SIMT Execution and Divergence Simulation

Write a C program that simulates SIMT execution with divergence masks.

Requirements:

• Define WARP_SIZE as 32 and create an execution mask (32 booleans)
• Initialize threads with values 0..31
• Simulate a conditional: if (thread_id % 4 == 0) { result = thread_id * 2; } else { result = thread_id + 100; }
• Show the execution in two phases with masks: Phase 1: threads where id % 4 == 0 are active (mask bits set), others masked off Phase 2: threads where id % 4 != 0 are active, first group masked off
• Calculate and print the divergence penalty: 2 serialized paths = 2x instructions issued
• Simulate a loop where some threads exit early (e.g., while (value < 20)), showing how early-exiting threads remain masked but the warp continues until all threads finish
• Print the active mask for each loop iteration
• Explain why SIMD (CPU vector) does not have this divergence problem in the same way