So what do you think will be AMD's biggest blunder with Zen? could be anything from marketing to memory bandwidth, slow overall performance, TLB2.0 bug or even yields though i highly doubt it, Samsung are pretty great unlike global shitters
or do you think everything will go perfectly and there's going to be Opteron's with 64 threads blowing the server markets asshole apart and consumer 8 core chips for sub-400 dollars, with either case each core performing like broadwell, Andy Bryant is just going to hang himself in his office and the environment goes to shit because all the market share AMD gained is going to be green power we lost from intel.
I don't think AMD will mess up with Zen.
Just look at their current GPU lineup, the performance per dollar is simply top notch
>>55334952
But it's an entirely different team of people working thousands of miles apart on different projects. the ATI team has actually missed APU deadlines in the past becaause they just kept focusing on discrete graphics, so it's arguable that division isn't even as on tight of a leash as they should be.
>>55334994
The company believes in that core idea though. The value will be there if the AMD name is on the product
>>55335011
The name of the game has always been value, even with their current fx chips (insofar as they can manage with those huge expensive slabs of silicon they call dies). And yes, Vega is probably going to be fantastic, but with basically the same people working on Zen as that worked on Bulldozer except for the chief who stepped down right after BD launched and the inclusion of the guy behind K8... what's to say this is going to be any more competitive than bulldozer was at its launch? Whether or not its a good value is irrelevant because it will be either way.. It has to be. No other option but to give a great value when you're trailing behind with 15% of the market.
My point is if AMD can't offer FULL core cpu's (no sharing resources cancer) that have enough IPC to challenge intel, at least faster than haswell, they will have failed miserably.
>>55334715
Mem bandwidth as usual, if anything.
Ipc
It always has and always will be
