New MC Bench up and running, testing to begin soon [Archive] - RRTech...The SOURCE for Extreme Liquid Cooling

mcoffey

01-25-2009, 07:44 PM

Thought you all might like to see this.

This rig has and will only have two purposes in life. Crunching WCG WU's, and CPU Block thermal testing and development. Once it set, everything gets locked down. No changes to OS, SW, Rad, pump, pump top, fans. No added chipset block, GPU cards blocks or anything else unless something breaks. This is all to facilitate and allow long term repeatibility for CPU block performance comparisons. If I decide to test i7 blocks, everything for that set up will be locked down and remain the same. I'll simply swap the boards with everything on them setup already.

Just got the new MC bench up and running the other night. Still have to get the CF temp measuring system set up and plugged in.

MB: Asus P5E X38.
CPU: Q6600 G0 @ 3.6
RAM: Patriot DDR-2 PC-1066 @ 800/CAS 5
PSU: Silverstone Strider 1000 watt
OS: XP Pro 64 bit with all updates installed at the time of the bench setup, then locked down. No other updates allowed.
Rad: PA 120.3 with shroud
Fans: Zalman ZM-F3's in push config.
Pump: Laing DDC-3.2 with EK top, rev 1.
Heat Loading SW: Prime 95/64 bit version/Build 4-v25.8
CPU Temp monitoring and logging SW: CoreTemp v99.4
Rig Temp monitoring HW: CrystalFontz 633 RS-232 head unit with Dallas 1 wire DS18B20 probes. 1 probe rad fluid out. 1 probe block fluid out (optional). 1 probe x3, one for each fan inlet ambeint air temp. 1 probe for rad air out temp. 1 probe room ambient air temp (optional).
Rig Temp monitoring and logging S/W: WinTest_633 logging to external Laptop with time sync.

That about covers it for now, here's some shots. It already crunching @ 3.6 and solid as a rock 24/7. I'll have it all setup nice and clean by the end of this week.

http://lh3.ggpht.com/_BTn9YHMni9Y/SX0NJIfp6HI/AAAAAAAAC5o/njkZMX-VZwY/s800/DSC_0047.JPG

http://lh3.ggpht.com/_BTn9YHMni9Y/SX0NLGCxJZI/AAAAAAAAC5w/j-T9PDLEelQ/s800/DSC_0048.JPG

http://lh5.ggpht.com/_BTn9YHMni9Y/SX0NMxRg7-I/AAAAAAAAC54/9aDgE_4zk8o/s800/DSC_0049.JPG

http://lh4.ggpht.com/_BTn9YHMni9Y/SX0NOn1chAI/AAAAAAAAC6E/Moq7m8qRpik/s800/DSC_0051.JPG

More pics along with in depth CPU Block thermal performance testing methodology to come.

Enjoy,

andyc

zlojack

01-25-2009, 07:54 PM

Nice one mcoffey!

I'm digging the setup :D

Septim

01-25-2009, 10:59 PM

looks good... me think some nice alum fan holder add-ons should round it off nicely...

Kayin

01-26-2009, 06:09 AM

Great setup, but why go with an outdated OS?

At any rate, I'm very interested in seeing this in action. I'll be building something similar soon... Mine just won't fire up. I'm doing one with aquarium heaters and a die simulator. I find it'll be easier than swapping stuff in and out of Mithril.

HESmelaugh

01-26-2009, 07:04 AM

Nice! What blocks will you be testing?

RedRaider

01-26-2009, 07:33 AM

Looks great,Andt...:up:

mcoffey

01-26-2009, 09:10 AM

Thanks everyone,

I'm hoping this thread will encourage others to set up benches for testing. It's not the black magic a lot of people make it out to be. Also, as far as efficency and high performance goes, nothing beats temps on a bench in open air:D For a few hundred, anyone can set one up and get test results as good as Martin or anyone else for that matter if the focus is on CPU block thermal performance.

Because for me, that's where it's at. I don't really care about flow rate and pressure drop unless it directly equates to keeping the CPU cooler. Just that simple for me. You'd be amazed at how much you can learrn about subjects such as flow rates and their direct effect on cooling the CPU by setting up a simple bench with the correct measurement instruments and methodology. After it's setup right, it's more repetitive than anything else.

