Register for free to join our community of investors and share your ideas. You will also get access to streaming quotes, interactive charts, trades, portfolio, live options flow and more tools.
Register for free to join our community of investors and share your ideas. You will also get access to streaming quotes, interactive charts, trades, portfolio, live options flow and more tools.
Haswell CPU.
gb
The $190 SKU is a major volume sweet spot. Early population of that position at launch tells me that Haswell ramp will be very steep.
gb
Yet the SRAM guys (can't remember the loudest one, "real men have fabs" guy...) were adamant that external caches based on the status quo were "forever."
Your comments are spot on of course. I can forsee some integration of DRAM, perhaps for similar power/performance reasons or perhaps even space savings in the extreme.
gb
This DRAM thing reminds me of the days prior to the 486. Back then SRAMs were selling like hot cakes with the 82385 and competitive controllers. Common wisdom was that SRAMs would never be integrated due to die cost.
gb
With 30 coauthors sounds like something serious that everyone wanted recognition for.
Although likely for GT3 I suspect it will see use more broadly although not as a main memory chip.
Too much investment to let idle.
gb
Intel's software tools group and developer relations group should benefit any such activity and due to the amount of re$ource$ should have a much larger impact than anything AMD could muster especially considering a 10x volume target compared to any AMD chip.
gb
"Levers" or couch cushions...
The last question in the call produced the most telling answer. The CFO admitted that they would need to "pull some levers" to maintain their cash balance. This quarter was full of such levers with the sale of the Austin campus and a firesale of previously written off Llano devices. Another questioner asked if there were more Llanos to be sold and the answer was yes but no mention of quantity.
Of course there was no description of the remaining levers...
I wonder if one of the analysts will actually spend the time to do a teardown of the financials to remove the silly bits to figure out what's really the state of AMD. OTOH, I wonder if it's really possible to do so.
gb
Their story has always been in the past "but, but, but...our graphics is better!!". Along with of course "we're cheaper." Not sure that graphics difference is going to count for much with this next generation. Cheaper will always win a few designs.
They're still thumping the HSA drum but I suspect the amount of effort and $$ to make much of that is beyond their dwindling means at this point...as if it ever had a chance.
gb
And all of that goodness could have been had with an optional touch interface rather a "mandatory" one. I've used Win8 now with my wife's notebook and hate the whole UI experience. There was absolutely no reason to make so many gratuitous changes.
And touch is just plain silly for desktops as well as for most traditional notebooks. I'm sure there's a place for it but there's no way it should have been the primary interface.
gb
No, it knew how to make what was DRAM back in 1982 when I joined Intel. It got out a year or so later.
What is now DRAM is quite different and I doubt Intel has a process that would be useful for it.
Parts ain't parts...
gb
The ham radio ap (like many) continuously filters a stream of spotted stations against a database using a set of user defined conditions. Near realtime response is highly desireable.
The RF simulations are most often done for analyzing antenna configurations.
Not everyone just does email...
My daughter's photoshop and lightroom work for her painting references seriously tax the new i7-3770 machine I just built her.
gb
I run some apps for ham radio on my quad 2.4g machine that regularly use 50%. Having the headroom keeps the machine responsive during these peaks.
A single number may not reflect what the machine really feels like.
When I run RF simulations the machine maxes out.
gb
dram is radically different in process from intel's logic process, not close.
intel has access to flash thru their relationship with micron.
gb
sarmad: unrealistic is hardly a strong enough characterization. bizarre fantasy starts to come closer...
gb
I swear I know Luongo from somewhere in my pre-retirement past. And the memory isn't favorable...
gb
AMD sees the era of Moore's law coming to a close
what a maroon! /bugs mode
gb
The real issue here is that these folks get a high news profile but have zero accountability. They get to shout fire in a crowded theater anytime they want but unlike some random street bozo they get high profile exposure for their comments and in the echo chamber of the internet it is visible and repeated everywhere.
