Why does AMD have to push advancement down the industry's throat? They've been trying to get software to be multi-thread/multi-core for years. Finally to get it in rendering API's they actually had to write it, prove it's worth in the market, and then hand it out to everyone for free. WTH!!!!
Is it simply because Intel has the advantage in per core performance and the influence they have?
I've wondered about this myself. And that's a really interesting thought and conversation. I think you're right though.
I came to the conclusion that the market suited Intel fine, with DX9 being a singlethreaded monster (only tamed by SandyBridge or better at 4ghz+), and DX11 not being much better in that regard, 1-2 thread limit on the graphics pipeline, it is not just as multithread capable as DX12, and many believe it is. So I think that's the biggest issue, other than just disinterest in changing things. Intel has built many technologies to be completely surrounded and marshaled by x86 to maintain its relevance. DX up through 11 is one of those things, but MS was going to get blown off the map if they didn't adopt it. If game engines were adopting Mantle and Vulkan came out while they sat on DX11.
Nvidia is a much smaller, and insignificant player at the mercy of those with x86, but had adapted to the environment exceptionally well, especially over the past 3 years or so. They're almost a software company (and their technology shows this TBH).
So if I had to say, I'd probably point the finger at Intel myself. And I like Intel in general. I'm a technologist so it's hard not to be thrilled with the stuff Intel and AMD produce, I can't say the same for Nvidia. All of the resources they've created and built up for the driver stack is just gone, almost overnight. AMD won't need top notch OGL/DX11 legacy support, the raw performance is already there enough to play any title. NV needs Windows 10 to be stopped, it would behoove them if its adoption collapsed.
I'm a bit shocked how well Mantle worked out for AMD, they had to do it and I didn't expect it to end so well for them. AMD just has not had success lately, but I'm warming up to them again. Of course the Nvidia war drums never cease, and some people will buy Intel+NV exclusively, always. Because well, frankly Nvidia are a bunch of losers without x86, so they have this Tegra crap as their CPU.
But for most of us who aren't just diehard Nvidia, I think now, and next year are the times to go back to AMD. 16nm GCN should grant a bit of a performance boost and resolve any perf/watt issues that may bother some guys. Zen will most likely be 8C/16T for "cheap"-er but it will push the CPU market forward for the first time since SandyBridge. And HBM APUs along with Intel's Iris Pro is going to change the landscape forever. You can see there's no place for Nvidia anywhere here, they've got to kill AMD soon or AMD+Intel will suffocate them with APUs.
AMD even has a fantastic new Linux driver model for the GCN stuff having been released, and being improved for the SteamOS launch. I work in Linux (desktop and server) daily, so this matters to me. Though I'm a firm believer that if you're not gaming on Windows(10), you're not doing it right. :ninja: At least today, it's changing pretty fast.
Having Mantle as a part of LiquidVR is just pure gold and will pay off bigtime. The whole Mantle story was AMD executing like a machine. None of the effort put into it was wasted at all.
Going forward, it's Nvidia who is at risk if they can't kill AMD soon enough. And it's not possible to pull off in the timeframe necessary to ensure their market, AMD doesn't owe any money till 2019.
So I'm optimistic on AMD even if that's a bold statement today. I also hope to have a 8C/16T Zen CPU with 16nm GCN card hooked up to a SteamVR/HTC Vive headset by 3Q2016. There's really nothing else out there worth spending the full cost of an upgrade on so AMD will receive all of my next ~$2,500 build.