Also in games that were originally intended to be released under DX10.1 hardware it outperforms the GTX280. These are games like Unreal 3, Assassin's Creed, GRID, and Bioshock before nVidia forced them to do operation short-bus patch to unoptimize it for AMD video cards.
Except for the fact it isn't. I did say the GTX280 is the best. But the HD4870 competes which it does. Review Also in games that were originally intended to be released under DX10.1 hardware it outperforms the GTX280. These are games like Unreal 3, Assassin's Creed, GRID, and Bioshock before nVidia forced them to do operation short-bus patch to unoptimize it for AMD video cards.
Nope, its 6% better then GTX 260 w/o AA. GTX 260 performes 10% better then 4870 under AA.
Also, only DX10.1 game from those you listed is Assassin's Creed. And I agree theres something shady going on there.
The reason I make a point about DX10.1 is because of the performance improvements on it. In the case of Assassins Creed before its nVidia support patch it was a 20% performance increase due to a function that renders Anti-aliasing more effeciently. I would take that Toms Hardware review with a grain of salt. The reason I used it originally is because Toms notoriously Intel/nVidia biased and even they agree on the HD4870's performance. But notice how in the subsequent reviews the performance on the HD4870 dropped? This is because they are using a platform that supports nVidia, with old Catalyst drivers that were made before the HD4xxx cards were released.
A much better platform for testing GPUs is on an AMD system due to its memory controller, and supporting both GPU chipsets. HyperTransport is very important with GPUs as it was shown that even the Quad Core Extreme was a slight bottle-neck due to data running through the FSB for the rescent generation of cards. Also like I said it can be tested on the GPUs native chipsets with both manufacturers having bad relations with Intel.
Still the reason I like the AMD offering is because it can play any game at a playable framerate on the highest resolutions and settings (aside from Crysis 30" that none can play at over 30fps without dual GPUs), and it offers technology that will be utilized in the future without the extra cost. That being its Tesselator, Crossfire integrated controller, and Stream Processors capable of calculating 4x as much information. They also have the hardware on them implemented that was what was originally suppose to make DX10 require a new OS iteration to implement with a performance increase and more effecient rendering process. However, MS scratched it because nVidia failed to offer this and now DX10 is pretty much an upgraded version of DX9.0c
Comments
Its the GTX280. However, the HD4870 is incredibly close in performance and above it in games that support DX10.1 at much less the cost.
This is a lie dude.
Apsolute best single chip is GTX280.
HD4870 is 10% SLOWER then GTX 260 (not 280 but 260)
Best card on the market is HD4870X2 fallowed by 9800GX2, but those are dual chip cards.
Except for the fact it isn't. I did say the GTX280 is the best. But the HD4870 competes which it does.
Review
Also in games that were originally intended to be released under DX10.1 hardware it outperforms the GTX280. These are games like Unreal 3, Assassin's Creed, GRID, and Bioshock before nVidia forced them to do operation short-bus patch to unoptimize it for AMD video cards.
Nope, its 6% better then GTX 260 w/o AA. GTX 260 performes 10% better then 4870 under AA.
Also, only DX10.1 game from those you listed is Assassin's Creed. And I agree theres something shady going on there.
http://www.tomshardware.com/reviews/radeon-geforce-comparison,2007-35.html
The reason I make a point about DX10.1 is because of the performance improvements on it. In the case of Assassins Creed before its nVidia support patch it was a 20% performance increase due to a function that renders Anti-aliasing more effeciently. I would take that Toms Hardware review with a grain of salt. The reason I used it originally is because Toms notoriously Intel/nVidia biased and even they agree on the HD4870's performance. But notice how in the subsequent reviews the performance on the HD4870 dropped? This is because they are using a platform that supports nVidia, with old Catalyst drivers that were made before the HD4xxx cards were released.
A much better platform for testing GPUs is on an AMD system due to its memory controller, and supporting both GPU chipsets. HyperTransport is very important with GPUs as it was shown that even the Quad Core Extreme was a slight bottle-neck due to data running through the FSB for the rescent generation of cards. Also like I said it can be tested on the GPUs native chipsets with both manufacturers having bad relations with Intel.
Still the reason I like the AMD offering is because it can play any game at a playable framerate on the highest resolutions and settings (aside from Crysis 30" that none can play at over 30fps without dual GPUs), and it offers technology that will be utilized in the future without the extra cost. That being its Tesselator, Crossfire integrated controller, and Stream Processors capable of calculating 4x as much information. They also have the hardware on them implemented that was what was originally suppose to make DX10 require a new OS iteration to implement with a performance increase and more effecient rendering process. However, MS scratched it because nVidia failed to offer this and now DX10 is pretty much an upgraded version of DX9.0c
Also the HD4xxx series processes AA beautifully, and infact better then nVidia offerings. Where these cards have pulled ahead is in benchmarks with AA on, and at higher resolutions.