Improving register allocation for subscripted variables
PLDI '90 Proceedings of the ACM SIGPLAN 1990 conference on Programming language design and implementation
Register promotion in C programs
Proceedings of the ACM SIGPLAN 1997 conference on Programming language design and implementation
Stereo Matching Using Belief Propagation
IEEE Transactions on Pattern Analysis and Machine Intelligence
A Taxonomy and Evaluation of Dense Two-Frame Stereo Correspondence Algorithms
SMBV '01 Proceedings of the IEEE Workshop on Stereo and Multi-Baseline Vision (SMBV'01)
Belief Propagation on the GPU for Stereo Vision
CRV '06 Proceedings of the The 3rd Canadian Conference on Computer and Robot Vision
Optimization principles and application performance evaluation of a multithreaded GPU using CUDA
Proceedings of the 13th ACM SIGPLAN Symposium on Principles and practice of parallel programming
Stencil computation optimization and auto-tuning on state-of-the-art multicore architectures
Proceedings of the 2008 ACM/IEEE conference on Supercomputing
A Note on Auto-tuning GEMM for GPUs
ICCS '09 Proceedings of the 9th International Conference on Computational Science: Part I
A cross-input adaptive framework for GPU program optimizations
IPDPS '09 Proceedings of the 2009 IEEE International Symposium on Parallel&Distributed Processing
Auto-tuning 3-D FFT library for CUDA GPUs
Proceedings of the Conference on High Performance Computing Networking, Storage and Analysis
Belief Propagation Implementation Using CUDA on an NVIDIA GTX 280
AI '09 Proceedings of the 22nd Australasian Joint Conference on Advances in Artificial Intelligence
High-accuracy stereo depth maps using structured light
CVPR'03 Proceedings of the 2003 IEEE computer society conference on Computer vision and pattern recognition
Automatic restructuring of GPU kernels for exploiting inter-thread data locality
CC'12 Proceedings of the 21st international conference on Compiler Construction
| Hi-index | 0.00 |
A CUDA kernel will utilize high-latency local memory for storage when there are not enough registers to hold the required data or if the data is an array that is accessed using a variable index within a loop. However, accesses from local memory take longer than accesses from registers and shared memory, so it is desirable to minimize the use of local memory. This paper contains an analysis of strategies used to reduce the use of local memory in a CUDA implementation of belief propagation for stereo processing. We perform experiments using registers as well as shared memory as alternate locations for data initially placed in local memory, and then develop a hybrid implementation that allows the programmer to store an adjustable amount of data in shared, register, and local memory. We show results of running our optimized implementations on two different stereo sets and across three generations of nVidia GPUs, and introduce an auto-tuning implementation that generates an optimized belief propagation implementation on any input stereo set on any CUDA-capable GPU.