As far as what blocks I'll test, the usual suspects. EK, D-TEK, AC and few others. I'm really more interested in showing people how it's done more than anything else, I already know which blocks perform better, I've been doing this for a while now under the radar:D

Great setup, but why go with an outdated OS?

At any rate, I'm very interested in seeing this in action. I'll be building something similar soon... Mine just won't fire up. I'm doing one with aquarium heaters and a die simulator. I find it'll be easier than swapping stuff in and out of Mithril.

Thanks,

Becuase that OS is the most controlable as far as a service paging the system and inturpting the loading SW.:up: That's what your after, more than how a CPU will act on any given OS. Becuase as far as thermal performance, doesn't really matter. You just want stability and repeatibility more than anything else. You'd be amazed at how much works been done on the subject and how much it took to nail that down:D I'll put it to you this way, good thing I' more into learning and sharing than I am getting credit for something:D

andyc

CrazyJoe

01-26-2009, 08:50 PM

... as far as thermal performance, doesn't really matter. You just want stability and repeatibility more than anything else.

YES !

Looks great ! Hopefully I'll get mine built up soon enough so I can post some test-bed goodness as well.

Again, Great job ! :up:

mcoffey

01-27-2009, 08:55 AM

Yeppers,

Just ordered the last few parts for the CF unit, and they should be here by end of the week, then a testing we will go:D

I love this P5E board, what a great little clocker this thing is and solid as a rock:up:

andyc

Baleful

01-27-2009, 09:17 AM

Looking good man. I can't wait until I'm able to do this kind of stuff.

mcoffey

01-27-2009, 09:30 AM

Yeah man,

One of the most interesting parts of the hobby, and you really gain the ability to cut thru the bullshit and find what really works. There's just so much mis-information out there right now being drug in from the past.

First little test I'll run is the effect, or lack of effect restriction plays in CPU Core Temps. Specifically, BP rotary 90 fittings:D Then on to some other suspected myths like EK Supreme blocks only good for CPU loops only, etc etc.

That's were the fun is, discovery.

andyc

smee

01-27-2009, 09:41 AM

http://lh4.ggpht.com/_BTn9YHMni9Y/SX0NOn1chAI/AAAAAAAAC6E/Moq7m8qRpik/s800/DSC_0051.JPG

Hey andy, setup looks great!

Was wonderin, in that pic is that a rubber mounting "arm" thing around the res?

I have one of the nice black mounts too, but my "arm" thing that clips around is plastic..... no point in having it really....... but I'd love a rubber one (if they make them).

mcoffey

01-27-2009, 10:16 AM

Hey andy, setup looks great!

Was wonderin, in that pic is that a rubber mounting "arm" thing around the res?

I have one of the nice black mounts too, but my "arm" thing that clips around is plastic..... no point in having it really....... but I'd love a rubber one (if they make them).

That's actually the original EK res , and mount that came with the res when purchased back years ago. All hard plastic.

andyc

smee

01-27-2009, 10:27 AM

That's actually the original EK res , and mount that came with the res when purchased back years ago. All hard plastic.

andyc

ahhh, ok it looks like rubber almost!

I wish they made em in rubber though.... it'd be so much better. :/
Anyways, thanks!

Nickel020

02-12-2009, 06:10 PM

What fluid probe are you using? Do the probes need to be calibrated?

I'm also getting a new bench table (my old ones are just a metal box with motherboard standoofs, not suited for watercooling). The LeChuck table that HESmelaugh also uses is only 100€ and offers basicalle the same features as that MC.

mcoffey

02-12-2009, 06:21 PM

Modified Dallas/Matrix probes for fluid, and no there pretty much set right out of the box.

andyc

BiNGE

02-12-2009, 06:30 PM

:D looks like we have the same techbench. Did you have to mod the back a little to fit the rad? I can't wait to see what you're up for benching :D Keep us posted.

Snyxxx

02-12-2009, 06:31 PM

Damn, Andy, I had forgotten about getting one of those MC benches.

