It is compatible, however, with only X570 providing PCIe 4 and B550 still only announced, people who have built Ryzen 3000 boxes (like me) aren’t going to likely jump on B550 now. We’re going to wait for the next upgrade cycle. I.e. lost opportunity.
And considering that unless you are pushing extremely high frame rates (think well above 120 fps), PCIe 3 isn’t your bottleneck right now. Games crank up the effects to the point that any extra demand on the PCIe bus tends to get balanced out by the extra work the GPU has to do. There are some latency wins, but they aren’t huge.
According to the Steam Survey, two-thirds of users are still on 1080p displays. Yes, 1440p monitors are getting cheaper, but it hasn’t really made a dent yet. But this sort of nit-pick ignores the forest for the trees: People are running on hardware that doesn’t need DSC, and doesn’t benefit from PCIe 4. If you can cost-drop (due to better binning, etc) and die shrink existing designs (i.e. like the RTX Super refresh did), you can provide performance gains for these budget customers for less outlay of resources, and those resources can be used more effectively and under less time crunch.
Especially if the new architecture simply isn’t adding functionality the target market is going to be utilizing anytime soon, and will likely replace by the time they will.
Not exactly a great sign that AMD can handle Intel/Nvidia level volumes, is it? This ties back to my previous point that perhaps rolling out a new architecture on budget GPUs and then attempting to scale it up wasn’t the best use of resources if Sony and Microsoft were so important.
Don’t disagree, but doesn’t excuse the issues on Windows.
I look at it more from this perspective: What is the scale and impact of those bugs? If those bugs only impact Mac, but don’t do more than annoy, that’s one thing. If a sizable chunk of the Windows user base is hitting black screens, that’s another. If nobody got a Ryzen 3000-series CPU that didn’t decide to boost and draw the max rated power when Windows decided to wake up the CPU to do idle processing, and it took 2 months for a proper fix to get rolled out, that’s bad.
I kinda expect that bugs that can be a deal-breaker for customers to be the ones fixed by launch or at least ASAP.
For me, the difference is this: I’ve had no complaints with my M395X-based iMac that I’ve used for years. I’ve had no complaints about my 580X, or even my Vega 56 in Windows or as a Mac Mini eGPU, or even in the Mac Pro (other than boot screens, and some signal strength issues when used with a DP switch, which I admit is niche). But the fact that the Navi line is flat out known for having issues, both in the 16” MBP, Windows users with RX 5500/5600/5700 cards, and now W5700X cards are showing up and throttling themselves to be slower than the 580X, I’m at the point where I’m really tempted to simply return the card unopened and wait this out. I’m not sure I agree that this is AMD doing better when I can forgive a GPU not being as fast as it needs to be to make Nvidia look bad, but I can’t forgive a component that doesn’t do it’s core job properly.
But really, the way that AMD has been treading water exposes them to risk as Intel starts entering the market. Nvidia doesn’t have to outrun the bear (Intel) here, it just needs to outrun AMD. I wouldn’t be surprised if Apple starts including Xe HP GPU options down the road as a hedge against AMD without submitting to Nvidia, especially if the Xe HP designs are as efficient as the leaks suggest.
