How to evaluate if my design/matlab code/algorithm can be dumped in controller or processor or fpga or gpu?

I would like to know how can I see what can i use for my application for example:

1. i am trying to do take image and do the segmentation using blob analysis to identify a object in an image or video. how to determine whether i would need a controller or microprocessor or fpga or gpu ?

Answer 1

[Shubham]

Hi Ashokraj,

To determine the appropriate hardware for image segmentation using blob analysis, consider the following:

Performance:

• Microcontroller: Low performance, suitable for simple tasks.
• Microprocessor: Moderate performance, good for general image processing.
• FPGA: High performance with real-time capabilities.
• GPU: Very high performance, ideal for complex tasks and deep learning.

Power Consumption:

• Microcontroller: Low.
• Microprocessor: Moderate.
• FPGA: Moderate to high.
• GPU: High.

Development Complexity:

• Microcontroller: Low.
• Microprocessor: Moderate.
• FPGA: High.
• GPU: High.

Cost:

• Microcontroller: Low.
• Microprocessor: Moderate.
• FPGA: High.
• GPU: Very high.

Recommendation:

• Microcontroller: For simple, low-power, cost-sensitive applications.
• Microprocessor: For moderate processing needs with balanced cost and complexity.
• FPGA: For high-performance, real-time processing with low latency.
• GPU: For very high-performance needs, complex tasks, and deep learning, if power and cost are manageable.

Choose based on your application's specific performance, power, complexity, and cost requirements.

[Ashokraj]

Thanks @Shubham, Is there any way i can simulate that in matlab to check the performance, complexity? For example there is a profiler in matlab which gives us timing. In order to implement this code we need to run the clock at x Hz, we need to have this X of RAM and X of ROM/Flash,etc?