I don't share concerns about market fragmentation. We aren't going to see the API that 2/3rds of the market uses disappear.
Yeah. I've mentioned this before, but it was the same situation with glide - even with glide in existence, D3D was the go to and only became more popular with time. Glide was more of a "value add" for 3dfx, and it was created initially because D3D was in it's infancy early on and wasn't usable. Hence there were tons of vendor specific APIs, because there was no D3D. Once D3D caught on, it only became more popular over time as compared to vendor APIs (such as the powerVR, rendition verite, 3dfx glide, S3's API, etc). Speaking of which, from what I remember Glide never provided outrageous performance increases. It *did* provide a performance increase, but it also enabled 3dfx specific rendering techniques which weren't available in D3D at the time - so the main benefit was better graphics. That was then and this is now: D3D has caught up in terms of features, although some would argue that should happen at a faster pace. Nonetheless the point remains. I'm not aware of Mantle offering any graphical features that D3D 11 doesn't.
Anyway, back to the point, fragmentation: there are two games we know of using Mantle in 2014 - thief and BF4. There are rumors of all frostbite games using Mantle, but it turns out that isn't true: NFS rivals won't be using Mantle (their devs posted this, the game also has a 30 fps hard coded limit) and DA: Inquisition is unconfirmed. PvZ zombie warfare is also unconfirmed. Star Citizen won't have all modules complete until 2015 and won't be a complete game until then. Oxide's game? There's no determined release date as of yet, unless that very recently changed. Someone correct me if i'm wrong there, but I had not seen a confirmed release date for it. The next Star Wars Battlefront game is slated for 2015.
I guess the point here is, there are 2 known Mantle games for 2014 and 2 possible games, in that the devs have not absolutely confirmed Mantle use. NFS Rivals will not get Mantle. With this being the case, market fragmentation won't be an issue - it's not like Mantle is being used in 100% of games. Far from it. We're talking less than 1% of games right now at this point. Which is fine, since it's a free value add for AMD users.
So with Mantle having limited use: this is where the benefits of Mantle become really strange. We know now that more or less, GPU limited situations show 0-5% improvement with Mantle in BF4 (per SKYMTL). Low end gets 10-15% improvement. I mean it's free improvement. But this isn't what people expected I don't believe, especially after seeing the 45% marketing slides from AMD.
You know, I hate to say it, but this is why you have to be fully skeptical. AMD has made so many promises that were missed over the years, and it really is hard to take what they say at face value after the entire bulldozer thing and all of the inflated claims from their marketing powerpoints. I really do truly hope that AMD takes a new approach in the future of instead of promising stuff that they can't deliver on, to just work quietly and release something awesome when it's done. When it's done, by all means, do your marketing thing at that point. But the pre-hype marketing and failing to deliver IMO doesn't really do their customers justice. Promises for HD3D, freesync, frame pacing (now broken for more than 2 years..) I could go on. The pre-hype pre-promise methodology needs to change IMO. And now we find out that AMD in fact will not hit their January promise for Mantle - honestly, while I expect early Feb, I am not surprised that they missed it. I had 7970CF in the past, so I have been there done that. Gets old after a while.