Now it is back in my head. :D

Thanks for posting the pics and I cannot wait to see your consistent results.

Nickel020

02-12-2009, 06:32 PM

Can you show me how you modded the probes? I'm not very happy with the standard probes that connect to my Aquaero (different probes give somwhat different values) and I'm therefore looking at getting a Crystalfontz as well.

Thanks!

Hondacity

02-12-2009, 10:06 PM

Snyxxx

02-12-2009, 10:20 PM

mcoffey

02-13-2009, 12:41 AM

Here you go:

http://www.realredraider.com/vbulletin/showthread.php?t=169&highlight=dallas

Yep,

That's pretty much the way I've done it for a while now. The main thing is, just make sure the epoxy covers up to where the metal part of the probe sets on the plastic housing. AS long as you do that, you can apply the epoxy any way you want. This is however the most pro looking one and application I've seen. He really did a super job.

:D looks like we have the same techbench. Did you have to mod the back a little to fit the rad? I can't wait to see what you're up for benching :D Keep us posted.

Nope,

Just attached the rad using the 4 middle holes to the outside of the bracket, then shroud, then fans, then filters.

I'm still snapping it in right now and doing validation tests for repeatability. I'm just about there. But that's another thing about good testing results. You have to learn and find out all the ends and outs of your test setup. Each setup tends to act different with it's own little quirks. The MC bench is really great though, and I set it up so I can pick it up and move it around.

andyc

Bigtime

02-13-2009, 01:11 AM

That looks sweet! Always wanted to do something like that, just never had the patience:rolleyes:

Nickel020

02-13-2009, 04:39 AM

Thanks for the link and the advice, that was exactly what I was looking for (and more than I expected)!
Do you know if the Crystallfontz display be usable with a flowmeter, like the ones you can use with an Aquaero?

mcoffey

02-13-2009, 08:44 AM

Only if they can do somthing like read and/or convert Fan RPM. Other than that, not really made for that.

andyc

mcoffey

02-13-2009, 02:58 PM

Well,

Here's an interesting little tid bit. I didn't run into it before because the last round of testing was all done within 1 c ambient air temp for all blocks and mounts. But now since I'm snapping the new bench in for validation, I ran another type test. I left the same block and mount on, recorded temps, then raised the ambeint rad air in temp by 7c, and did another run, with everything exactly the same.

Now you would expect to subtract the difference in temps from the core temps and be about the same average core temps for the two tests, but it wasn't. I was about 1.2c higher on average core temps than I was on the first run. That's throwing me for a loop somewhat, so I need to get to the bottom of it before I start testing. I'm pretty sure I know the cause is, but I have to find out exactly what's going on, and a way to account for it when comparing temps at different times during a typical run.

I know it's not the probes, they are linear, but the difference in average core temps aren't They don't scale 1 for 1 as the ambeint temps rises.

AT least that's my intial findings, I'll post some data points as I isolate it, and can repeat the anomaly.

andyc

Snyxxx

02-13-2009, 03:13 PM

Andy,

That is very interesting. Now that you mention it, it does not surprise me the ambient air versus core temps are not directly proportional. I would think the cooling capacity of a radiator is non-linear with ambient temps.

mcoffey

02-13-2009, 03:43 PM

Yeah,

What I think it really brings into question is testing validity for previous results if not accounted for. But like I said, have to nail it down first to make sure it's consistant.

So much fun testing stuff and discovering. It's amazing how sensitive these Dallas probes are. You can blow on one and see the temps jump for a second or two. Move your hand within 2 inches of one and the temp will rise.

Amazing considering a 633 serial set up with a few probes can be had for less than 70 USD. And one hell of a fan controller too.

andyc

Snyxxx

02-13-2009, 04:02 PM

Yes, it seems there is no one package that has it all. I have:

bigNG - Great hardware - lousy software support - uses "ordinary" temperture probes.

Aquaero - Great hardware - great software - uses "ordinary" temperture probes.

I do not have the crysatfontz package, but wow are these temperature probes accurate in comparison to the others. Too bad you cannot use the Dallas probes with the above two packages.

