- Joined
- Feb 29, 2012
Hello, first post on this board so bit of a mini intro. I'm researching for a future build (which may not happen for 12 months but almost certainly will happen within 24 months.. one holdback being waiting to see how nextgen techs pan out and such) and so I am exploring my options. Budgets DO matter, ie I want the best value for money, but so does performance... if I cant reach performance targets without spending twice as much, then I spend twice as much. But my performance targets can sometimes be a moving target, after playing I realize "I want more" and so i'm trying to plan out an upgrade path just to understand how much more can even be expected. (ie - if I want 20% more performance and have $800 to spend on cooling, is it even possible or generally no?)
I'm wanting to know what kind of rules of thumb are out there of how much faster one can expect to go using phase change methods than using high quality watercooling. For all uses - CPU, GPU and chipsets.
So here's what i'm thinking (and if there's some problem please point it out):
Step One: solid watercooling rig that essentially brings temperatures down to darn near ambient under load.
Step Two: subambient watercooling rig preferably using some form of antifreeze (I assume?) to get below freezing, some type of chiller used for the cooling loop.
Step Three: large peltier into waterblock type systems getting below common antifreeze points.
Step Four: phase change cooling systems
What % of improvement can one expect at each level? Ie - how much further does one tend to be able to take a CPU, or chipset, or graphics card with the different methods? In the jump from "solid air cooling" to "extreme phase change" has anyone ever achieved a 60% improvement or 40% or what exactly do the results tend to be? (it sounds like the ultimate limits are like liquid nitrogen hitting 8ghz on processors which is around a 60% improvement over the stoutest air reaching 5ghz for alot... i'm curious if the same has been attempted with video cards) I'm trying to figure out how much you spend at each step just to go from 20% to 30% to 40% for instance. Can phase change give me a 6ghz i7 or a 1.5ghz geforce 580? Can that be achieved with peltier or subambient water?
And is there a point where more cooling is overkill (ie chips limited by safe nonsuicidal voltages, or just inherent stability limits) or will cooling ALWAYS get a boost in performance?
For general information i'm only interested in 2-8hr runtimes (don't think a thermal sink could be big enough or 'temporary' solutions like liquid nitrogen - I don't need 24/7 running, just extended gaming sessions... i'd considered large water resoirvoirs filled with ice if I want to stretch the limit for a perhaps shorter gaming session though) and i'm not interested in destroying chips (dropping another $1000 to replace damaged hardware isn't my goal) or shortening lifespan to the 'weeks' level. Just pushing performance well past what's available with simpler systems.
I'm wanting to know what kind of rules of thumb are out there of how much faster one can expect to go using phase change methods than using high quality watercooling. For all uses - CPU, GPU and chipsets.
So here's what i'm thinking (and if there's some problem please point it out):
Step One: solid watercooling rig that essentially brings temperatures down to darn near ambient under load.
Step Two: subambient watercooling rig preferably using some form of antifreeze (I assume?) to get below freezing, some type of chiller used for the cooling loop.
Step Three: large peltier into waterblock type systems getting below common antifreeze points.
Step Four: phase change cooling systems
What % of improvement can one expect at each level? Ie - how much further does one tend to be able to take a CPU, or chipset, or graphics card with the different methods? In the jump from "solid air cooling" to "extreme phase change" has anyone ever achieved a 60% improvement or 40% or what exactly do the results tend to be? (it sounds like the ultimate limits are like liquid nitrogen hitting 8ghz on processors which is around a 60% improvement over the stoutest air reaching 5ghz for alot... i'm curious if the same has been attempted with video cards) I'm trying to figure out how much you spend at each step just to go from 20% to 30% to 40% for instance. Can phase change give me a 6ghz i7 or a 1.5ghz geforce 580? Can that be achieved with peltier or subambient water?
And is there a point where more cooling is overkill (ie chips limited by safe nonsuicidal voltages, or just inherent stability limits) or will cooling ALWAYS get a boost in performance?
For general information i'm only interested in 2-8hr runtimes (don't think a thermal sink could be big enough or 'temporary' solutions like liquid nitrogen - I don't need 24/7 running, just extended gaming sessions... i'd considered large water resoirvoirs filled with ice if I want to stretch the limit for a perhaps shorter gaming session though) and i'm not interested in destroying chips (dropping another $1000 to replace damaged hardware isn't my goal) or shortening lifespan to the 'weeks' level. Just pushing performance well past what's available with simpler systems.
Last edited:
That's the kind of quickie info I was looking for. It's sounding like normal ambient watercooling may be the best just due to the condensation issues itself vs the performance boost not as much more as i'd want... instead of spending $800 for cooling, could just start with faster parts like i7-3960X or just wait for the ivy bridge or haswell or similar. It sounds like watercooling should be good enough... until the next models come out and then just upgrade instead of spending it all on ultracooling with phase just to match the performance at extreme electricity costs ahead of the curve.
