I'm going to sign off for a while so why don't you get all of your worthless, content-free, no support opinions out of your system while I'm gone and head back to the ARM marketing department to do whatever it is you shills do when you are not clogging up the message boards of stocks you have no positions in...
'So apparently the Abu Dhabi Investment Authority isn't willing to pony up.'
Only because there is no volume need now but they have the cash potential to do so when needed. AMD sold G-F a right donkey with IBM's flawed 32nm process and I think they are just finding that out and realizing they won't be needed all these new FABs ... at least for now ;-).
"A pair of quad-core 2+ GHz 28nm ARM A15s joined by their CCI-400 Cache Coherent Interconnect at 5-10W. Funnily enough Nvidia are planning to do just that with their top-end Wayne model around 2013."
It will all be over but the crying and the shouting by 2013. And it will be the ARM shops doing both. Too little, too late, too slow. The market is slipping away from ARM as their fabrication coasts to a stop...
'And what ARM processor will you put up against a 77 watt Ivy Bridge processor?'
A pair of quad-core 2+ GHz 28nm ARM A15s joined by their CCI-400 Cache Coherent Interconnect at 5-10W. Funnily enough Nvidia are planning to do just that with their top-end Wayne model around 2013.
And what ARM processor will you put up against a 77 watt Ivy Bridge processor? A Tegra 3? No thanks, I'll pay for the extra 75 watts.
:) Tegra 2 wasn't the best implementation so we'll see with Tegra 3, but it doesn't look good. That said, if the claimed benchmarks are anywhere close you wouldn't need many Tegra 3 to outperform Ivy Bridge in parallel work loads.
As a side, have a nosey at:
[Oracle? You're trying to muddy the water with Oracle? Oracle is a software company, primarily and substantially. Trying to compare Oracle to the ARM vendors is completely illogical and inappropriate.]
"There is lots of cash to fund the fabs."
[Now this is blatantly false. The number of companies able to build fabs has been dropping and will likely drop to a handful because of the immense cost. As costs continue to rise, it gets harder and harder to handle the capital expenditure.]
Not sure how to respond. Did you know Oracle bought Sun? Have you seen the Sparc roadmap? Where do you think these chips are made?
My point about Oracle was that there are other large players who need high performance and have the pockets to fund any fab they need. What about Samsung? What about IBM?
[No, ARMs road to 14nm is anything but clear. The only thing we know about it, is that it's long. Really long. You hardly see anything about ARM's 14nm eforts. What is clear is that no one will be breaking ground on a 14nm ARM fabrication plant for years. And there are the unresolved issues about whether they will have to do both EUV and 450mm wafers at the same time. If there is anything clear on ARM's 14nm efforts please post it up.]
Yes, 14nm will be years away. However, there is a roadmap:
[And the reason that ARM hasn't revealed its 14nm manufacturing roadmap is because they don't have one. Heck, we can't even guess at a year when it might arrive. And don't make me post up for the 5th time the information about ARM's potential server market share being less than 2.5 percent and that's 2 years from now. No, there is no good news for ARM for years out on fabrication. ARM isn't going to fade to black any time soon. But they are going to be giving up market share and that big lofty P/E ratio...]
Like I said I am cynical about ARM and server. The numbers you are quoting are from Intel. The ARM vendors (as you would expect) are quoting numbers that are a lot, lot higher.
Global Foundries has scrapped a new construction plan and an expansion plan. So apparently the Abu Dhabi Investment Authority isn't willing to pony up.
Thank you for the clear example of the crisis that the fabless model is bringing to those who thought it would save them. Instead it's precluding their ability to compete on an ongoing basis...