You're running a logistics company. Trucks wait 6 hours at loading docks. Where's the parallelism in that system?
You're asking the wrong question — your 6-hour wait isn't a parallelization problem, it's a serial bottleneck you engineered into the system.
Six hours. That's not a bottleneck, that's a parking lot. When I bet NVIDIA on CUDA in 2006, everyone thought I'd lost my mind. Gaming chips? For general computing? But I saw what they missed: the future isn't serial, it's parallel. And I ...