Results Analysis

With the GTX 465 retailing at around £15 more than a stock HD 5850, the card needed to get off to a good start. Unfortunately, the GTX 465 failed to beat a stock HD 5850 in any of our Dirt 2 tests. At 1,680 x 1,050, which is the resolution of a 20in or 22in monitor that this kind of card is likely to be paired with, the GTX 465 was 4-6fps slower when comparing the minimum frame rates. The same kind of gap was apparent at every resolution - while both the GTX 465 and HD 5850 can play Dirt 2 in all it glory even at 2,560 x 1,600 with 4x AA, the HD 5850 is clearly the faster card in this game.

STALKER: Call of Pripyat has always favoured ATI cards, and given the results of our Dirt 2 testing we weren’t expecting an upset with the GTX 465. Again we saw a clear lead from the HD 5850, with minimum frame rates 5-8fps higher than those of the GTX 465. Notably this meant that the game was smooth and playable on a HD 5850 even at 2,560 x 1,600 but was a bit suttery on the GTX 465 with its minimum of 19fps.


Crysis really sorts the men from the boys, despite being nearly 3 years old. However, we were disappointed when we saw a minimum frame rate of 17fps at 1,680 x 1,050 (with no AA) compared to the 21fps of the HD 5850, and were even more so when the HD 5850 was faster when we did apply 4x AA. No matter what resolution or AA level we used, the GTX 465 was never faster than the HD 5850.

This was surprising, as typically Fermi cards hold their performance much better than HD 5000-series cards when applying AA in Crysis, but the GTX 465 was just too weak, likely because of the low stream processor count and the skinnier memory interface. It's also worth noting that neither card could play Crysis in DX10 mode with Very High settings even at 1,680 x 1,050 with no AA - you'll have to make image quality sacrifices to get a playable frame rate no matter which card you use.

The GTX 465 showed some signs of superior speed over the HD 5850 in Bad Company 2, with faster frame rates at 1,920 x 1,200 and 2,560 x 1,200 when using 4x AA. In both cases, the extra speed of the GTX 465 was significant - at 1,920 x 1,200 the game ran at a much smoother 27fps minimum rather than 24fps. Both cards could play the game comfortably at the key resolution of 1,680 x 1,050, however, with the HD 5850 wining out by a noticeable 8fps margin when no AA was used, and managing a higher average frame rate when we did use 4x AA (when both cards posted identical minimum frame rates of 33fps).


Power consumption was broadly in line with what we expected: our test system drew 159W from the wall while idle and 305W while under load with the GTX 465 installed, which was a significant 68W more than when we used a HD 5850.

We were pleased to find the cooler for the Asus GTX 465 much quieter than that of the GTX 470, and it did an acceptable job of keeping the GPU cool. Our thermal testing showed delta Ts of 24°C at idle and 56°C when loaded heavily.

We couldn’t let the Asus GTX 465 get away without being overclocked, so we installed the bundled Smart Doctor program and set about seeing what we could do. Unfortunately, Smart Doctor wasn't able to raise the clock speeds of our GTX 465, despite the program thinking it had. In the end we installed the excellent Zotac Firestorm overclocking tool which had no problem giving the card a much needed boost.

After an afternoon of tinkering (it’s a hard life) we settled on a core speed of 765MHz (up from 607MHz), and pushed the stream processor clock from 1,215MHz to1,530MHz, and the memory from 3.2GHz to 3.46GHz. This represented an impressive 26 per cent overclock to the GPU, which gave us an increase of 2-3fps in Crysis.


This boost was enough to push the GTX 465 past the HD 5850 where previously the two were tied, though we did have to force the cooler’s fan to 80% speed to keep the overclock stable which made the card very loud; we wouldn’t want to live with the card at these settings. It’s nice to know there is a bit more in the tank if you need it though.

Conclusion

Whichever way you cut it, the GTX 465 is a disappointment. This Asus card is pretty much the bog-standard model, with other manufacturers using custom coolers and PCBs rather than modding a GTX 470 card. However, while the card is £15 more expensive than a HD 5850, it's noticeably slower in the vast majority of our tests. Only at specific combinations of resolution and AA setting does the GTX 465 make any case for choosing it over a HD 5850. To exacerbate the problem, the GTX 465 card is more power hungry and produces more waste heat than its ATI rival.

Some heavy overclocking solved the speed issue to an extent, but re-introduced the problem of noise - to keep the overclocked GPU cool we had to set the fan to blow fast and loud - and there's no guarantee that every card will overclock as well as ours did. While Asus has done what it can with this card, Nvidia's GTX 465 GPU is too expensive or too slow (or possibly both) and until that changes, isn't worth looking at. We'll be keeping our eyes peeled for the forthcoming GTX 460 for a genuine mid-range GeForce card.

• Performance
• x
• x
• x
• x
• x
• x
• x
• -
• -
• -
• 7/10

• Features
• x
• x
• x
• x
• x
• x
• x
• -
• -
• -
• 7/10

• Value
• x
• x
• x
• x
• x
• -
• -
• -
• -
• -
• 5/10

• Overall
• x
• x
• x
• x
• x
• x
• -
• -
• -
• -
• 6/10

Score Guide