I am trying to build a workstation specifically to run MD simulations with intel’s i7-14700K (24 core, 32 thread) and an Nvidia RTX 4090 (16384x CUDA cores and 512x Tensor cores). Currently, I’m planning to get 128 Gb of RAM and wanted if it makes sense to have so much memory or should it reduced to 64Gb or increase it 256 Gb.
Has anyone here, tried a similar setup and have any suggestions on optimizing the performance?