Nvidia GeForce GTX 465: Positioning
We covered briefly the key technical aspects of the GeForce GTX 465 in our preview and found a lot of factors to be worried about. Our apprehensions were eventually confirmed in our practical tests. Now we want to compare the new card once again in more detail with the senior models from the same series as well as with its predecessors and market opponents.
There are indeed few reasons for us to be optimistic about it. The new card is inferior to the Radeon HD 5850 in terms of fill rate and texture sampling rate. In fact, the latter card is twice as fast as the GeForce GTX 465 at texture sampling although their official price differs but slightly ($299 and $279, respectively). This may have been an oversight of the developers or, perhaps, they just had no other way to introduce a successor to the GeForce GTX 275 but the latter is about twice as fast as the GeForce GTX 465 at processing textures, too. The difference in fill rate is not so huge, yet the GeForce GTX 465 cannot be expected to deliver high performance at high resolutions considering its fill rate of below 20 gigapixels per second and low memory bandwidth. We have already made sure of this in our preview. In fact, releasing a new-generation graphics card with lower key specs than those of the previous-generation card from the same category is an alarming symptom that confirms Nvidia’s having found itself in a trap with the GF100 chip. The company has had to lower the performance of the new GPU too much to create a GF100-based graphics card comparable to the Radeon HD 5850 in price.
We guess the developers should not have reduced the number of stream multiprocessors to 11, especially as they did not manage to increase the core clock rate over the GeForce GTX 470’s. 12 active multiprocessors would provide a total of 384 ALUs and 48 texture-mapping units, which would make a difference for the GeForce GTX 465. Instead, the new card’s GPU has been cut down too much. Lacking a lot of computing and texture-mapping resources, the GeForce GTX 465 cannot be as fast in modern games as its price suggests. Of course, the GF100 has advanced geometry processing features including cutting-edge tessellation blocks which are obviously superior to what the ATI Radeon HD series currently has to offer, but the history of GPUs shows that focusing on one technology at the expense of others has never been a successful development strategy. This strategy won’t work with the GeForce GTX 465, either, especially as game developers do not hurry to give up nice-looking textures and gamers, to give up playing at high resolutions. A decent mainstream graphics card should be a well-balanced product, but this is something that the Radeon HD 5850 has and the GeForce GTX 465 has not, being inferior to previous-generation cards in some parameters.
At the moment of its announcement the new card has a recommended price of $279 but we still think that the GeForce GTX 465 should cost less in order to succeed on the market. A lower price would make it more appealing for users but less profitable for Nvidia because any graphics card based on the current 40nm revision of the GF100 chip is expensive to make and can hardly be sold at $200 or something for a profit. This may only make sense if Nvidia wants to utilize a large number of defective GF100 chips which would have to be wasted otherwise. Besides, Nvidia might win some time until it has a new inexpensive GPU that might replace the GF100. We’ve got some information that it is for such a card that the name “GeForce GTX 460” has been reserved. So, the GeForce GTX 465 looks like a temporary solution.
It’s time now to have a look at the graphics card itself. We’ve got an off-the-shelf version of it now instead of the presale sample we benchmarked in our preview. It is called Zotac GeForce GTX 465.