I have a question ablout use shadowmap to implement visible function.How to choose a appropriate size for shadowmap?
I want to use shadowmap to do visible determinate,the other patchs with the shoot patch. It is the same as your mention.:-)
This Demo is compute radiosity in GPU.version 0.0015
The algorithms is:
Use three 32bit fp texture store the patchs's attribute.
Is that you made reference Local PRT?
I don't understand the means of the convolution coefficients in Loacl PRT.
I have a idea about use SH Lighting and PRT in Key Frame Motion.Compute the PRT of the vertex per-key frame,then blend dot(SH Lighting,the PRT of...
In XBOX,use XGSetVertexBufferHeader to set Texture as Vertex Buffer.
Use this method,you can render to vertex buffer(use the same data with the...
Your means that the R9700 disable the z-optimizations (early z reject, hierarchical z) when doing depth-fail rendering?
cache or mem?
The z-cache only store the z info.
The effect of the z-cache should be only in z-clear and z-test.
Why improved Z-Cache can optimize...
How many bytes in the Z-Cache?
"The optimizations in HyperZ III+ mostly affect the improved Z-Cache, which...
"This means you "merge" a multisample surface (depth/stencil) with a non multisample surface (texture) "
Why did you merge the depth/stencil...
Your means is render the thermal object uses 2 sequential dependant textures fetches ?
I have a mistake. :cry: :cry: :cry:
It is a night vision,not a infrared thermal.
"glow buffer" is the second render target?
The second render target is only contain the glowing object associated glowing texture,then do post...
"However, it's much faster to touch memory that's in the same memory page (inside the RAM, not the AGP/CPU page) than to cross pages. "
How many bytes are there in the memory page of GPU?
How many in NV25 and how many in R300?
I am sorry for my poor english.
XBOX GPU's Vertex Shader
Like PS2's VU?
I test the efficiency of degenerate triangles on R300,the result is:
first 20000 normal triangles+ 396 degenerate triangles, 515 fps...
The result of my test is:the optimized code with big tris is faster than the unoptimized code with big tris,and the optimized code with small tris...
When i reduce the tri size,only changed is the fill-rate,the throughput of T&L is not changed.The optimised code should be faster after changed...
That is a sample.
In my code,parallel strips between each PrimeVertex are 100;
"two small-poly cases" What is your means?
I am very...
Before and after reduce the size of triangles,All triangles are in screen.No clip and no cull.
I think the fill-rate of optimized code is the...
my optimize method is:
suppose the post-T&L wertex should be 15 vertexs,
30 31 32 33 34 35 36 37 38 39 40 41 42 43 44
First,the optimised code is faster.The strange thing is when i reduce the size of triangles,the unoptimised code is faster.
What is wrong?I don't...
First ,after using 20,000 triangles,I got two results,optimized code is 255 fps, and no optimized code is 177 fps.
If the method of optimized is...
The post-T&L vertex cache of XBOX GPU is FIFO,the ATI And NVIDIA is too.And the theory of optimized is all-purpose,i think. :lol: :lol: :lol:
0, 1, 1, 2, 2, 3, 3, 4, 4, 5, 5, 6, 6, 7, 7, 8, 8, 9, 9, 10, 10, 11, 11, 12, 12, 13, 13, 14, 14,15,15,0 (degenerate triangles to prime the...
Notes on the post-T&L cache
It's worth noting that tristrips automatically organize vertices so as to get a large part of the benefit of the...
render degenerate triangles,resulting in no pixels(in degenerate triangles) actually being drawn.The cost that save process shading vertex by use...
use degenerate triangles to fill post-T&L vertex cache by especial order of the vertexs.
I think the method should be efficient.
But i don't understand why get inverse result by reduce the size of triangles?
Why the no optimized code is faster than optimized code after changed the size in my project?
thanks for your reply. :idea:
The ATI's vertex cache is FIFO or LRU? :shock: :shock: :shock:
I suppose.I couldn't find any information about the ATI's vertex cache.
The NVIDIA GPU's vertex cache is FIFO,so i suppose the ATI's vertex...
the cache is FIFO,not LRU
0, 1, 1, 2, 2, 3, 3, 4, 4, 5, 5, 6, 6, 7, 7, 8, 8, 9, 9, 10, 10, 11, 11, 12, 12, 13, 13, 14, 14 ,0
my optimize method is:
sample,suppose the post-T&L wertex should be 15 vertexs,
30 31 32 33 34 35 36 37 38 39 40 41 42 43 44...
anyone can help me :?:
Why the no optimized code is faster than optimized code after changed the size?
I use the method raised by Mike Abrash(<<Xbox Vertex Performance>>).But I find some strange things.
I suppose the post-T&L...
R9000pro DX9 Catalyst 3.0
Why tell the NV25's texture cahces is "Dual Texture Caches"?
and the NV25's texture caches is 256k or 512k?
How many bytes fetch when cache...
I install the new driver,the stencil is ok :D
Thank Chalnoth and WhiningKhan. :wink:
I found the R9000 only support D3DFMT_D16,D3DFMT_D24X8 and D3DFMT_D32 DepthStencil Buffer in my pc.
I try to use D3DFMT_D24S8 or D3DFMT_D15S1 to...