2 Dec 2015 7:16
#17044
Trying to benchmark the linear algebra code with large matrix sizes. I have the 4770k CPU with 1600MHz DDR3 RAM but am curious if 3000-3200MHz DDR4 RAM will bring the benefits for the problem solving such equations as the RAM could be a bottleneck. And if it will then CUDA solvers on our graphics cards multicores which use DDR5 at 7000MHz might help even more but that's a big change in software side. So let's try first if DDR4 idea works.