7.2. Principles of Parallel Processing Architectures

Try these first

1. 1.

   Define SIMD.

   Hint

   Think of a single process applied to many items at once.

2. 2.

   What is task parallelism?

   Hint

   Consider what happens when multiple jobs are done at the same time.

3. 3.

   What does SIMD stand for?

   • Single Instruction
   • Multiple Data
   • Single Instruction
   • Multiple Devices
   • Simultaneous Instruction
   • Multiple Data

   Hint

   Focus on how the name indicates the relation between instructions and data.

4. 4.

   True or False: MIMD allows different processors to perform the same instruction on different data.

   • True
   • False

   Hint

   Think of the flexibility of MIMD compared to SIMD.

5. 5.

   Evaluate a parallel processing architecture for an AI task requiring simultaneous image and text analysis. Which model (SIMD or MIMD) would you choose and why?

   Hint

   Consider the flexibility needed to process different types of instructions.

6. 6.

   Design a scenario using data parallelism to reduce computation time for a deep learning model. Specify how tasks would be divided among processors.

   Hint

   Think about dividing large datasets into smaller, manageable blocks for processing.