Rage3D Discussion Area

Rage3D Discussion Area (http://www.rage3d.com/board/index.php)
-   General Hardware (http://www.rage3d.com/board/forumdisplay.php?f=11)
-   -   -Post pics of your computer- (http://www.rage3d.com/board/showthread.php?t=33744307)

aviphysics Oct 4, 2012 01:46 PM

Quote:

Originally Posted by BababooeyHTJ (Post 1337054015)
Yeah, but in that direction you would have the heat that was just blown off the first rad blowing right through the second.

The way I look at it is that your airflow should be high enough so that the air does not undergo a signifigant temperature change durring its path through the radiators. This would at least be the ideal for optimal cooling.

You can optimize under non ideal conditions by putting the radiators in series and blowing the cool air through the second radiator in the series first. This maximizes heat flux by maximizing the mean temperature differential betwen the air and radiator. I believe you would also want the radiators configured so that the coolest part of the second radiator is in front of the coolest part of the first radiator. I hope that was clear enough. If not, you can just google counter current cooling which is basically what I am talking about.

The symmetry of the problem suggests to me that putting a divider between them wouldn't really help. You would still have a pressure build up between the fan and the divider that would be aproximately equivalent to the effect of the fans fighting eachother.

What might be more help is removing the middle puller fans. This would make more room for air to flow between the radiators. Fans are much better at pushing than pulling anyway so it doesn't seem like the puller fans would be that helpful. You basically never see a puller fan on commercial hardware and I think you would see it at least sometimes, if it offered a signifigant benefit.

All that said... it may not be to much of a problem in the first place, thus the question mark in my post. It is just a question of whether the gap is big enough and how well the fans can tollerate the pressure build up.

My intuition on the matter comes from several years working a large water cooling system for General Atomic's DIII-D experimental fusion reactor. I haven't done any of the math on this one so as I said, it might not make much of a difference.

shadow001 Oct 4, 2012 01:48 PM

Quote:

Originally Posted by aviphysics (Post 1337054308)
I just gave a little friendly jab to the ant hill. I am not the one that kicked it over.

The two systems just beg comparison, being that they are both over the top water cooled dual socket multi GPU mammoths.


Well, he'll have the ability to handle 32 threads in total, as he's got 16 physical cores and hyperthreading will be enabled, not to mention the fact that his processors are better IPC wise so that should add another 20% extra performance....My setup on the other hand, has 12 physical cores and handles 24 threads, they are older CPU's that aren't as good in terms of IPC, but are running at 4 Ghz constant, and 4.3 Ghz with turbo mode when not all cores are being used.

I'm guessing it'll be a tight fight overall....Better IPC with newer processors having more cores and threads on his setup, versus older processors with fewer cores but running at much higher clock speeds.

Quote:

Originally Posted by Redeemed (Post 1337054315)
Oh I know and understand that. Still don't like it. :lol:

Just means I'll never have such a system is all, even if I could afford it. :)

SR-2 motherboards are now cheaper than they used to be, and the Xeon chips they use are sometimes on special....The last i saw were X5680's wich can easily overclock to 4.5 Ghz, on special for 750$ and these are processors that cost 1500$+ when they launched, and they're basically clearing socket 1366 inventory now that the E5 series for socket 2011 xeon's are available for the last few months, and so is the EVGA SR-X motherboard on the market, but Asus also makes one too that has overclocking capabilities too.


Anyhow, if it wasn't for the fact that Intel launched these chips with completely locked multipliers, i'd be all over it for sure, and i might still go for the SR-X motherboard but only when Intel releases the ivy bridge based Xeon's, wich it seems that they'll have a full 12 cores in each processor( for the high end models), so a pair of those totals 24 cores and handles 48 threads.

Even if intel pulls the locked multiplier trick a second time, going from a setup with 12 cores and handling 24 threads, to one with 24 cores and handling 48 threads will be an obvious upgrade, even if the chips are running in the low 3 Ghz range.

aviphysics Oct 4, 2012 01:57 PM

Quote:

Originally Posted by shadow001 (Post 1337054339)
Even if intel pulls the locked multiplier trick a second time, going from a setup with 12 cores and handling 24 threads, to one with 24 cores and handling 48 threads will be an obvious upgrade, even if the chips are running in the low 3 Ghz range.

Now if we jumped into the way back machine I believe we would find that you justified the extravagance of your machine with the argument that it would take a very long time for even normal high end systems to match it and therefore require an upgrade to run games at max settings.

It would seem that performing a major upgrade in the near future would mean all the people that called your system a ridiculous e-peen enhancement were right.

