The Decisive Moments
The Decisive Moments
블로그 | RSS구독 | 관리자 |  
oranzePerformance & Practicalities #01
<< Contents >>
─────────────────────────────────────
1. Resource Management
2. ID3DXEffect & Rendering Sequence
3. Batching IO Buffers & Textures
4. Shader Instructions

<< 1. Resource Management >>
─────────────────────────────────────
Device Lost, When It Occurs?
- By Power Management System
- Switch to H/W Rasterizer to S/W Rasterizer
- Choose Different Video Adapter on The System
- Change Desktop Display Mode
Device Probe/Resource Manager
Restoration of Dynamic Rsources of VB/IB/Textures

<< 2. ID3DXEffect & Rendering Sequence >>
─────────────────────────────────────
ID3DXEffectStateManager for Pure Device & None-Pure Device
Rendering Sequence
- User Defiend Priority
- ZR/ZW/AT/AB/D/1DM/2DM/3DM
Determining Sort Ordering
- Current: FX -> VBDECL -> VB -> IB -> TEXTURE
- Preset Sort Functors: FX -> TEXTURE -> VBDECL -> VB -> IB, ...
- Registable User Defined Sort Functor

<< 3. Batching IO Buffers & Textures >>
─────────────────────────────────────
Batching Small IO Buffers
Batching Small Textures (Especially for Tiny Light Maps)
Multi-Streaming: G + T + A + I
Real-Time VB Stream Processor

<< 4. Shader Instructions >>
─────────────────────────────────────
IDirect3DQuery9
Performances of Shader Instructions
- Common: add, mad, mul, dp3, dp4
- Sparse: cmp, rsq, rcp, frc
- Macros: nrm, pow, lrp, crs, sincos
Arithmetic VS Texture
- To Use a Less Accurate Approximation of The Function
- Precision Modifier: _pp
Pitfalls & Rumors
- ps_2_a, mad, texkill, _pp
─────────────────────────────────────

<< References >>
─────────────────────────────────────
ShaderX3 2.6 Improved Batching via Texture Atlases
ShaderX3 3.1 Optimizing DX9 Vertex Shaders for Software Vertex Processing
ShaderX3 6.1 An Extensible Direct3D Resource Management System
ShaderX3 6.5 Drawing a Crowd
ShaderX3 7.1 In-Depth Performance Analyses of DX9 Shading Hardware
ShaderX4 5.1 Tips and Tricks for D3DX Effects-Based Renderers
ShaderX4 8.6 GPU Performance of DirectX 9 Per-Fragment Operations Revisited
GPUGems1 C28 Graphics Pipeline Performance
GPUGems2 C03 Inside Geometry Instancing
GPUGems2 C05 Optimizing Resource Management with Multistreaming
GPUGems2 C46 Improved GPU Sorting
─────────────────────────────────────
이 글의 관련글(트랙백) 주소 :: http://oranze.com/tt/rserver.php?mode=tb&sl=20
이름 ::   비밀번호 ::  
홈페이지 ::  
비밀글로 등록




[이전 목록]   [1] ... [10][11][12][13][14][15][16][17][18] ... [31]   [다음 목록]
 
▒ 검색
▒ 카테고리
분류 전체보기
Free Talk
▒ 글 보관함
▒ 달력
<<   2024 Oct   >>
S M T W T F S
293012345
6789101112
13141516171819
20212223242526
272829303112
▒ 최근에 올라온 글
▒ 최근에 달린 댓글
▒ 최근에 받은 트랙백
貢籃膽뺏,google탤츰
貢籃膽뺏,google탤츰
▒ 링크 사이트