I made a benchmark a few months ago for a friend of mine working on computational physics. She needed to compute the determinant of a lot of matrices to perform

I made a benchmark a few months ago for a friend of mine working on computational physics. She needed to compute the determinant of a lot of matrices to perform MCMC sampling and wanted to know how much GPUs could speed up things. For this task, one V100 was equivalent to a 100 cpu cores, so a speed up of 100,000 is definitely not coming only from using GPUs vs CPUs.