The temptation to upgrade so soon when you claimed your system would last for so long is exactly what I was poking a little fun at.

shadow001 Oct 4, 2012 02:07 PM

Quote:

Originally Posted by aviphysics (Post 1337054350)
Now if we jumped into the way back machine I believe we would find that you justified the extravagance of your machine with the argument that it would take a very long time for even normal high end systems to match it and therefore require an upgrade to run games at max settings.

It would seem that performing a major upgrade in the near future would mean all the people that called your system a ridiculous e-peen enhancement were right.


You'd be absolutely correct in that assesment that i did say that, but this is in the context of going beyond gaming environments and offering enough performance boost over what i currently have already....I gave up on game developers ever going beyond 4 cores unless the next generation of consoles have more than 4 cores.

The other issue is the price of the upgrade, as the very highest end Xeon's for 2 socket motherboards are close to 2000$ a piece when they first launch, and i'm sure that the ivy bridge Xeon's will be priced that way....Add the SR-X motherboard to the shopping list and with taxes and shipping costs, i'll be looking at 5000$ easy, even if i can reuse the case and all the cooling gear(except for the CPU water blocks of course)....

Suffice to say that i won't be happy if the performance increase was only marginal( say 15~20% faster). for the amount of money involved in the upgrade, and i am putting it mildly...:lol:

DarthBeavis Oct 4, 2012 03:06 PM

shadow001 I will give you credit for packing in a ton of stuff into a small space, not easy to do

I am curious about how you are getting airflow to the mobo especially the ram.

Since I am an NVIDIA guy, I operate on the paradigm of GPU > CPU.

Not sure why you have five pumps. Two D5s with everything (parallel GPUS but still in series with rest of blocks and rads) in series is just fine. At least this is my experience . . .I think I have build most H20 systems than most people so I have a ton of experience and have tried all sorts of loops.

BababooeyHTJ Oct 4, 2012 03:45 PM

The thing that I find funny about shadow's rig is that its loaded with overkill like the five D5 pumps but then he goes and fills the case with a bunch of $5 sleeve bearing fans. :bleh:

I'm having trouble keeping my three 7950s under 50c with the fans spinning nice and slowly. I can understand the need for so much rad space for four 7970s.

caveman-jim Oct 4, 2012 03:55 PM

Quote:

Originally Posted by DarthBeavis (Post 1337054432)
Since I am an NVIDIA guy, I operate on the paradigm of GPU > CPU.

you mean you're and NVIDIA enthusiast or you're an NVIDIA employee?

DarthBeavis Oct 4, 2012 04:09 PM

Quote:

Originally Posted by caveman-jim (Post 1337054509)
you mean you're and NVIDIA enthusiast or you're an NVIDIA employee?

Enthusiast who they sometimes pay to do special projects (1099 contractor). I DO NOT work for them inhouse at all (though I would jump at the chance if ever offered).

The context of this thread is enthusiast as I truely believe the paradigm versus being a paid shill.

caveman-jim Oct 4, 2012 04:16 PM

gotcha :D:up:

bman212121 Oct 4, 2012 04:33 PM

Quote:

Originally Posted by DarthBeavis (Post 1337054524)
Enthusiast who they sometimes pay to do special projects (1099 contractor). I DO NOT work for them inhouse at all (though I would jump at the chance if ever offered).

The context of this thread is enthusiast as I truely believe the paradigm versus being a paid shill.

They're not going to hire you if you keep "Ruby" around. :p

DarthBeavis Oct 4, 2012 06:59 PM

Quote:

Originally Posted by bman212121 (Post 1337054545)
They're not going to hire you if you keep "Ruby" around. :p

lol I gave Ruby to my former Crossfit trainer when he and his (ex) wife were at my house for a party last year.

shadow001 Oct 4, 2012 08:11 PM

Quote:

Originally Posted by DarthBeavis (Post 1337054432)
shadow001 I will give you credit for packing in a ton of stuff into a small space, not easy to do

I am curious about how you are getting airflow to the mobo especially the ram.

Since I am an NVIDIA guy, I operate on the paradigm of GPU > CPU.

Not sure why you have five pumps. Two D5s with everything (parallel GPUS but still in series with rest of blocks and rads) in series is just fine. At least this is my experience . . .I think I have build most H20 systems than most people so I have a ton of experience and have tried all sorts of loops.


What i did was setup the loop in such a way as there's excess airflow coming from the four radiators placed on the bottom half of the case(3 are for the GPU's and 1 for the CPU), and exit's from the top of the case as we all know hot air tends to rise anyhow....The memory itself is running at the speeds it's rated for and using the usual 1.65 volts( it's DDR 3 at 1866Mhz, 8-8-8-24), and i haven't had any issues so far, then you see the top radiator wich is also for the CPU loop too and all the radiators are double or triple thick and took an entire gallon of water to fill up and bleed...:lol:


As for the pumps, it is overkill indeed as i have a ton of flow, but they're there for redundancy more than anything else even if i run the system overnight should a pump fail....That's where you see the 2 front mounted pump controlers on the lower 5 1/4 bays wich one is for the CPU loop and the other for the GPU loop, so i can slow them down or speed them up if it's a really hot day.


