what is the point of this article ? I get it if you are warning people from today cards , but this is so unprofessional .
I'd say that a review like that is actually far more professional than just looking at things immediately at their release date.
Because it zooms out a little and provides you longer term perspective.
And I was actually quite impressed with just how well argued it was, avoiding any cheap bias as far as I could tell.
Of course you'd hope that vendors have learned from past mistakes and won't repeat them outright, but when you are trying to imagine how things might evolve, this article provides you with tons of things to think about.
AMD's bet on HBM or AMD going to 512 bus width even on the normal R9 290X showed great engineering skills (to my knowledge nobody has done a 512 bit PCB design since), yet also that it doesn't pay off, when you can get so much more with texture compression or an extra layer of cache: brute force often just doesn't pay off economically.
It's simply a fact of life that AMD couldn't hedge more than one or perhaps two bets: they were too small to win on all fronts and resorted to some rather cheap relabelling to crank up revenue.
If you carefully read all these tech sites, chances were you couldn't get hoodwinked, but I'll have to admit that occasionally I still got snookered, e.g. with a Kaveri A10-7850 which never really delivered on anything.
That R9 290x was my last AMD GPU after decades of ATI/AMD (it went into a Phenom II X6 system). My first ATI was the Mach8, an IBM 8514/A clone and my first accelerated graphics cards after dozens of VGAs. In fact I started with an EGA on my 80286 and I even personally ported X11R4 on a TI TMS 34020 GPU.
I've given AMD a chance on every PC purchase over the last 40 years because I like the underdog and believe it's essential to have. I've actually tried to have AMD/ATI and Intel/Nvidia side-by-side whenever possible, but in the GPU arena, there just wasn't an able bodied contender for a long, long time.
At least AMD has won the last few big CPU rounds, but that seems to matter less and less as nearly everything that requires real number chrunching power, just runs on GPUs anyway and I have 16-22 CPU cores from teams blue and red just waiting for my RTX 4090 to do the work.
I've actually gone all out and started buying some Intel ARC 770 GPUs, because they seem to be the more "under" doggies today. I guess it's a good thing they are so terribly cheap, because they sure aren't that terribly good: nor at graphics nor at GPGPU. Supporting alternatives to the leading contender is just a necessity and I'll continue to do so within the limits of my means.