Agree. AMD has recently won a small amount of share back, but the market for discrete client GPUs is not where the action is. It's in datacenter servers and supercomputers. The real threat to AMD probably isn't Nvidia, it's Intel's latest integrated Xe graphics, which will likely take a big bite out of the low end of the discrete client GPU market. The discrete Xe part, whenever it comes, if it comes, is also more likely to hurt AMD than Nvidia.
Geek.
PCIe 4.0 does have an advantage for client computers, but it is doubtful customers will see it per se. For a given amount of throughput PCIe 4.0 consumes less power than 3.0, and halving the number of conductors for a given throughput requirement makes board design easier, though I suspect 4.0 may need a more expensive board material.
Ignore fab process "node numbers". Intel has had so many variations of their 14nm process even Intel engineers have admitted in the press they lose track of which one their product is using. Intel's new 10nm is impressive technology, as are their new "11th Generation" CPUs, but they are still behind the AMD/TSMC team overall IMO. Being late to market with PCIe 4.0 in the client market was mildly uncompetitive, in the server market it was pretty inexcusable with the latest SSDs and 100Gb+ networks. Here's a relatively non-technical blog on the new Intel client CPUs:
Author: Boyd Phelps, Corporate Vice President of Client Computing Group
medium.com
Here's a great analysis on the effects of fab process on different portions of ICs based on the circuit type:
Transistor count and transistor density are often portrayed as technical achievements and milestones. Many vendors brag about the complexity of their design, as measured by transistor count. In reality, transistor count and density varies considerably based on the type of chip and especially the...
www.realworldtech.com