Same idea goes for the PSU's, wich in my previously case, i was using a single silverstone strider 1500 watt PSU and it let go in less than 18 months, especially with all the overclocking and over volting...The system just draws too much power for a single PSU, so i split the loads roughly in half with one handling 3 video cards and all accesories for the GPU water loop + SSD's and hardrives, and the other PSU handling the first video card up top and the CPU's and the motherboard and all the accesories for the CPU water loop( pumps, fans, controlers).


Basically each PSU is drawing maybe 700~800 watts a piece as a worst case scenario where both the video cards and CPU's are getting hammered to 100% loads, and leaves quite a lot of headroom should CPU's or video cards continue to consume even more power in the future.....The basic 70% load rule still applies if you want the PSU's to last for the long run, and the silverstone proved it nicely when some of the rails just let go.


I picked the corsair AX 1200's because they're single rail designs for the 12 volt feed, so no need to worry about overloading a specific rail.

shadow001 Oct 4, 2012 08:12 PM

Quote:

Originally Posted by BababooeyHTJ (Post 1337054496)
The thing that I find funny about shadow's rig is that its loaded with overkill like the five D5 pumps but then he goes and fills the case with a bunch of $5 sleeve bearing fans. :bleh:

I'm having trouble keeping my three 7950s under 50c with the fans spinning nice and slowly. I can understand the need for so much rad space for four 7970s.


They're actually scythe fans....Just saying..;)

DarthBeavis Oct 4, 2012 08:26 PM

Quote:

Originally Posted by shadow001 (Post 1337054700)
What i did was setup the loop in such a way as there's excess airflow coming from the four radiators placed on the bottom half of the case(3 are for the GPU's and 1 for the CPU), and exit's from the top of the case as we all know hot air tends to rise anyhow....The memory itself is running at the speeds it's rated for and using the usual 1.65 volts( it's DDR 3 at 1866Mhz, 8-8-8-24), and i haven't had any issues so far, then you see the top radiator wich is also for the CPU loop too and all the radiators are double or triple thick and took an entire gallon of water to fill up and bleed...:lol:


As for the pumps, it is overkill indeed as i have a ton of flow, but they're there for redundancy more than anything else even if i run the system overnight should a pump fail....That's where you see the 2 front mounted pump controlers on the lower 5 1/4 bays wich one is for the CPU loop and the other for the GPU loop, so i can slow them down or speed them up if it's a really hot day.


Same idea goes for the PSU's, wich in my previously case, i was using a single silverstone strider 1500 watt PSU and it let go in less than 18 months, especially with all the overclocking and over volting...The system just draws too much power for a single PSU, so i split the loads roughly in half with one handling 3 video cards and all accesories for the GPU water loop + SSD's and hardrives, and the other PSU handling the first video card up top and the CPU's and the motherboard and all the accesories for the CPU water loop( pumps, fans, controlers).


Basically each PSU is drawing maybe 700~800 watts a piece as a worst case scenario where both the video cards and CPU's are getting hammered to 100% loads, and leaves quite a lot of headroom should CPU's or video cards continue to consume even more power in the future.....The basic 70% load rule still applies if you want the PSU's to last for the long run, and the silverstone proved it nicely when some of the rails just let go.


I picked the corsair AX 1200's because they're single rail designs for the 12 volt feed, so no need to worry about overloading a specific rail.

redundancy = 2 pumps, 5 pumps = ocd :D

nycdarkness Oct 4, 2012 08:29 PM

Counts number of fan controllers, similar to the pump situation.

DarthBeavis Oct 4, 2012 09:06 PM

Quote:

Originally Posted by nycdarkness (Post 1337054718)
Counts number of fan controllers, similar to the pump situation.

I am using Corsair Cool Link instead of that type of controller.

http://www.corsair.com/corsair-link/...hting-kit.html

http://www.youtube.com/watch?v=i5auv...layer_embedded

shadow001 Oct 4, 2012 09:45 PM

Quote:

Originally Posted by DarthBeavis (Post 1337054714)
redundancy = 2 pumps, 5 pumps = ocd :D

Remember though, there's 2 water loops.....One for the GPU's and another for the CPU's and motherboard blocks, so each loop gets 2 pumps.


The fifth pump just comes with the reservoir that i was using from a previous system that was also water cooled...:lol:

shadow001 Oct 4, 2012 09:46 PM

Quote:

Originally Posted by nycdarkness (Post 1337054718)
Counts number of fan controllers, similar to the pump situation.