I am fine with the ordinary temperature probes like the Bitspower ones, but they are not that accurate. I have four of them and they all show a different temperature when side by side up to 2°C in difference. But I basically offset them in software to get them all the same, so it works. I just hate knowing something better is out there I cannot use.

mcoffey

02-13-2009, 04:08 PM

Yeep,

The best thing about the Dallas probes for testing is the sensitivity or resolution @.0625c. That's what makes them so well suited.

andyc

mcoffey

02-13-2009, 11:38 PM

Well,

Here's the first pass, and everything seems to be working well. I just need to do one more thing, add a reversed shroud to act as a air box and filter. Then I think we're all set.

Here's some shots of the data sheet setup I use to plot the temps and such. Basicaly data from the Core Temp sheet, and the 633 .cvs log file are used to input info in this sheet. Then a sheet for each test is archived with the average temps of all runs being logged to a master testing sheet for comparison.

Front page with data points

http://lh6.ggpht.com/_BTn9YHMni9Y/SZZTG-Gon8I/AAAAAAAAC-k/-u7aKokjYsc/s800/AC%20DI%20v1.xls%202142009%20121343%20AM.jpg

First graph of all data as a spot check to make sure nothing went south during the run

http://lh4.ggpht.com/_BTn9YHMni9Y/SZZTEW4EZUI/AAAAAAAAC-E/3ZtGxfkJhPc/s800/AC%20DI%20v1.xls%202142009%20121252%20AM.jpg

Shot of the Core Temp only graph. Notice how the 2 sets of cores rarely work independently of each other, which is one of the thing I really dislike about using 65mn chips for testing. 45nm work much more indivigually.

http://lh4.ggpht.com/_BTn9YHMni9Y/SZZTE4tGi8I/AAAAAAAAC-M/jFsmxo7-hYU/s800/AC%20DI%20v1.xls%202142009%20121301%20AM.jpg

Very usefull graph for spot checking fan intake temps. Notice how the left fan intake spiked towards the end. Probably the furnace kicked on. Just goes to show how you can't use some probe stuck in the room somewhere when trying to test. Room air temp, and fan intake temps are never the same, or even close most of the times.

http://lh6.ggpht.com/_BTn9YHMni9Y/SZZTFX1mLuI/AAAAAAAAC-U/Px4q40D-Uew/s800/AC%20DI%20v1.xls%202142009%20121310%20AM.jpg

This is a very useful graph too. It compares average core temps with average rad air intake. Notice how steady the bottom plot and top plot match up. See how the top plot dips down every once and a while? That's just reflecting the core temp average for cores 0-1 moving from 57.0 c to 56.5 c. Now look towards the end, see the bottom line move up slightly? Now look at the top line again towards the end, see how it stops dipping down, and the couple of peaks going up. That was the furnance kicking on as seen in the other graph when the left fan spiked. The setup doesn't miss a thing when the probes have a sensitivity of .0625c:D

http://lh6.ggpht.com/_BTn9YHMni9Y/SZZTGE8LDVI/AAAAAAAAC-c/l5hUkh5AayU/s800/AC%20DI%20v1.xls%202142009%20121319%20AM.jpg

More to come later.

andyc

mcoffey

02-14-2009, 10:21 AM

And here's why you want to chart everything on a data sheet to check for validity.

This was a pass I did when I first got up this morning, and one of the problems when using software to load the CPU for testing. This whole run is invalid. Something was paging the system and disrupting cores 2-3 from loading 100% via Prime 95. That in turn causes the temps of the cores to go down more than they should and skew the data. It's normal to see cores drop 1c off and on as they trend up or down, but they should never drop 2c. In this case it's happened way to many times, and will skew the core temps average to rad air in average. Also, the paging kept cores 2-3 from trending up with core 0-1 towards the end.

In this case, I didn't reboot to clear everything out before I did the run, and some service paged the system while I was running the test.

