and 1/2" ID (3/4"OD) tubing with compression fittings.
what do you think?
I dont know which brand tubing to get so suggestions would be nice. Also, with the CPU water block i posted a link to a list of them. I want the EK Supreme HF but it comes in different variations such as full copper with gold plating, full nickel, full coppper, and things with plexi or acetal. is there a difference in performance or do they just look different. also, i am not sure about getting the GPU waterblock. it is expensive and idk if i need it. i may overclock my GPU some more but im not sure. i may just get the other stuff and get it later on but then i have to redo the setup. one last thing: what addatives should i buy for the water? thanks for the help!
Sounds like you might want to read through the water cooling sticky, if you haven't already done so. You have some decent choices there, but if you are planning on running your CPU and GPU in that loop, you might consider different radiator options. The Swiftech MCR rads are great (in fact, I use 2 of the ones you listed) but the fact you want to OC your video card (which appears to be a GTX460 and you didn't list the CPU...you might consider at least a little more radiator or running 2.
Tubing...is a matter of choice. I've only used Tygon and Feser and like both. Others like Masterkleer or Primochill, which I've also heard are good.
Compression fittings...look great, but expensive in most cases. They do not offer any improved performance, just FYI.
As for blocks being plated, etc. Doesn't help performance at all...simply antithetical value. There are some issues with EK GPU blocks and flaking nickel plating...you might want to read up on those before you make that choice for either CPU or GPU. I think ortoklaz has a thread on the forum about the EK nickel plating issues.
the res above is for dual loops ,how about this one or similar; http://www.performance-pcs.com/catalog/index.php?main_p...
for CPU/GPU loop and say 2 rads you only need 1 D5 (MCP655),if you absolutely must have EK blocks ..get pure copper
please post your specs and the case you plan on using,budget
Yeah, good call on the res...I simply glanced at it. However, I think it has plugs to run as a single loop res, but it really is designed as a dual loop res. Even if you ran dual loops, you could still run a single res...I know there are some people that do this as well simply for the increase of flow for a dedicated loop.
would a single bay reservoir be big enough? also, would a reservoir that doesnt go in the drive bay be recommended? if so where would i place it? i like the bay reservoir because i know where it goes and it is secure
the reason i chose that reservoir is it has the aluminum faceplate which i want so it still matches the case
I think there is an XSPC res that is exactly the same but for a single loop implementation.
if you dont recommend the EK block what do you recommend? i was looking at the Swiftech Apogee XT.
Either are a good choice.
i only plan on running one rad (unless 2 offers a huge performance upgrade) so which one radiator would you recommend?
Have you determined your thermal input for the loop in watts? If a CPU only, it would only be your CPU (pump is somewhat negligible in the grand scheme of things for most people) or if you add the GPU, you will need to add that as well and then choose radiator(s) and fans to dissipate this heat.
Skinneelabs.com has some of the best benchmarking and breakdowns of radiators and other watercooling gear
You can go with a bay res, or you can get a top/res for most pumps...however this usually voids the pump warranty. You can also get a bay res that includes a pump mount. There are some XSPC res's like the one you originally posted that allow this, as well as other brands.
For heat load, you can typically find the watts at full load on the web, usually from the manufacturer's site. For instance, my Q6600 Core2 Quad (old, yes I know...but she's still a beast) has a max TDP of 105 watts (which is stock). You can do a little math and figure out what overclocking might run you to, but it also involves more voltage, which is an algorithmic-curved increase due to the addition of speed and added voltage and thus, more heat watts being produced. Short story: most CPU's these days don't see over 150 watts unless you are really pushing them incredibly hard. (eh...might be 150-200w, but depends on chip/speed)
GPUs are similar...you can look up their max draw in watts (which is almost a 1:1 for draw and heat output for what we are concerned with) and find this number as well. Again, for instance, each of my GTX260's has a max load draw of around 180 watts (again...at stock). I run 2 of them in SLI...so that's 180w x2 + CPU (which is OC to 3.4...probably close to 200w just for argument's sake) = ~560 watts being put out as heat...and into my water loop.
(Sorry, had to add a little...) Determining actual heat/watt output isn't exact, but you can get some decent ideas Googling your components and their load specs. Some tests have gone as far and used multi-meters and power draw devices to help calculate these numbers, so you might find more exact details for specific hardware/chips at a certain speed, but they still are considered 'guesses' as it might take nominally different voltage changes from one system to another to accomplish the same reported clock speeds.
Now, that being said, you need to look for radiators that will meet/exceed these heat dispersal values with specific fans speeds and typically with a generalized concept of your loop's flow rate. All of this combined together represents a basic overview of how you determine your Delta in the water loop. (which is also discussed in the forum sticky...look for Delta T by Conumdrum...he did the write-up) <--went ahead and linked you
This is just a quick overview...you can Google the TDP of your components and get a pretty good idea of what you will need to cool components in a loop.
(This is also the perfect argument of why to have a rock solid power supply for a moderate to high-end rig...to have consistent, good power for these components).
I have a similar loop to what your looking into building and in my opinion your on the right track with your parts Im running ek supreme hf, koolance full blocks on the gpus, koolance mobo block mcp655 pump thermochill TA 120.3 rad 3 120mm scythe 133 cfm fans with fesser bi distilled water. My cpu sits at 35c gpus @ 29c. personally i think the 1/2 tubing does make a difference and so does the size of your res simply do to the fact you have more fluid in your loop and it takes longer for the ambient temperature to rise, just an opinion. Also im using the primochill flex tube and would recommend it. a good rule of thumb is 120mm rad surface per water block
personally i think the 1/2 tubing does make a difference
The difference in flow and temps between 3/8"ID and 1/2"ID is very minimal. You might see a little improvement with 1/2"ID. I have a link in the sticky that has some information surrounding this question.
so does the size of your res simply do to the fact you have more fluid in your loop and it takes longer for the ambient temperature to rise, just an opinion
Well, you are correct on the part about temps taking longer to rise. At some point though, any loop will reach a point where it reaches operating temps (idle being the coolest working temps) and will not get cooler unless you forcefully introduce cooler temps or turn the loop off. Even a 55 gallon drum as a reservoir would rise to a state of working temps at some point.
a good rule of thumb is 120mm rad surface per water block
Actually, 2x120mm is a good rule of thumb with most components these days. This is one reason most LCS coolers are less capable to cool a hot CPU (they only use a 120mm rad) and the pump is incapable of decent flow rates. I typically recommend a 2x120mm rad per CPU or GPU in the loop, but you can get by at stock clocks with a single 3x120mm rad. With some newer rads being 140mm, you start to allow for some different cooling options, but the 120mm is a good starting point in terms of cooling capacity and general planning.