Great piece. Was wondering whether to get an NVLink bridge for $500+ for my two A4500s for inference. This pretty much confirms it only really rescues you if you have nonsensical PCI-E topology.

Very informative article. I was hazy on where NVLink sat in the local inference stack. This makes it very clear: to adapt consumer hardware that lacks multiple full x16 PCIe lanes.

nvidia supported it for 17 years and amd still supports it but only on older cards

Wasn&#x27;t that actually a very short lived period of time? Especially with gains up to 70%, and you had to contend with new stutters?

I don&#x27;t miss it.<p>I got a pair of 9800GTs back in &#x27;08 to play with SLI.  It only helped with some games, and despite having all those video output jacks the rig only supported two monitors.<p>By 2015, the essentially-free iGPU part of an i3 Skylake CPU was ~better than the SLI rig.

I badly miss the days when I could buy a GPU and then three years later buy the same GPU for a hundred bucks and boost my framerate by 70%.