Like I said before, there's a lot more to good testing than the gear you use. A lot of what people consider to be mount variation turns out to be other things. This would be a good example of how something else could make one mount look better than the other, when the cause was something totally different. You really have to pay attention to your data and results. Couple something like this with average ambeint air temps not scaling correctly in a linear fashion, and it's easy to see how you can start chasing your tail.

http://lh3.ggpht.com/_BTn9YHMni9Y/SZbsjcK3JOI/AAAAAAAAC_c/wlbLYdx1TDk/s800/%5BAC%20DI%20v2.xls%5DCore%20Temp%20Average%20Char t%201%202142009%20110842%20AM.jpg

andyc

Nickel020

02-14-2009, 02:19 PM

Nice find! I didn't know this had this much of an impact
Do you think it might be usuefull to use safe mode for testing (does Prime run in safe mode?) since less processes run in safe mode?

Another option might be to increase the process priority of Prime.

mcoffey

02-14-2009, 02:39 PM

Not sure about safe mode, but you do have to be carefull about raising the priority to high on the loading SW, because then it can cause logging to drop out here and there thus missing data which causes a whole other problem.

andyc

Hondacity

02-14-2009, 05:29 PM

im not sure if you're going to test the i7

but its got lots of heat...which is suitable for watercooling...

i got some pics of idle and load power usage..if interested...i'll post it

mcoffey

02-15-2009, 12:21 AM

im not sure if you're going to test the i7

but its got lots of heat...which is suitable for watercooling...

i got some pics of idle and load power usage..if interested...i'll post it

Oh sure, no doubt about it. Just a little more of that Intel Koolaid and I'm all over it, but just not yet:D

andyc

Hondacity

02-15-2009, 01:31 AM

great..i like an open mind into tech

here's my idle..this is lowest power consumption i can get...whats connected to my psu..is the mobo+ram+raptor+videocard..thats it..the videocard is at stock...
http://i24.photobucket.com/albums/c5/hondacity25/image003.jpg

when running occt at 4.3ghz ht on at vcore 1.55
http://i24.photobucket.com/albums/c5/hondacity25/image001.jpg

the cpu is really hot...

i remember having the q9550....when overclocked..it can never pull 350watts from the killawatt...

remember we don't have the fsb...its all integrated to the imc inside the core i7 cpu...

this is just my observation...

i think this is also the reason why the evga x58 classified..has dual eps connectors on the motherboard...its just sick how this processors eat current...

here's the link to the evga classified dual eps connector

http://vr-zone.com/articles/evga-x58-sli-classified-preview/6361-3.html?doc=6361

mcoffey

02-15-2009, 01:37 AM

shoooo..that's a big chunk of current...wow..bonus times coming pretty soon, so might just be a good time to drop some coin on a i7 setup and a couple of 295's. really nice stuff.

thanks for the post,

andyc

mcoffey

03-15-2009, 12:32 PM

http://home.comcast.net/~mcoffey22/pwpimages/DSC_0002.JPG

http://home.comcast.net/~mcoffey22/pwpimages/DSC_0002.JPG

Silverion77

03-15-2009, 12:36 PM

so nice

Hows the mounting system?

mcoffey

03-15-2009, 01:44 PM

so nice

Hows the mounting system?

We'll see, that's the KL 350 mounting system shown in the pic, I still have to switch completely over to the HK mounting system. But I can tell you now, the HK pretty much kicks ass, with half the pressure drop of the KL 350. Based on what I've seen so far, I'd chose a HK over a KL block on a i7 setup.

here's a capture of HK best mount against KL 350 best mount. All just initial testing, not conclusive.

http://lh4.ggpht.com/_BTn9YHMni9Y/Sb1VoRYZNyI/AAAAAAAADOM/Ybz3QXL-XCU/s800/Microsoft%20Excel%20-%20Book1%203152009%2032228%20PM.jpg"

i7 920, 4.0, HT on, 20x200, 1.3v, Prime 95, 8K small in place FFT's. Again, this is without the HK3 mounting system.

http://home.comcast.net/~mcoffey22/pwpimages/HK%201200-1.JPG

Here's a shot of the plots I use to check the data integrity. The core on a i7 seem to act independently which helps alot for accuracy and slight differences between blocks.

http://home.comcast.net/~mcoffey22/pwpimages/i7%20HK%20v1.xls%203152009%2061941%20PM.jpg

andyc

HESmelaugh

03-15-2009, 02:37 PM

Great thread, andyc! :up:
Looking forward to more results! Especially that HK vs. 350 comparison is very interesting. :)

mcoffey

03-15-2009, 02:49 PM

Tell you what I find even more interesting is the delta t difference. Just goes to show, more than one way to skin a cat. HK3 does it via high flow and optimal base/mid plate design, KL does it thru impingement and fluid acceleration over the pins/core.

