nemesismk2
Diamond Member
Life in the high end is tough, isn't it?
AMD needs to kill its big cores now.
Why should AMD listen to an Intel fan?
Life in the high end is tough, isn't it?
AMD needs to kill its big cores now.
It's Digitimes, let's not act like this is a fact until confirmation.
What is a concern is that we've seen nothing at all from Intel on 14nm. For as long as I can remember they have been showcasing their node advances as soon as there has been silicon to show, but so far they haven't shown anything on 14nm. I'll be surprised if Broadwell doesn't slip into H2 2014 as Intel puts everything behind Atom.
I don't know, but logic would tell me Kaveri = GF, consoles = TSMC
The decision to withhold Broadwell from the desktop market and only deploy it to the mobile segments is highly suggestive of 14nm having issues getting the drive currents and reliability where they need to be in order to field broadwell chips that are reasonably clocked (comparable to Haswell/Ivy at least).
14nm might be spanking Intel's R&D team at the moment, at least enough for them to effectively delay releasing the high-performance portion with Skylake to the desktop and instead adding in a "haswell refresh".
Fully agree here. Digitimes is not exactly known for its reliability.
Vr-zone reported a delay too.
.AMD Kaveri supply may be delayed to early 2014
It just looks like an uncited copy of the digitimes report to me, however. That said I wouldn't be surprised if Kaveri was delayed a few months.
Is it necessarily that, though? I would suspect more that Intel is prioritising their initial fab capacity, and giving it to the 14nm Atom instead of a desktop Broadwell. Remember we're only seeing 22nm Atom after 2 desktop releases at 22nm- could they not just have switched their priorities, and given "first dibs" to their ultramobile parts?
Think about it for a minute, Intel does things to make money.
They aren't going to prioritize a lower ASP and lower profit SKU like Atom over the possibility of selling higher profit and higher margin chips like the broadwell equivalent of a 4770K...unless something is preventing them from financially benefiting from selling that broadwell equivalent of a 4770K (like low yields or unobtainable clockspeeds).
When Plan A isn't working to plan and you find yourself falling back to Plan B, you don't tell the world they are getting Plan B. You tell them you meant to do Plan B all along.
Prioritizing Atom at the expense of delaying desktop Broadwell only makes financial sense of the yields for broadwell on 14nm are going to be so low that the cost and margins of a desktop broadwell end up worse than the margins for a low-ASP chip like Atom. (which is basically what I am assuming is the root reason for the sudden slip in desktop Broadwell plans)
Atom may be a lower ASP part, sure, but think about the strategic significance of it. Intel is well aware that the desktop market is in a slump (potentially for the foreseeable future), and the mobile phone market is exploding. Getting their latest and greatest mobile part out on their world-beating process as fast as they possibly can will give them the best chance of making inroads into this new, high-volume market.
The strategic incentive for creating a mobile part on the newest, lowest powered process and finally cracking the mobile market is far higher than the incentive to slightly improve margins on the enthusiast-oriented desktop parts.
Prioritizing Atom at the expense of delaying desktop Broadwell only makes financial sense of the yields for broadwell on 14nm are going to be so low that the cost and margins of a desktop broadwell end up worse than the margins for a low-ASP chip like Atom. (which is basically what I am assuming is the root reason for the sudden slip in desktop Broadwell plans)
Why should AMD listen to an Intel fan?
Think about it for a minute, Intel does things to make money.
They aren't going to prioritize a lower ASP and lower profit SKU like Atom over the possibility of selling higher profit and higher margin chips like the broadwell equivalent of a 4770K...unless something is preventing them from financially benefiting from selling that broadwell equivalent of a 4770K (like low yields or unobtainable clockspeeds).
When Plan A isn't working to plan and you find yourself falling back to Plan B, you don't tell the world they are getting Plan B. You tell them you meant to do Plan B all along.
Prioritizing Atom at the expense of delaying desktop Broadwell only makes financial sense of the yields for broadwell on 14nm are going to be so low that the cost and margins of a desktop broadwell end up worse than the margins for a low-ASP chip like Atom. (which is basically what I am assuming is the root reason for the sudden slip in desktop Broadwell plans)
Personally, I just don't see people flocking to $1000 convertibles and consequently Broadwell.
There will be a desktop Broadwell - just not an LGA Broadwell.
No LGA Broadwell makes perfect sense.
Think about it for a minute, Intel does things to make money.
They aren't going to prioritize a lower ASP and lower profit SKU like Atom over the possibility of selling higher profit and higher margin chips like the broadwell equivalent of a 4770K...unless something is preventing them from financially benefiting from selling that broadwell equivalent of a 4770K (like low yields or unobtainable clockspeeds).
When Plan A isn't working to plan and you find yourself falling back to Plan B, you don't tell the world they are getting Plan B. You tell them you meant to do Plan B all along.
Prioritizing Atom at the expense of delaying desktop Broadwell only makes financial sense of the yields for broadwell on 14nm are going to be so low that the cost and margins of a desktop broadwell end up worse than the margins for a low-ASP chip like Atom. (which is basically what I am assuming is the root reason for the sudden slip in desktop Broadwell plans)
Or it makes financial sense to build less 14 nm fabs because you can supply desktop cpus from your old 22 nm fabs. And when mobile chips move to 10 nm, desktop can use the 14 nm fabs and so forth. Makes perfect sense to me.
Speak for yourself I'd be quite keen on a convertible with 1080p screen, decent iGPU and a powerful dual-core Broadwell to replace my netbook- I'm starting to get sick of this low-res screen, and the number of games that just don't work properly on it. (If you can't actually see all the entries on a menu, clicking on them is pretty hard...)
They aren't going to prioritize a lower ASP and lower profit SKU like Atom over the possibility of selling higher profit and higher margin chips like the broadwell equivalent of a 4770K...unless something is preventing them from financially benefiting from selling that broadwell equivalent of a 4770K (like low yields or unobtainable clockspeeds).