Impications of Mantle and AMD's slower per core but more core strategy...

[AtenRa]

We'll see. I can see why AMD is pushing this. 6-7 years later there will still games being developed for GCN but 6 years on the PC is a quantum leap time. If we're stuck with GCN for the next, this means stagnation to the PC market. Also the use of specific branches of code will alienate small developers, and we'll soon see AMD specific titles. And the PC doesn't need that kind of optimization, as you can always throw more hardware at the problem.

There is no win for the consumer here. Be locked to AMD isn't a win. Stay with the same technology for the same time of the console cycle isn't a win. Cheaper console ports aren't a consumer issue. It's a win situation to AMD GCN in the short term, and a loss for everyone in the medium term.

GCN is scalable, you can add more CUs, more rops, more TMUs, more Tessellations units etc. You can have a 20nm 4000+ cores with better optimized ROPs/TMUs etc next year and so on.

 

[cbn]

If there is anything that can help or slow down the erosion on the x86 pc market its solitions like mantle.

In general, I think that would be true for the gaming segment of x86 desktop/laptop....if Mantle pans out.