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.
I got a pair of 9800GTs back in '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.
By 2015, the essentially-free iGPU part of an i3 Skylake CPU was ~better than the SLI rig.
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.
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%.
Wasn't that actually a very short lived period of time? Especially with gains up to 70%, and you had to contend with new stutters?
nvidia supported it for 17 years and amd still supports it but only on older cards
Oh NV Surround, how we miss thee
I don't miss it.
I got a pair of 9800GTs back in '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.
By 2015, the essentially-free iGPU part of an i3 Skylake CPU was ~better than the SLI rig.
Thanks