It would seem that the HK is a much more efficient design IMO. Really digging this block right now.

andyc

HESmelaugh

03-15-2009, 02:54 PM

I agree. It looks like the HK 3.0 manages a combination of low restriction and great temps that we haven't seen in other blocks before. At least, I haven't.

mcoffey

03-15-2009, 06:17 PM

Here's something interesting, and I glad I was able to capture it. Anyone thinking about adding a 2nd pump, or ubber strong pump with a free flowing block, might want to reconsider after reading this data.

Basically, I set up the bench with two new D-5's with EK v2 tops. I started the test with both pumps running, captured the data, then shut down the other pump capturing that data. Screen shots speak 1000 words:D

Check the fluid temp increase while the air in temp basically stayed the same.

With a free flowing block like the HK3 the only benefit from using two pump is redundancy. At least that's what the data shows, and the way I see it.

http://home.comcast.net/~mcoffey22/pwpimages/Microsoft%20Excel%20-%20Book1%203152009%2071032%20PM.jpg

And here's a shot that shows it all, or why someone would think 2 pumps would work better. When you first turn the 2nd pump off, you'll see a slight jump in core temp averages, but then core temps will level off, and start to decrease. If I continued to run the test for a hour after shutting down the 2nd pump, I would imagine you'd see slight better performance with 1 pump only. I shut the 2nd pump off at the 17:30 mark, and you can see the temps jump just a slight bit if you look carefully.

http://home.comcast.net/~mcoffey22/pwpimages/%5Bi7%20HK%20v2-2%20pumps%201pump%20off.xls%5DRad%20Air-Core%20Temp%20Averages%20Chart%201%203152009%20726 46%20PM.jpg

Pretty interesting stuff.

andyc

Langer

03-15-2009, 06:21 PM

Thanks for sharing that. Interesting results. I would have expected more from a 2nd pump with this block.

RedRaider

03-15-2009, 06:24 PM

:bird:KL-350

Hondacity

03-15-2009, 06:32 PM

mcoffey

03-15-2009, 06:35 PM

what was your radiator fan setup?

PA 120.3, ZM-F3's @ 1250

:bird:KL-350

Gotta tell ya, with what I know now, no way would I get a KL 350 with the restriction, and I haven't even began tweaking the orientation for best operation. The HK3 is just another example of a great German engineered block. Very smart design, and I know they didn't just stumble across it. They put some work in on this one.

andyc

Snyxxx

03-15-2009, 06:59 PM

Nice work. While your results are much more accurate, I found the same conclusions when testing 1 vs 2 pumps. Not worth it except for redundancy in all but the most restrictive systems.

zlojack

03-15-2009, 08:00 PM

Hmm...so my I guess I won't need two pumps on my CPU loop after all. Shit.

I already ordered them :D

mcoffey

03-15-2009, 09:22 PM

Hmm...so my I guess I won't need two pumps on my CPU loop after all. Shit.

I already ordered them :D

Still nice to have, and I run dual D-5's on two of the loops in my main build. Nothing wrong with running dual or hi performance pumps, just not the magic LC bullet some folks make it out to be. Not with todays gear and pumps.

andyc

zlojack

03-15-2009, 10:11 PM

Still nice to have, and I run dual D-5's on two of the loops in my main build. Nothing wrong with running dual or hi performance pumps, just not the magic LC bullet some folks make it out to be. Not with todays gear and pumps.

andyc
Yeah...down the road, if I finish my build with the pedestal and a bunch more rads, then the second pump on the CPU loop will probably be helpful.

