Intel don't care about consoles or any other low margin business for that matter. They were in with OG Xbox when it involved bugger all extra work and got out again when it looked like involving any significant amount of work. Their fab process is literally peerless so anyone else hitting 10 nm for 2018 is highly doubtful in my book. Nvidia fired off two duds in a row with OG Xbox NV2A and RSX while also being litigous so I can't see anyone touching them and the long awaited Denver with a barge pole.
Surprised at how bullish some folks are on the cost for 16GB of HBM, I can only see this if the costs and fail rates are at least 2x better than just slapping more GDDR chips on the m/b. Is it really that simple to implement?
Cost is still king in console land on all sides and I can't see the bean counters in s/w being too keen on 3 year console refresh cycles. As a cynic I would also expect another mechanical drive for system storage with maybe 64 or 128 GB of flash cache (likely as a hybrid HDD). X-point is off the table as it's not an industry standard and Intel will be making too much money caning it off the server space if it takes off to be bothered with the low margins in consoles
Surprised at how bullish some folks are on the cost for 16GB of HBM, I can only see this if the costs and fail rates are at least 2x better than just slapping more GDDR chips on the m/b. Is it really that simple to implement?
Cost is still king in console land on all sides and I can't see the bean counters in s/w being too keen on 3 year console refresh cycles. As a cynic I would also expect another mechanical drive for system storage with maybe 64 or 128 GB of flash cache (likely as a hybrid HDD). X-point is off the table as it's not an industry standard and Intel will be making too much money caning it off the server space if it takes off to be bothered with the low margins in consoles