AMD

It’s practical to use their existing test suite. It allows for comparison against other parts they have previously tested.

It's not good idea to compare hardware performance from different tests against each other. Even if they test the same game at same settings, testing procedure can be different, and that makes those tests kinda incomparable.

Just look at 5800x3d performance in this video and compare it to 5800x3d performance their video about 14900k from 1 month ago:

• Baldur's Gate 3, Ultra 1080p: 132/100 vs 145/106
• TLOU Part 1, Ultra 1080p: 139/116 vs 152/129
• Assetto Corsa Competizione, Epic 1080p: 175/143 vs 161/128
• Spider-man Remastered, 1080p High + High RT: 144/119 vs 122/93

I'm all in for using few years old CPUs in benchmarks to see how they compare to newest CPUs, but if that was the goal of the video, they would've also tested newest CPUs.

Plus plenty of people are still playing games from the past 5 years.

Of course they are, plenty of people are still playing games older than 5 years too. I don't have a problem with testing older games, but when someone say "5 years later", "today's games", "to see which platform aged the best", "in 2023", when most of the games aren't "today's", aren't from "5 years later" CPUs were released, weren't released "in 2023", and say us nothing about how CPUs aged in 5 years, I find it confusing and kinda dishonest.