Experimental radiator \build log!!!

Page 30 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
toolmaker_03

toolmaker_03

Honorable
Mar 26, 2012
2,650
0
12,960
well here are the radiators, my problem is the configuration of them, simply too many possibilities, so if you could find the time to give your opinions, I would really appreciate it.
I also need advice on a type of shroud for the radiators, the purpose is to inshore good air flow through the radiator.
and lastly how to attach the fans to the radiators?



all advice is welcome and thank you

 
Solution
rubix_1011
I don't use a flow meter, never have, never will.

Having a picture of a flow meter does nothing to support a finding, especially when you cannot be 100% conclusive that it is reading correctly or you are accurately converting what is being reported. Does the flow meter registering on the pump without restriction show a flow rate that is close to the rated flow of the pump with your conversion?

If so, great, but if so, you're still running lower than 1.0 gpm on your loop, meaning your delta-T MUST be offset to calculate your cooling potential. Meaning - you have to over-rad to accomplish the same cooling ability that you could achieve with a higher flow. Adding cost of 3x 360 radiators ($60, example) starts to get expensive...
Sort by date Sort by votes
Upvote 0 Downvote
Cause apparently the Dominators are a higher quality stick with better heatspreaders, hence why they are much more expensive.
Unless your overclocking the memory, I would just go with a cheaper kit.
Recommend these, same performance as the Vengeance models, but without the massive heat spreader (which is pointless unless you extreme overclock memory).
http://www.newegg.com/Product/Product.aspx?Item=N82E16820231610
 
Upvote 0 Downvote
Ok so I am thinking that maybe it is a little hard to understand exactly why I went to a parallel setup with this rig. The issue for me was the restrictive element within the blocks that I own. For example the XSPC ray storm water block has many thin grooves that water must pass through to leave the block this causes the flow in the system to drop drastically in any loop, there for it is the determining factor for overall flow of the system. 35mm rads are the same way, as far as the restriction on the system flow, they will be the determining factor in systems that do not have restrictive blocks. Now for the video blocks they have a pin matrix that the water must pass through again this is highly restrictive on a system, so I paralleled the video blocks in the hopes of increasing the total flow through each one. Example let’s say that the total flow through a block is 2500RPM you will not get that flow through the second block of the same type when connected in series the total flow through both cards will drop again now making the total flow at 1500RPM for the system. But if you have them in parallel you can get 2500RPM through both blocks assuming that the total flow coming into the cards is above 5000RPM. I hope that this helps, thanks again tool_03
 
Upvote 0 Downvote


i wish my block wasnt so high flow, i literally have to turn my pump down all the way using a fan controller to keep it quiet and get better temp results. the mcp355 runs 5000rpms, speed isnt always good when trying to cool everything. sometimes a drop in speed is necessary depending on what the pump runs at and how much pressure they can handle
 
Upvote 0 Downvote
Correct I had the same problem with my 480 blocks with the flow at 5000RPM my temps where quite high I found that the sweet spot for most blocks is between 1500RPM and 3000RPM any slower or faster you begin to have diminishing results on temps.
 
Upvote 0 Downvote
Well I went out and got some dominator 1600Mhz memory for this rig and the water blocks for them. Now that completes the last loop as far as hardware and fitting are concerned. So I am ready to build the final loop in the system but I am going to wait until I have the processor for the Intel board. That way when I do the board swap I will build the final loop into it as well. The loop will have a 360 rad a Z-multi tank 250 reservoir a pump and the 3 SSD cooling blocks and the memory blocks I will mount the rad on top of the case and the reservoir off the back of the case only the tubing and pump will be in the case. Still trying to keep it looking nice and clean so lots of fittings.
 
Upvote 0 Downvote
Finally I found a place that has fully explained what I have been trying to with this build so have a look and let me know what you think
http://dragonsgaming.wordpress.com/2011/04/06/water-cooling-video-cards/
The same logic applies to the radiators
 
Upvote 0 Downvote
Interesting, I always thought that having blocks in series was the better option, as that way you aren't halving the flow rate within the blocks. Seems that halving the resistance has more of an effect than keeping a full flow rate.

Particularly what I found surprising was the massive temp on GPU 2 when the cards were in Series, I was under the impression the water would be moving too quickly to make any significant difference. Maybe because he was just cycling ambient temp water through, and not actively cooling the water through a radiator so there is no equilibrium point where cooling and heat output cancel out.

A test using a radiator would be best as that represents what most (ie, 99%) of how water loops are set up, but its given me a bit to think about.
 
Upvote 0 Downvote
well my loop has radiators and the temp on GPU2 with this set up is on average 1C lower than GPU1 if that helps any and I had the same issue when running in series with this video card the temp on the second card was 20C higher under load than the first but with that setup my flow was at 900RPM a trickle still moving but not much it was to slow to control temps on the second card the first card was fine that is why I went through all of this.
 
