High performance SIMD marching cubes isosurface extraction on commodity computers

被引：12

作者：

Newman, TS ^{[1
]}

Byrd, JB ^{[1
]}

Emani, P ^{[1
]}

Narayanan, A ^{[1
]}

Dastmalchi, A ^{[1
]}

机构：

[1] Univ Alabama, Dept Comp Sci, Huntsville, AL 35899 USA

来源：

COMPUTERS & GRAPHICS-UK | 2004年 / 28卷 / 02期

基金：

美国国家科学基金会;

关键词：

isosurface extraction; Marching Cubes; SIMD; vector-parallelism; microcomputers;

D O I：

10.1016/j.cag.2003.12.008

中图分类号：

TP31 [计算机软件];

学科分类号：

081202 ; 0835 ;

摘要：

Approaches to achieve high performance for volume visualization via the popular Marching Cubes algorithm on commodity x86 personal computers are described. The approaches utilize Single Instruction Multiple Data (SIMD) Within A Register (SWAR) capabilities available in the streaming media extensions of recent-generation x86 CPUs. The SWAR capabilities are exploited in a manner that enables simultaneous processing of multiple volume data points, edges, and cells. In addition, the approach maintains the rendering quality of a standard serial implementation of the Marching Cubes. An average overall speedup of nearly four times is achieved (compared to an unoptimized standard implementation that uses conventional serial processing). (C) 2003 Elsevier Ltd. All rights reserved.

引用

页码：213 / 233

页数：21

共 46 条

[41]

Talla D, 1999, 42ND MIDWEST SYMPOSIUM ON CIRCUITS AND SYSTEMS, PROCEEDINGS, VOLS 1 AND 2, P18

[42]

Thakkar S, 1999, INTEL TECHNOLOGY J Q, VQ2, P1

[43]

TUNG YS, 1999, P 1999 IEEE INT C MU

[44] OCTREES FOR FASTER ISOSURFACE GENERATION [J].

WILHELMS, J ;

VANGELDER, A .

ACM TRANSACTIONS ON GRAPHICS, 1992, 11 (03) :201-227

[45] Efficient parallel out-of-core isosurface extraction [J].

Zhang, HJ ;

Newman, TS .

PVG 2003 PROCEEDINGS, 2003, :9-16

[46] Scalable isosurface visualization of massive datasets on COTS* clusters [J].

Zhang, XY ;

Bajaj, C ;

Blanke, W .

IEEE 2001 SYMPOSIUM ON PARALLEL AND LARGE-DATA VISUALIZATION AND GRAPHICS, PROCEEDINGS, 2001, :51-58

← 1 2 3 4 5 →