Foam Board Case

How is the cooling (airflow) in your case? I know a lot of people overlook this but air must flow in, over the coolers and then out. If you get a place where warm (hot) air is producing eddy currents that's guaranteed to jigger the effects of the cooling system as they are feeding (some) warm air back into the ducting. We can go into all sorts of esoteric methods but that's the first (and usually) lowest cost option.

When the system is running you can use a small smoke match (they are used to check airflow in boilers) to see if air is being sucked in at one end of the case and blown out of the other. You should have a negative air pressure (suck) at one end and positive at the other (blow) so cool air (typically about 22 celcius or 72 Fahrenheit) is getting through the system at all times.

It's useful to have a temperature gauge (and it can be a simple thermometer like you might find in a fish tank) at the exit to see what the temp of the exhaust air is - higher is better if the fans are working properly.

Additionally, I'm leery of falling back on Peltier effect coolers - at least in the first instance. The Peltier effect uses a power to create a differential temperature but the device itself has to be cooled down - which rather defeats the object. There are some excellent third-party cooler systems from places like Overclockers but they tend to concentrate on the CPU.

The difficulty with a large heatsink such as we might use on a CPU is that the vertical mounting - dating back to the very first personal computers - means they will over-stress the card and probably break it.

One possibility (I haven't tried but they do exist) is an external graphics card adaptor. Some of them are so simple they can sit on the desk and pull as much clean air as is available without the constraints of a case. Looks a bit Heath Robinson though.

Matt did a breathing PC which is rather cool (sorry, couldn't resist that quip) - and that might be worth a look.

A lot of us tend to forget that air is a fluid and fluid dynamics is a very difficult mathematical subject. Something of note - if you push too much air into the system, you will overwhelm the exhaust which must be capable of removing (as near as possible) the volume of air pushed in. If not, the pressure in the case will rise and the eddy currents will form, potentially causing areas of "dead air" where the air is moving only very slowly.

Air also has mass (something else we rarely consider) - but you can think of it as water with a very low mass. In fact, a lot of the active cooling effect is due to gaseous water in the air flowing through the case. Water is weird stuff (even though it's a very simple chemical we take for granted). It takes a lot of energy to heat the stuff up which is why water cooling is so ubiquitous in ICE cars and engines; although the water is cooling the oil which is in direct contact with the moving parts. Another option is CO2 but that's not practical for a home computer. Laptops use a chemical to conduct heat away from the chips (and I expect your GFX card does too) - so it can be passed to a larger heatsink which is cooled by forced air.