Tom's Hardware's AMD Q&A session clarified some matters.
http://www.tomshardware.com/forum/id-1863987/official-amd-radeon-representatives.html
Mantle will be discussed in detail, AMD's Mantle partners revealed (or most of them) and some Mantle performance numbers shown.
Also revealed about Mantle is that in every case the developers approached AMD and not once did AMD approach a developer(starting with Johan Andersson) and that since the Hawaii event EA/DICE presentation AMD has a surge of developers wanting to know more about Mantle. If EA/Dice is "
Super exited about Mantle!" (appeared on a slide exactly like that) ... well, it's time to find out more!!
Kaveri will be addressed in detail and benchmarks revealed.
The 290 and 290X support system unified addressing. Yet unrevealed but all but certain is that Kaveri will also support system unified addressing.
........
It's now clear Mantle developer support is already strong and will be growing and Mantle performance gains appear to be substantial with Oxide describing 'dramatic' performance gains and Dice waxing "
Super" excited about Mantle. And of course AMD's Product Manager in a post event Q&A saying the 290x running BF4 Mantle will 'ridicule' Titan.
So what happens when Mantle demonstrates a 20-30% (or more) performance delta when AMD's AIB pricing is already cut throat and driving Nvidia to slash prices?
Nvidia at the high end is already at a 20% die size cost deficit. When aftermarket coolers come out for the 290 cards, how much margin is left on the 780 chips to match the R9 290's cost/performance? Add Mantle into the equation, with the outlook of broad and deep developer support, will there be any margin left at all at the high end (or anywhere) for Nvidia if they try to maintain cost/performance parity?
Nvidia is being backed into a corner with it's teeth pulled and claws clipped.
It's obvious Rory Read has everything in place to go for Nvidia's throat in the consumer GPU space, hence the extremely aggressive pricing. The time for a tacit agreement with Nvidia to keep GPU prices high to the benefit of both is over and the time to break Nvidia on a price/performance rack has arrived.
It just won't make sense to buy Nvidia with a flood of Mantle games on the way. And as BF4 shows, once Mantle is incorporated into an engine it can be added
post release to games, so
ALL next gen PC games are candidates for Mantle.
Post APU13 will see Nvidia AIB sales plummet and AMD AIB sales skyrocket. The developers will see that, know it is a long term thing because Nvidia has no answer to Mantle and more developers will get on the Mantle bandwagon.
Tim Sweeney is not immune no matter how much he dislikes Mantle because the Unreal Engine is being widely used for next gen games and
those developers have nothing against Mantle and are going to start
demanding Epic add a Mantle path to the Unreal engine so their games will be competitive with Mantle enabled games.
I suspect Epic and Ubisoft will both be on board with Mantle before the end of the year.
That's not even taking Kaveri into account which adds a whole new dimension of hurt to Nvidia's situation. What kind of performance would a fully HSA/hUMA capable Kaveri + R9 290 combo with system unified addressing be capable of? Is there anything in the Intel/Nvidia stable that would be remotely cost/performance competitive on Mantle games?
Now AnandTechs AMD CENTER makes all kinds of sense. AMD is where all the action will be at and Anand is in on the ground floor.
Interesting times.