Thermal Performance

We've changed our thermal testing procedures to one that we feel more accurately represents real world gaming. When your in a gaming session there are peaks and troughs in demands that are put on your GPU. These can be in the form of dying and getting a loading screen, messing around in the menu system or simply travelling around part of a map where there aren't any explosions of other pwnage to render. During these stages, your graphics card cooler will dissipate any excessive heat that was built up during more graphically periods of game play.

Synthetic benchmarks such as FurMark thrash the GPU constantly. Not only is this not reflective of how a GPU will be getting abused in gaming, it's such a hardcore test that any GPU under test is almost guaranteed to hit it's thermal limit, the mark at which the card's firmware will kick, speeding up the fan to keep the GPU withib safe temperature limits. For this reason, we were getting results that were more dictated by the cards' firmware than the cooler, as no matter how good the cooler was, FurMark was going to keep pushing until the GPU was hotter freshly microwaved do-nut jam.

Instead, we now load up a level of Crysis in DirectX10 mode at 1,920 x 1,200 with 4x AA and leave it as it stands, resisting the urge to jump on and start playing. Every time the character is killed, the game loads again automatically from the last save point and the cycle is repeated. This process more accurately replicates the peaks and troughs of a gaming session. We leave it for an hour until the temperature has stabilised and then compare the delta Ts.



As Nvidia has never sent us a reference design for the GTS 250, possibly a result Tim's Blog about the GTS 250 TipeXXX Edition that we manufactured ourselves in the labs. Fortunately though, local London retailer YoYo Tech were kind enough to lend us one so we could compare the results from the Galaxy to a reference card.

Overclocking