We'll see what happens. I doubt it will hurt too much.

mcoffey

03-16-2009, 12:36 AM

Opps, almost forgot the new pump setup.

http://home.comcast.net/~mcoffey22/pwpimages/DSC_0006.JPG

andyc

mcoffey

03-22-2009, 08:57 AM

Seems like the same thing with the KL-350. No gain when using two pumps. Same setup as before.

i7 920, 4.0 Ghz, 1.30v, PA 120.3, ZM-F3's@1250 RPM, dual D-5's with EK v2 tops.

http://home.comcast.net/~mcoffey22/pwpimages/Book1%203212009%20101551%20PM.jpg

2nd pump shut off at around 20:30. Allowed the temps to normalize for 15 minutes, and captured the data. About as simple a test as there is.

http://home.comcast.net/~mcoffey22/pwpimages/i7%20KL%202pumps%20one%20OFF.xls%203212009%2010174 5%20PM.jpg

I don't know where others are seeing the reported gains from, I can't see it:shrug:

andyc

zlojack

03-22-2009, 09:06 AM

So basically the only benefit of a second pump is reduncancy.

Hmm...maybe I won't do a dual-pump loop on my CPU. Less hassle.

mcoffey

03-22-2009, 09:16 AM

So basically the only benefit of a second pump is reduncancy.

Hmm...maybe I won't do a dual-pump loop on my CPU. Less hassle.

I guess it has to do with the extra heat dump from the 2nd pump negating the added flow:shrug: Theoretically it should work, but practically speaking it doesn't with the pumps we use. AS long as your flows over 1 GPM, I think your good to go. Which isn't to say there's no benefit with 2 pumps, or people shouldn't run them, just not the road to thermal performance everyone makes it out to be.

I run a lot of dual pump setups, but for redundancy primarily.

andyc

Nickel020

03-22-2009, 08:56 PM

If you look at HESmelaugh's CPU cooler review then you see that cooling performance does no scale linearly with flow but the marginal performance benefit decreases as flow rate increases:
http://www.dexgo.com/index.php?site=artikel/view.php&id=356&rubrik=Hardware&seite=9
At 300 l/h (roughly 1GPM) the curves are already pretty flat and you need quite a bit flow to get another 1C out of those coolers. The added heat dump of the pump makes up for the better temp by improved flow rate though, if you think of it then the 24W of a D5 is quite a lot, that's like more than 10% of the heat output of a heavily overclocked CPU. Even if only 12W go into the water, that's a 5+% increase in total heat output (i.e. CPU + pumps).

Chicken Patty

03-23-2009, 12:12 AM

mcoffey

Let me ask you something. How do you mount the res and the pump together like that. What fittings do you have to get??? Thanks.

mcoffey

03-23-2009, 12:53 AM

2 BP D-Plugs, that's all it takes. I use the mid size. Sidewinders carries them.

andyc

Chicken Patty

03-23-2009, 12:58 AM

2 BP D-Plugs, that's all it takes. I use the mid size. Sidewinders carries them.

andyc

Thanks.

These are it?

http://www.sidewindercomputers.com/bishg14si1id.html

mcoffey

03-23-2009, 01:03 AM

Thanks.

These are it?

http://www.sidewindercomputers.com/bishg14si1id.html

Those are the longs which I use on some stuff like SLI connectors. On my pumps and rads I use the mid size like these. They also make a mini version, never tried those.

http://www.sidewindercomputers.com/bishg14sidse.html

They are handy as hell to have around. You can actually connect a 1/2 fitting on both ends and use them as quick disconnects, but they don't stop flow.

Chicken Patty

03-23-2009, 01:11 AM

mcoffey

03-23-2009, 01:24 AM

thanks dude. I might be getting a top for my MCP 655, ill see if I can use those to mount it to my res :D Thanks for the help bro.

No problem man, that's what it's all about:up:

andyc

mcoffey

03-24-2009, 05:54 PM

I should have been a plumber, I'm having way to much fun with these BP fittings. They run a little more quieter like this when pushing fluid hard, and works perfectly.

http://home.comcast.net/~mcoffey22/pwpimages/DSC_0005-2.JPG

And I'm starting to get better at this close up photography stuff:D All about the f stop baby.

andyc

Snyxxx

03-24-2009, 06:54 PM

I see now the hybrid impeller in the new DDC pumps. Orange for the gold ol' DDC-2 days and blue for the new generation.

