Serial

[Hardass]

Going to be adding a second EK-FC CSQ Block to a 7970 I need a connector for them and not sure which to get. Can anyone explain the difference between the Parallel and Serial Connectors?
Parallel:
http://www.performance-pcs.com/catalog/index.php?main_page=product_info&products_id=34577
Serial:
http://www.performance-pcs.com/catalog/index.php?main_page=product_info&products_id=35695

 

[MattMan657]

I actually asked this very question a couple of weeks ago.

If I'm wrong, someone please correct me. Parallel is when you have both ports on the GPU blocks connected to each other which increases flow rate but reduces overall pressure.

Serial is when you only have one port connected to each other, not as high flow rate, but higher pressure.

From what I gathered, it depends on the build. How restrictive is your loop, etc. But it sounded like it doesn't make a huge difference at the end of the day.

 

[Hardass]

Thanks, think I will go with Serial.

Reply from another forum.
In a parallel connection, there is no defined water flow direction in the waterblocks.
Obviously water has to go in somewhere and come out somewhere, but once inside the block it can flow or disperse randomly, returning or flowing out of each channel opening in the block.

In serial the water flow has a defined in and out and pathway through the block channel.

If you click through the pictures supplied in the Performance link, you'll see the water pattern nicely in the included diagrams.

I've always used a serial scheme in my SLi or Crossfire watercooling.

 

[funsoul]

Go parallel. It provides more even heat dissipation.

 

[Hardass]

Go parallel. It provides more even heat dissipation.

Really?

 

[funsoul]

yes...for example...in a quad setup...
- Parallel (actually, unless you have a specific bridge, this is actually called semi-parallel)
Water goes in to card 1 and flows through card1's block to card2 in serial (so card2 will always be warmer than card1) but flows parallel (not through card1's block) to card3 (so card3 will be same temp as card1). The water flows in serial from card3 to card4 (so, like above, card4 will always be marginally warmer than card4). Expect cards1 and 3 to be a tad cooler than 2 and 4.

- Serial
The water flows from 1 to 2 to 3 to 4 so card2 will be a little warmer than 1, 3 will be warmer than 2 and 4 will be warmer than 3.

 

[MattMan657]

Does it make much of a difference in a two card setup?

 

[funsoul]

Only marginally but if you're pushing your system (and, in general) parallel is the preferred method (imho).

 

[funsoul]

Going to be adding a second EK-FC CSQ Block to a 7970 I need a connector for them and not sure which to get. Can anyone explain the difference between the Parallel and Serial Connectors?
Parallel:
http://www.performance-pcs.com/catalog/index.php?main_page=product_info&products_id=34577
Serial:
http://www.performance-pcs.com/catalog/index.php?main_page=product_info&products_id=35695

PS- Get the first one...same as one of the ones I use. If you're ever thinking of using more cards that have waterblocks using the same connectors, just get a 4-way...about the same price and you just plug the holes you're not using (the bridge comes with enough iirc). Eliminates having to buy additional bridges.

 

[Hardass]

Thanks for all the help. Cannot order until I found out the spacing of pcie slots on the X79-UD3. Will be adding a third card so may just get the connector for 3 cards and block off the 3rd until needed.

 

[Tatsugen]

I was just recently looking up information for this setup as well. Found a few posts that goes into the thermal dynamics behind WC (math), this one, make sure you read though the whole thing though, the first post had his math wrong. And then, this one, post 11 explains it a little more.

It comes down to personal preference. In theory, parallel should do better, and it actually does cool the cards more evenly. The only thing I can think why series would be better, is if something you are running only utilizes 1 card, (and you have that card before the other one) it cools it better than it would in parallel.

I'm still new at this stuff, so, if I said anything wrong, someone correct me please.

 

[S_I_N]

done both methods with my 7970's and neither ran cooler than the other I'd say go serial and be done with it.

 

[Hardass]

Thanks guys, going Parallel order the hardware.

 

[RGE]

Really doesnt matter. The temp difference from 2 cards in serial is about 0.5C, takes 400W to heat water 1C at 1.5gpm, so for 200W cooling it is 0.5C high water temp to GPU2 than GPU1. Parallel, though you get same water temp, it is at 1/2 flow so each card will probably see about 0.3 to 0.5C higher temps. When testing done with calibrated probes, typical outcome is
Serial gpu 1 temp 46.5, gpu 2 temp 47.
parallel gpu 1 temp 47, gpu temp 2 47.

So basically discussing 0.5C regarding gpu. Serial cpu probably 0.3 to 0.5C higher from lower overall flow.

Bottom line, personal preference, do whichever is easier or more aesthetically pleasing, because you wont be able to measure, let alone see any temp differences. Unless you want to spend $300 on calibrated probes to detect the less than 0.5C difference.

If you have 4 gpus with full cover blocks, parallel is better, not for gpu temps, but for cpu temps because 4 full cover gpus might decrease flow down in range where you might get 2C higher cpu temps.

 

[Clovett]

Bottom line, personal preference, do whichever is easier or more aesthetically pleasing, because you wont be able to measure, let alone see any temp differences. Unless you want to spend $300 on calibrated probes to detect the less than 0.5C difference.

If you have 4 gpus with full cover blocks, parallel is better, not for gpu temps, but for cpu temps because 4 full cover gpus might decrease flow down in range where you might get 2C higher cpu temps.

I just used a port off my Apogee HD for the parallel, seemed easy enough and I think it helps my flow. I don't have any fancy $300 calibrated probes, but I do have an infrared heat gun that is a big help on tracking down issues and checking to see what is really being cooled and by how much. And before everyone rants about how they aren't as accurate as the calibrated probes realize 1 degree doesn't matter and that all readings will be off the same. Just take them from the same distance.