ATI Dynamic Brancing Demo
Posted by: Newsfactory on: 07/02/2004 10:24 AM [ Print | 7 comment(s) ] · 9241 views
Just got word that Humus has finished up his New dynamic brancing demo!
Nvidia can consider themselves pwned. Here's a demo that does dynamic branching, without the need for pixel shader 3.0, and still receives the huge performance boost that pixel shader 3.0 dynamic branching supposedly gives when utilized in a similar fashion.
Requires:
Pixel shader 2.0 Vertex shader 1.1
Please Note: Dynamic Brancing runs on by default on Radeon X800 cards. On Radeon 9500 and up, and on GeForce 5200 and up, The result is that early-out speeds things up considerably. In this demo the speed-up is generally in the range of two to four times as fast than without early-out. And with more complex lighting, the gain would have been even larger.
Dynamic Brancing Demo
Nvidia can consider themselves pwned. Here's a demo that does dynamic branching, without the need for pixel shader 3.0, and still receives the huge performance boost that pixel shader 3.0 dynamic branching supposedly gives when utilized in a similar fashion.
Requires:
Pixel shader 2.0 Vertex shader 1.1
Please Note: Dynamic Brancing runs on by default on Radeon X800 cards. On Radeon 9500 and up, and on GeForce 5200 and up, The result is that early-out speeds things up considerably. In this demo the speed-up is generally in the range of two to four times as fast than without early-out. And with more complex lighting, the gain would have been even larger.
Dynamic Brancing Demo
« Intel Desktop Control Center 1.2.0059 · ATI Dynamic Brancing Demo
· Battle of the 2.4 GHz Hammers - AMD?s CPU Sockets Compared »
Comment
|
hinkle Unregistered |
#54721 Posted on: 07/02/2004 04:41 PM
Too bad that this isn't really dynamic branching and that it cannot be used in games like FarCry because this method is too limited. |
Comment
|
The_truth Unregistered |
#54724 Posted on: 07/02/2004 05:26 PM
It will be used in games, its being coded right now. |
Comment
|
hinkle Unregistered |
#54725 Posted on: 07/02/2004 06:01 PM
sorry, don't believe ya and of course you can't name any game which will include this little piece of code? NDA and all that 
since it doesn't seem to be working with nvidia cards, what good would this hack bring? |
Comment
|
digitalwanderer Unregistered Posts: 3 Joined: 2004-02-22 |
#54727 Posted on: 07/02/2004 07:49 PM
Er, uhm...Humus is a coder for ATi now, they'll be using this technique with developers to improve ATi's performance. The reason it doesn't help performance for the nVidia cards is because of the way they handle stencil operations in the pipeline. There is a good explanation of it in the thread at Beyond3D. (Around page 7 I think) This is a real technique that really does do the same thing as dynamic branching, but apparently a bit quicker and in SM2.0.
Nice one Humus, you never cease to amaze me. |
Comment
|
The_truth Unregistered |
#54729 Posted on: 07/02/2004 08:27 PM
Not working on one IHVs cards wouldnt stop it being used, after all DX9 didnt work properly on nvidias last product range, yet we still have DX9 games! SM3.0 isnt supported by ATI but we will have SM3.0 in games, so i dont hink you understand how game coding works... |
Comment
|
hinkle Unregistered |
#54738 Posted on: 07/03/2004 02:54 AM
Check again at page 12. Humus method does work with nvidia cards. Of course ATI is already using this technique... too bad that Humus just stated now that he did not even talk about it with ATI
But I sincerely expect valve to delay HL2 another year just to implement this "new" technique. |
Comment
|
hinkle Unregistered |
#54739 Posted on: 07/03/2004 02:56 AM
Your last sentence makes no sense whatsoever but I agree... just to make you happy |
