HIS vs EVGA Midrange Gaming Face Off Review

Midrange Gaming Shootout

Metro 2033

The game based on the book by Dmitry Glukhovsky is part first person shooter, part survival horror, without really being either. It’s a hugely frustrating game for us because it’s so reliant upon the lighting effects that the frame rates never really reaches playable on any hardware, yet it doesn’t look so stunning that you’re willing to put up with the shoddy frame-rate just to gawp. Even the much maligned (by me) Crysis looked good enough we put up with it’s woeful performance. 

Single GPU – Stock Speeds

Our first big “huh” moment. Advanced PhysX enabled and yet the GTX460s lag massively behind the Radeons. Clearly the PhysX is so advanced and puts such a strain on the card that it hasn’t enough puff left to do the graphics. Whereas the Radeons dump the physics onto our Core i7-950 and therefore can at least have a good go at providing something more than a slide-show.

 

Single GPU – Overclocked

The incredibly variable nature of the Metro 2033 engine stands out here with both the GTX460s gaining 3 frames or so from the overclock to their GPU, yet the HD6870 loses frames. It is so frustrating. Hopefully our twin GPU testing will iron out the rough edges.

 

Dual GPU – Stock

Yes and no. Your second 1GB GTX460 is worth a piddling 5FPS, and yet the second 768MB GTX460 gives a 14FPS boost. If only getting 5FPS extra from an SLI 1GB GTX460 is strange, then the Crossfire HD6850 gives over 100% improvement. There are so many things to try and fathom about this game it should be based on a Jules Verne book.

 

Dual GPU – Overclocked

Finally we get a result remotely approximating what we’d expect on a theoretical basis when compared to the single stock card. The big GTX460 finally gets back ahead of its sibling, the HD6850 doesn’t really benefit from the overclock and the HD6870 is quite a beast. All is once again as it should be in the world.

Even if a 4GHz 950 and two overclocked HD6870s don’t manage to get Metro 2033 past 60FPS average.