Benchmarks of speed of using CUDA UVA-transfer vs DMA-controller (CUDA 5.0 and C++03)
Source code requires: CUDA >= 4.0 and C++03
Recomended: CUDA 5.0 and C++03 in MSVS 2010
Comparison bandwidth for memory transfer for blocks 128 Bytes - 128 MB: CPU-RAM -> GPU-RAM -> CPU-RAM:
- standard MPP-approach with using DMA-controller
- with GPU-Cores by using Unified Virtual Addressing
Result of benchmarks on GeForce GTX460 SE CC2.1 see in: result.txt