All the channels are being used for fans and nothing is wasted....:p

Kasucode Oct 4, 2012 10:33 PM

Quote:

Originally Posted by DarthBeavis (Post 1337054432)
shadow001 I will give you credit for packing in a ton of stuff into a small space, not easy to do


Not trying to brag or anything :) ive put 3 loops with 3 D5 pumps, a 120.3 two 120.2 and a single 120 rad in a TJ07 was my 2nd time doing watercooling around 4+ years ago. ive done many water loops sense.




Anyways I like your setup and ive always been a fan of your previous builds. Im looking forward to seeing it come together. I was thinking about hitting you up. im working on a big project something i been wanting to do for years.

mullet Oct 4, 2012 11:35 PM

Kasucode, that thing is a beast!!!

DarthBeavis Oct 5, 2012 12:03 AM

Quote:

Originally Posted by Kasucode (Post 1337054807)
Not trying to brag or anything :) ive put 3 loops with 3 D5 pumps, a 120.3 two 120.2 and a single 120 rad in a TJ07 was my 2nd time doing watercooling around 4+ years ago. ive done many water loops sense.




Anyways I like your setup and ive always been a fan of your previous builds. Im looking forward to seeing it come together. I was thinking about hitting you up. im working on a big project something i been wanting to do for years.

beautiful work - I consider you a veteran bro ;) I have seen your projects here and there as well. Your beginning work is way above my beginning work.

Hogrider Oct 5, 2012 12:20 AM

Forget everything else in this thread... Beavis got rid of his mannequin! :cry:

BababooeyHTJ Oct 5, 2012 04:56 AM

Quote:

Originally Posted by shadow001 (Post 1337054701)
They're actually scythe fans....Just saying..;)

I thought that you were using Yate loons, which scythe fans did you pick up?

mullet Oct 5, 2012 09:31 AM

More pictures less text, k thx bye.

aviphysics Oct 5, 2012 10:11 AM

Quote:

Originally Posted by DarthBeavis (Post 1337054741)


I still think that you would be better off without the pulling fans on your radiators to free up more space for airflow in the middle. I remember seeing an analysis years ago that showed that putting fans in a push pull configuration doesn't help much anyway.

DarthBeavis Oct 5, 2012 11:01 AM

Quote:

Originally Posted by aviphysics (Post 1337055067)
I still think that you would be better off without the pulling fans on your radiators to free up more space for airflow in the middle. I remember seeing an analysis years ago that showed that putting fans in a push pull configuration doesn't help much anyway.

Will see what the performance is like

VW_Factor Oct 5, 2012 11:21 AM

Quote:

Originally Posted by aviphysics (Post 1337055067)
I remember seeing an analysis years ago that showed that putting fans in a push pull configuration doesn't help much anyway.

Thats funny.

Even on my H60 I noticed a difference. Although I am using some quiet fans that while at their top speed have good pressure, I don't ever run them at full speed, so I don't really know.

What I do know, is that I noticed a good 6c difference with a push/pull on it.

BababooeyHTJ Oct 5, 2012 12:26 PM

Quote:

Originally Posted by VW_Factor (Post 1337055164)
Thats funny.

Even on my H60 I noticed a difference. Although I am using some quiet fans that while at their top speed have good pressure, I don't ever run them at full speed, so I don't really know.

What I do know, is that I noticed a good 6c difference with a push/pull on it.

It really depends on the configuration. From the tests that I've seen with some rads it doesn't make much of a difference but with others it can. I guess that the XSPC EX series really benefits. A lot of thicker rads really seem to benefit.

One test that I found to be surprising was on Martin's liquid lab where on his test bench an xspc rs 360 with fans in push pull out performed the RX 360 with a single set of fans.

I have a feeling that logical is probably right about the fans. Those pull fans really do look like they're fighting with eachother. Although the sr-1 is does like higher speed fans so I wouldn't be surprised if it did benefit from push/pull in general.

bman212121 Oct 6, 2012 09:58 AM

Quote:

Originally Posted by DarthBeavis (Post 1337054655)
lol I gave Ruby to my former Crossfit trainer when he and his (ex) wife were at my house for a party last year.

Apparently she got jealous. :D

BababooeyHTJ Oct 7, 2012 10:21 AM

I just added a new pump and radiator. I really need to buy a different res. That swiftech 35x res was a nightmare to fill and bleed.



All times are GMT -5. The time now is 12:56 AM.

Powered by vBulletin® Version 3.6.5
Copyright ©2000 - 2022, Jelsoft Enterprises Ltd.
All trademarks used are properties of their respective owners. Copyright 1998-2011 Rage3D.com