Chicken Patty

03-24-2009, 07:35 PM

mcoffey, looks very nice bro :up:

mcoffey

03-24-2009, 07:38 PM

I see now the hybrid impeller in the new DDC pumps. Orange for the gold ol' DDC-2 days and blue for the new generation.

Yep,

Amazing how you can see the orange clearly even though the pumps are running:D

mcoffey, looks very nice bro :up:

thanks man,

andyc

mhw100

04-06-2009, 07:08 PM

Those are the longs which I use on some stuff like SLI connectors. On my pumps and rads I use the mid size like these. They also make a mini version, never tried those.

http://www.sidewindercomputers.com/bishg14sidse.html

They are handy as hell to have around. You can actually connect a 1/2 fitting on both ends and use them as quick disconnects, but they don't stop flow.

That seems like a great idea to use as a quick disconnect so as to drain the CPU part of the loop. It looks as if you've done jus that in the pics at the start of this thread.

1. Is it a D-Plug you have on the CPU in/out tubing coming/going from your waterblock or is it transfer base like this:
http://www.sidewindercomputers.com/bishsimubabp.html

2. If it is a D-Plug, then what female fitting do you need to put on the end of the D-plug i.e. the required fitting needs to have a barb on one end and a female coupling on the other?

3. My understanding was that anything on a loop will restrict flow. I can't tell from the pics on the SW website but they look like their a little restrictive on thin inside or is this just an illusion?

Thank you.

mcoffey

04-06-2009, 09:36 PM

1. yeah, both end of the D-Pligs are male, so you need some type of female coupler if you want to attached barb fittings to them. The one you have listed is exactly what I used on both ends of the D-Plug so I could connect 2 BP g1/4 1/2" OD fittings.

2. See answer above

3. no more restrictive than the threaded end of a BP 1/2" fitting.

Note: they're OK as a quick disconnects, and I use them that way on my bench. But they are not "stop" fittings that stop flow when disconnected.

andyc

mhw100

04-06-2009, 11:14 PM

Thanks Andy.

mhw100

04-06-2009, 11:38 PM

Just one other question Andy if I may.

I notice you have a tline running between the pump and the rad and then another tline between the rad and the cpu. I appreciate you are taking measurments off of these points but would it not be better to have the tline(s) between the res and pump so as to avoid any restriction other than the rad. In other words, would you not want the full force of the pump to hit the cpu without the impediments of the tlines? ...or is it as you've answered in 3. above wherein you state it is a mild restriction by virtue of the threaded end of 1/2" fitting?

mcoffey

04-06-2009, 11:56 PM

Hondacity

04-06-2009, 11:58 PM

dude...get off the restriction craze will ya...doesn't mean crap for most systems. You're kind of making a big deal over nothing:) Just use the fittings or whatever else you need to get the loop together the way you want, and don't worry about it.

andyc

+1 restriction is overated... as long as water is flowing...and your rad + fan combo is good you're good. if you're not satisfied get another pump

smee

04-07-2009, 12:02 AM

+1 restriction is overated... as long as water is flowing...and your rad + fan combo is good you're good. if you're not satisfied get another pump

+2, I used to worry about restriction too much myself, until i realized it was all bogus and the difference between a 90 degree fitting and no 90 degree fittings was hardly a thing.....

mhw100

04-07-2009, 07:16 AM

mcoffey

04-07-2009, 08:54 AM

mhw100

04-07-2009, 09:02 AM

NP...just want to you not to worry about stuff so much and focus more on getting the rig up and having a good time.

Planning's good, but it's pretty easy to think yourself into a tizzy if not careful.

Enough of that Dr. Phil shit from me.:D Now post a build log, get that puppy built and have a blast. Let us know if you have any other questions.

andyc

Thanks Andy - you've been a big help to me on a variety of different technical matters and it's always good to hear the calm voice of reason from an experienced veteran.