Palit Radeon HD 4870 X2 2GB Video Card Review

Temperature Testing

For this heat test we went about things a bit differently since there is currently no program available which monitors and logs temperatures for the new ATI cards. So, for this test we loaded the card with 3DMark’s Batch Size Test at the highest triangle count to put a constant and high load on both GPUs for 1 hour while keeping two instances of GPU-Z’s Sensors tab open. Right before the 30 minute test finished, we quickly Alt + Tab to the desktop and quickly take a note what temperatures the cores peaked at. While this may not be the most scientific method, we are reasonably sure about its accuracy so take it as you will.


Idle temperatures on this card are actually very good for a dual chip card but it is very easy to tell which of the cores is further from the fan. While we were sure to see some extremely high temperatures, it was surprising to see the hotter of the two cores never breaking the 88C barrier after a whole hour of high load. The cooler of the two cores was well within the norms due to its proximity to the torrents of cool air being moved by the fan. Overall, this isn’t too bad of a result but we can see what effect the PCB layout and heatsink design has on this type of two core design; one core will always run significantly hotter than the other.

Something else that should be brought up is that this card gets very, VERY hot on every surface so make sure that you let it cool down for at least 10 minutes before even thinking of touching it. It also outputs waves of heat so be prepared to sweat like a pig if your room isn’t well ventilated. Indeed, by holding a temperature probe 2” away from the exhaust grille we measured a whopping temperature of 52C. This is great for those Canadian winters but it seriously sucks in the dead heat of summer.

Acoustical Properties

If you have sensitive hearing you may want to look the other way instead of buying this card. While it is not s loud as the GTX 280 cards we have had in our possession, the HD4870 X2 does make its presence known if your game audio isn’t turned up to medium. It isn’t an annoying sound but it is rather a loud “whoosh” which is accompanied by a blast of hot air. Since the air within the heatsink needs to move over two separate fin assemblies while maintaining its speed, the single fan naturally has to spin up to some pretty high RPMs. If you have a case with a perforated side panel (the Gigabyte Aurora 570 I use is a perfect example of this) you may want to think about blocking it off if you insist on absolute silence.

This kind of performance always has its tradeoffs and as has always been the case with high powered cards, that tradeoff is usually found in the acoustical footprint of any high end card.

Power Consumption

For this test we hooked up our power supply to a UPM power meter that will log the power consumption of the whole system twice every second. In order to stress the GPU as much as possible we once again use the Batch Render test in 3DMark06 and let it run for 30 minutes to determine the peak power consumption while letting the card sit at a stable Windows desktop for 30 minutes to determine the peak idle power consumption.

Please note that after extensive testing, we have found that simply plugging in a power meter to a wall outlet or UPS will NOT give you accurate power consumption numbers due to slight changes in the input voltage. Thus we use a Tripp-Lite 1800W line conditioner between the 120V outlet and the power meter


The first thing that is sure to catch your attention is the fact that idle power consumption of the HD4870 X2 is seriously lowered over the Crossfire setups to the point where it hovers below that of dual HD4850 cards. It seems like ATI has finally got PowerPlay working in this card at least so hopefully they will get it to work in the rest of the lineup as well.

What completely stunned me was the fact that peak full load power consumption actually decreased a bit over a pair of HD4870 cards even though it has 1GB more memory in total. This left me scratching my head a bit but it is possible that the single PCB layout is more efficient than two separate cards on two separate PCBs connected by a Crossfire bridge. Thinking there was something amiss, the test was run twice with the same result so it seems our assumption of a more efficient design could very well be correct.

Since these power consumption figures leave the system memory and the processor pretty much alone, we would recommend that you use a good 700W or higher power supply with an 8-pin PCI-E connector if you are planning on using this card.