I am using 2x3090 NVLinked GPU on a Dual Intel Xeon 2630 v2 (16 core, 32 threads) setup that only has 32GB RAM currently. All drives are NVME. Any ideas if there is a bottleneck - like the amount of RAM is not enough, and or the processors are too slow for the paired 3090s? I feel the IPR sessions have not shown any significant performance increase on the same scene after switching from 2xTitan RTX NVLinked. Although, before I had a 3rd GPU dedicated for viewport which I dont have anymore =(
This is on Maya (Windows 10), with the latest Arnold build.