Number of Cores vs Number of Threads


I was wondering if i have to chose between two processors; the first one has more physical cores than the second one but the second one has more threads which would be more suitable for deep learning? I am currently working on some experiments in reinforcement learning and after training(which is done on GPU), the network plays against itself, which is done mainly(if not entirely) on the CPU.

For parallelization purposes on CPU for ML tasks, you can think of logical cores exactly as you would physical cores – I.E, an intel with 6 physical and 12 logical effectively has 12 cores.