Couple questions, first regarding the practicality of water cooling my system, and second being big one: effective radiator and mounting techniques
Currently, my rig is spec'd as follows:
C2D e6420 (2.13ghz stock, 8x266), currently @ 8x400, 3.2ghz
4gb, 2x2gb DDR2-800
GTX 285
3 HDDs, 2-160GB, 1-1TB
I've had the CPU over 3.4ghz, 8x430, and with cooling management I should be able to hit 3.6 with little effort. Thermal management regarding opening up voltages has been the deal breaker. Idle temps over 50 I am not pleased with, and that's not difficult if we don't have air on, and my second floor room at my parents house in the summer gets easily around 85ºF or higher if air is not on. At school, the air conditioning is always on, was easily able to keep the 3.4ghz OC stable.
My Northbridge is the killer, it's still passive, probably going to zip-tie a small fan to drop the temp a few ºC.
I think I can drop the vcore some, probably bring the CPU temps down a bit.
Current temps, on a decent night outside, not super warm in my room:
System: 43ºC
NB: 52ºC
CPU: 47ºC (hottest core, basic app load, average 5-10% CPU utilization)
GPU: 53ºC
Now, the issue is... I'd love to maybe see a 3.6-3.8ghz CPU OC, but with a C2D, how much more real world performance increases is that going to bring, and is it worth the cost of a couple hundred in a water cooling setup. Yes I know cost justification is a personal thing, just kind of venting how I am approaching this, and want to get some outside opinion on this to help with my decision.
Bringing the NB down a bit in temp, and working out the clock rate/voltage, I might be able to get 3.3-3.4ghz on air and still be manageable in the summer. Though I'd rather just play with the voltage and bring the current temps down, leave the current OC, versus keep the temps the same by upping the clock rate, which even managing the voltages would definitely keep the current temps the same.
BTW, using a Scythe Ninja B cpu cooler, one 120mm fan pushing across CPU toward system rear, one 120mm fan on the back, pulling air out of the system, and is 2 inches from the CPU cooler, directly in line, pretty good push/pull set up. Would be creating a terrible whirlwind by adding a dedicated pull fan, not much space.
Case is an NZXT Lexa, original series.
So with current configuration out of the way, that'll help with the justification decision.
Most importantly, I tend to upgrade the system components every few years, gaming is the main focus of the power, everything else, regardless of how power hungry an app is, I just get by with how it is, and power hungry apps I don't use much for anything serious. Playing around in Photoshop from time to time, encoding a video on occasion for PS3 playback or burning, or decoding for video playback on my system.
I enjoy the concept of overclocking my video cards, but really, with the current tech inside of them, how much real world gain can be had from putting an extra 50-75mhz on the core clock of the GPU, or shaders or memory. Whole video card generations seem to be where the real extra FPS come from, maybe a handful at most from just overclocking, or have I had bad experiences? When some games are only getting 60fps or less on an as-is card, is that low of clock changes going to increase more than 5fps in a game?
Now, moving on to the , "if I go this route, how am I going to accomplish it" part...
As I stated, I have an original series NZXT Lexa. Difference in the original and current series, no holes on top for mounting an external radiator. Case is small though, about 17" or so between the top structure "lips".
Currently, I've kind of decided that if I go water cooling, I'm going to skip the video card, just because the better blocks are specific to each card, and an expensive replacement with each upgrade. That, and since the GTX 285 is voltage-locked, no internal voltage regulator (EVGA even offers a voltage modifier utility, in warranty to boot, but has stated the 285 is not supported, nvidia reference design cut costs by locking the voltage), I am left to thermal management, and the luck of the board, with overclocking. Considering the cards can handle a great deal of heat, and my recent trials with OCing the board haven't granted a great deal of extra power (evga card is already overclocked from reference), it's really just a question of how much thermal output I'm comfortable with. The fan is amazingly quiet even at high load, so I'm satisfied with no water cooling block on the card. Easier to troubleshoot by removing the video card if nothing is attached as well. Not to forget the whole upgrading thing, that's the part I upgrade most often. Typically every other chip generation.
With that in mind, I'd be keeping any loop to this: CPU, then North Bridge. Maybe adding Mosfets and Southbridge into the equation. Don't have any insane performance hard drives, so that might not be a worthwhile venture.
Big question for this topic though: what size radiator is really going to be the most I'd really need. Let's say, my loop is: CPU, North Bridge, MOSFETs/Voltage regulators, then to the water cooling components.
Looking at my case, and the fact that I move my system at least a total of 4 times a year between summer and winter breaks, and sometimes to a LAN gathering with my buddies in the summer and winter. I wouldn't really want to deal with a lot of external complications if moving it a bit.
Plus, would I need a 2x120 or 3x120 if keeping a relatively modest cooling setup?
Could I gain good performance and overclock potential, with a single 120mm fan radiator?
If not, would mounting style is the best, considering my case? Would a rad/res/pump external system, with flow-lock couplings, be the better route, and just kind of set it on top/across the top of my case? Those units are not cheap, and the cooling blocks and everything else still need to be purchased as well.
Basically, I look to it as a fun hobby, but I don't want it to be simply a mean of keeping the current configuration with lower temps. If I do it, I want to enjoy the hobby aspect but also make it practical and give potentially more overclock room, by helping manage temps from voltage.
And then there's the desire to sometime in the relatively near future make the upgrade to i5/i7 when prices become more realistic, if performance gains are worthy. As a gamer, the big emphasis tends to be the GPU, and just making sure there are no components holding the system back.