Upvote 0 Downvote
This has nothing to do with it being a CPU block or a GPU block, this has to do with water blocks that have a Pin Matrix, or a Grove Matrix within them, making the block itself restrictive by design. Because of this restriction as you add each component in series the total flow of the loop will continue to drop by half of what you had. So you can see how this can add up to disaster really quickly as new components are added the flow continues to drop, all the components begin to suffer from this after the first, from lack of sufficient flow to remove the heat from the unit as it is produced.
 
Upvote 0 Downvote
Let Me Give some history on water block design back in the late 90’s when water cooling had a pretty good start, and the first blocks produced where designed to be open blocks with little dips in the bottom of the block to cause some turbulent flow inside them. But this design also allowed for great flow through the blocks themselves so to slow the system down the radiators where deliberately made to be restrictive with in turn increased there efficiency or ability to remove heat from the system. Today they have started making the blocks restrictive and now offer radiators that are not restrictive, like the Monstra series and other thick radiators like them. There are a few single pass 35mm radiators out there that do have good flow through them as well, but for the most part thin radiators are restrictive by design. So today you have an issue with the products being offered to the general market today, for the most part all of the most common products bought for a custom water cooling system are restrictive by design and if linked in series will have terrible results. I have seen this personally and have read about it on other threads although they may not have understood the problem they quickly realized that one existed. Temps where through the roof on all the components after the first one, is usually the first clue, or issue that will occur. This issue is even encountered in the sticky under the duel loop verses single loop scenario as long as the system never gets any larger than the CPU and one Video card the issue is not very apparent, but once they add the second card to the system the temps one both cards went through the roof. They called this setup an extreme situation but in reality anyone wanting to do a SLI or crossfire setup will run into this issue.
 
Upvote 0 Downvote
Ok so I like the corsair dominator memory and I have the water blocks for them. I found a set of 4 with 8-8-8-24 timings, but its only 16 gigs or 4 x 4 gig sticks, they also have a 32 gig set 4 x 8 gig sticks but the timings are a step slower at 9-9-9-24 cant seem to find any real info that says one is faster over the other though. So at this point it looks like stability will be the deciding factor on memory type.
 
Upvote 0 Downvote
Memory performance doesn't have much impact to overall system performance, vast majority of the time its not worth paying for faster RAM as it has such little benefit over your standard 1600Mhz CL9. If you are going for faster memory anyway, in most applications you want a faster frequency, not lower timings.
But regardless, their Dominator sticks, they had better overclock given their price tag and that their under water :lol:/
 
Upvote 0 Downvote
Have not had much luck with stable over clocks on memory, I would be happy with memory that does not cause the system to reboot or give a blue screen in the middle of game play! I just like to play games man and the less that I have to deal with in order to do so, the better.
 
Upvote 0 Downvote
Man I am stocked let me lay out this build and how I will finish it out.
DX79SR MoBo
i7 3970X CPU
32gigs 1600 memory
2 240gig intel 520 SSD's in raid 0
For now 2 X 580's but when they are released, I will get 2 X 790's to complete the build.
Now I can only hope that this will hold me over for a few years. :na:
 
Upvote 0 Downvote
So here are my thoughts I will start by leaving the setup the way it is now, I will only add a 3rd card in parallel with the two already there in the loop, but if I need more radiator to cool all of the cards. I will add a 3rd radiator in parallel with the two radiators I have now, and I know that will be enough to cool all of the video cards.
 
Upvote 0 Downvote
Ok so I have been reading the manual on the DX79SR motherboard and it says that it is capable of 3 way SLI at 16x across all three slots instead of the normal 8x across all 3 so my question is two things one will this feature work with my video cards, and secondly what kind of benefits will this provide for the system I do not understand all of the features of a PCIe 16x 3.0 enabled motherboard?
 
Upvote 0 Downvote
1x,4x,8x and 16x all specify the number of physical lanes between the PCI slot and CPU, which determines the bandwidth between them (think of how wide a highway is). More lanes, the more bandwidth and therefore the more data (cars on a highway) can be moving at any particular time.

PCI Gen3 offers more bandwidth per lane than PCI Gen2 does, about double. So a PCI-3 1x slot is equivalent in bandwidth to a PCI-2 2x slot (theoretical anyway, 2x slots don't exist). PCI-3 4x is equivalent to PCI-2 8x and so on.
In terms of actual benefits, for graphics cards there is pretty much none. A PCI-2 8x slot is more than enough for even the strongest of current cards, PCI-3 is largely a future proofing measure and to pave the way for PCI based storage mediums.

All the current gen cards (GTX 600, HD 7000 and the brand new GTX 700 series) use PCI-3 interfaces and can access that greater level of bandwidth, though you wont see any real benefit for it.
 
Upvote 0 Downvote
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom