memory, core, network bound performance comparisons - example RTX-A6000 outperforms due to VRAM bandwidth

[obriensystems]

For example, a simple iterative unsigned integer based 64 to 128 bit precision loop that does not need floating point and does not interact with VRAM during iteration should perform the same. However certain batch algorithms will benefit from increased VRAM bandwidth like on the RTX-A6000 over the RTX-A4000