研究目的
Investigating the performance and power consumption of several different numeric algorithms implemented on iPhone device, comparing CPU and GPU based solutions in the context of medical image processing applications.
研究成果
GPU based implementations are significantly faster than CPU based ones for complex image processing tasks, with similar power consumption in some cases. The choice between GPU and CPU based solutions should consider the specific requirements of the application, especially in medical contexts where efficiency and battery life are crucial.
研究不足
The study is limited to iPhone devices and specific image processing tasks. The power measurement method requires specific setup and does not account for all possible variables affecting power consumption.
1：Experimental Design and Method Selection:
The study compares GPU and CPU based image processing algorithms on an iPhone 8 Plus with iOS
2：4, focusing on matrix multiplication, thresholding, and Canny’s edge detection. Sample Selection and Data Sources:
Test images of surgical equipment with varying detail levels were used.
3：List of Experimental Equipment and Materials:
iPhone 8 Plus, iOS
4：4, GPUImage2, OpenCV. Experimental Procedures and Operational Workflow:
Each test scenario was executed in a loop for 10,000 times to measure efficiency, and power consumption was logged over 10 minutes.
5：Data Analysis Methods:
Performance was measured by execution time, and power consumption was logged using sysdiagnose tool to analyze battery voltage, instant amperage, and temperature.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
