Processing time does not depend only on the hardware configuration but also on the processing options used and the dataset, which is defined mostly by the number of images of your project, the resolution of the images and how many features are found in the image. Also depending on what you want to reconstruct, which results you want to obtain, how accurate, etc. there are several factors to consider.
Since the processing depends on many parameters, it is hard to estimate the processing time, and there is no formula to calculate it precisely.
With the following hardware configuration, using Windows 7 64Bits, 67 images at 14MP and the default processing option:
CPU: Intel® Core™ i7-4930K CPU @ 3.40GHz
GPU: NVIDIA GeForce GTX 660
you should expect about:
Step 1 Initial processing: 5min
Step 2 Point cloud densification (without 3D mesh): 20min
Step 3 Orthomosaic: 35min
Please note again that with a different project but the same number of images, with the same computer configuration you could obtain different processing time. For example, the image content effects on the processing time. The more keypoints are found on the images, the more time it will take.
I believe the following posts can bring you more insight into benchmark assessment and processing time: