Good performance for the price; true sub-$300 Fermi card.
Draws more power; not quite as robust a performer as the HD 5850.
In the raging battle between AMD and Nvidia over DirectX 11 supremacy, AMD has had a decided edge in price/performance ratios, if not raw performance. Now, Nvidia aims to rectify that with the GTX 465.
Like the GTX 470 (and even the GTX 480, for that matter), the GTX 465 uses the same Fermi chip, with key functional units disabled. This may be by choice or because of yield issues, given the massive size of Nvidia’s latest progeny. Whatever the case, it allows Nvidia to bring a card to market that’s generally priced just a little less than AMD’s sweet-spot card, the Radeon HD 5850. We’ve seen prices for the PNY card at around $280, as opposed to an average price ranging from $290 to $300 for the HD 5850.
The PNY XLR8 GTX 465 is the first sub-$300 Fermi-based card.
So, what does the 465 give up relative to its bigger siblings, the GTX 475 and GTX 480? While there are indeed fewer shader cores (see chart), which implies lower performance in shader-heavy applications, game performance is also likely to be affected by the reduced number of ROPs in the final output stage. There are also fewer texture units, but that’s actually balanced by the ROPs, and by the smaller memory interface.
How does that play out in actual performance? To find out, we compared the GTX 465 to a stock (not overclocked) Radeon HD 5850. Both cards are running the latest drivers at the time of the review: version 257.15 for Nvidia and Catalyst 10.4 for AMD. The XFX card comes in at around $295, while the PNY 465 GTX can be found for $280.
The result is a wash, with both cards winning some and losing some—although it’s worth noting that when the Radeon HD 5850 wins, it wins by fairly large margins, while the GTX 465 ekes out just marginal wins when it pulls ahead.
Some of this, of course, is due to drivers. Fermi drivers are still relatively immature, and each driver release from Nvidia has produced notable performance increases. Power usage on the GTX 465 was still higher at full throttle, though the gap is narrowing. So the GTX 465 won’t eat power supplies for lunch.
In the end, however, the GTX 465 is still a cut-down version of the GTX 480 (which in itself is somewhat crippled). It’s great that Nvidia has brought the price down to a more affordable level, but we’re looking forward to the next set of Nvidia GPUs, which will be built from the ground up for midrange and budget-class cards. That will tell us how effective Fermi really is as a scalable architecture.
In the end, if you buy your midrange card based on flipping a Maximum PC coin or simple brand loyalty, you’ll be fairly satisfied either way.
|GTX 465 ||GTX 470 ||GTX 480|
|Shader Cores ||352 ||448 ||480|
|Texture Units ||44 ||56 ||60|
|Core Clock ||607MHz ||607MHz ||700MHz|
|Shader Clock||1,215MHz ||1,215MHz ||1,401MHz|
|GDDR5 Memory Clock||802MHz||837MHz||924MHz|
|Frame Buffer Size ||1,024MB ||1,280MB ||1,536MB|
|Memory Interface ||256-bit ||320-bit ||384-bit|
|PNY XLR8 GTX 465 ||XFX Radeon HD 5850 |
|Unigine Heaven 2.0 (fps) ||19 ||14 |
|Battle Forge (fps) ||40 ||40 |
|Dirt 2 (fps)||57||62|
|Far Cry 2 / Long (fps) ||66 ||65 |
|Far Cry 2 / Action (fps) ||56 ||54 |
|Tom Clancy's HAWX (fps)||69||76|
|Crysis (fps) ||19 ||27 |
|DX11 Aliens vs. Predator (fps)||20||25|
|Just Cause 2 Concrete Jungle (fps)||31||32|
|STALKER: Call of Pripyat (fps)||26||30|
|Power Usage (Idle)||146W||142W|
|Power Usage (Full Throttle) ||285W ||261W |
Best scores are bolded. Our test bed is a 3.33GHz Core i7-975 Extreme Edition in an Asus P6X58D Premium motherboard with 6GB of DDR3/1333 and an 850TX Corsair PSU. The OS is 64-bit Windows 7 Ultimate. All games are run at 1920x1200 with 4x AA.