iMac 24" Mac OS X 10.6.1
2.93GHz Intel Core 2 Duo Processor
4GB 1067MHz DDR3
NVIDIA GeForce GT 120
Parallels Desktop for Mac, build 4.0.3846
Lower is better, 그래프를 그려보려 했으나 급 귀찮아짐;ㅋㅋㅋ
참고로 내 Samsung NC10은 2.2219 / 1.0652 / 0.7304 / 1.2767 / 2.6096 / 1.6822
Conclusion :
- Floating point operations(LU and FFT) on a native MATLAB were 5~10x faster than that of the parallels virtual machine.
- Irregular memory access(FFT) was even more faster in native MATLAB.
- File access(ODE) showed somewhat similar performance, actually the virtual machine was faster.
- Integer operations (Sparse) in virtual machine was faster than floating point operations but still 3x slower than native.
- Graphics was not a big advantage being native, MATLAB on bootcamp xp might be better, but not installed now.
2.93GHz Intel Core 2 Duo Processor
4GB 1067MHz DDR3
NVIDIA GeForce GT 120
Parallels Desktop for Mac, build 4.0.3846
Lower is better, 그래프를 그려보려 했으나 급 귀찮아짐;ㅋㅋㅋ
참고로 내 Samsung NC10은 2.2219 / 1.0652 / 0.7304 / 1.2767 / 2.6096 / 1.6822
Conclusion :
- Floating point operations(LU and FFT) on a native MATLAB were 5~10x faster than that of the parallels virtual machine.
- Irregular memory access(FFT) was even more faster in native MATLAB.
- File access(ODE) showed somewhat similar performance, actually the virtual machine was faster.
- Integer operations (Sparse) in virtual machine was faster than floating point operations but still 3x slower than native.
- Graphics was not a big advantage being native, MATLAB on bootcamp xp might be better, but not installed now.
