ThreadPool implementation in C is at same performance or slower when compared to non parallelized code I've been working on implementing a ThreadPool in C to optimize the processing of my program, which involves various tasks related to neural network linear algebra. However, I'm encountering performance issues when using the ThreadPool. Surprisingly, the ThreadPool is not providing the expected speedup and, in some cases, is even slower than sequential processing. To investigate this, I conducted a comparison with CUDA, and the CUDA implementation shows a significant speedup. However, I'm...