Meanwhile their counterparts play the swings they create, no doubt with considerable front-running.
gb
Intel has for many years shown their process improvement curves showing how quickly a process matures. For more than a decade now each process has gotten better faster than the one before it and has achieved lower defect density.
They don't publish actual numbers of course, just relative comparison.
gb
I believe it was Mark Bohr several years ago that gave a presentation on how Intel would merge their up to then separate process development roadmaps to converge at 22nm so that they could more easily move designs and libraries up and down the power/performance scale.
Of course the analysts would pay zero attention to the importance of that and just assume that any other foundry could do it as well if they hadn't already.
Break out the clue bats...
gb
And until they ramp on 20nm, they won't have any idea of what they will need to do to make their hybrid 20/14 stuff work.
Lately their ramps have been much slower than forecast.
IMO we're at an inflection point if Intel ramps 14nm as promised.
gb
Looks like everyone but Intel in that forum and doesn't sound like the wannabees are very happy campers. Meanwhile Intel will be ramping 14nm soon.
gb
And I use my desktop for simulations as well as high information density displays across two 24" monitors as does my daughter and occasionally my wife.
Touch is a non-starter for these. Yet anyone using Win7 foregos the security and performance improvements of Win8 which were non-trivial.
Not everyone can use touch, not everyone needs touch, not everyone wants touch.
geddit??
gb
Sorry I thought SA referred to SemiA rather than SeekingA.
gb
You seriously think someone who was in an "executive position" would ever post such on SA or any similar site?
gb
a throng of tablets...
a flock of phones...
a drove of devices...
gb
Hardware sales slid by 23% for the period; Oracle had previously forecast the unit to be flat to down 10% for the quarter.
Intel can continue to organically improve their graphics and leave the high-end to NVidia as long as NVidia can afford to develop and produce it.
However even without a major decline in PC volumes that will be increasing difficult for NVidia with the transition to a greater percentage of mobile and a continued improvement in integrated graphics for both desktops and mobile.
At some point the ROI just isn't there for a major graphics effort by NVidia. I don't claim to know when that is.
gb
Depends on what they call "20nm" I suppose but if it is a real shrink to 20 from 28nm then I think it's unlikely volume worthy much before 2015.
Everyone but Intel has had a real problem with anything below 32nm especially when adding features to hit the expected process improvements.
gb
I don't recall GS ever saying anything good about Intel. If Intel does good "it can't go on." If Intel doesn't do good "I told you so."
They appear to be only focused on playing whatever buy/sell opportunities they can create through their "advice."
gb
Interesting bit about the "special charge" of $50M. They apparently had the property on the books at $200M and could only get $150M for it.
I wonder how much else of their assets are similarly over valued by 30%.
Might make short work of their remaining shareholder equity if properly accounted for.
gb
You might want to revisit TSMC history wrt hitting volume production milestones. Better than Glofo but...
gb
What "all that capacity?" Did you listen the to the CC?
The analysts got all bent out of shape about the additional spend that was going to happen to get the next size pizza ready.
They completely muddled it in with the normal Capex spend. No amount of explanation could straighten them out.
This "unusual" bump occurs about once every 15 years or so to get over the next bump in technology. Of course that's far beyond the collective memory of the analysts.
Clueless...
gb
They didn't ask me...but it would take a huge effort to port ARM stuff for the benefit of maybe a year or so production. That cost would be much better put to use in just moving to x86 including some sort of emulation as was done in the PPC to Intel move.
I think the effort to fab ARM would be a waste of time and money for both Apple and Intel.
gb
The problem is that the analysts want Intel to fill major fab capacity with foundry since they're convinced x86 and/or PC is deadmeat.
Intel isn't going to build competitor's chips which arguably Apple/ARM is.
Intel isn't also going to build a ton of jellybeans without getting paid for the advanced capabilities their process lead is worth.
So the analysts have stupid (and I use that word advisedly) expectations. They are by and large morons.
The Altera deal is an example of doing it right.
The x86/phone thing is taking longer than anyone would like but it will be successful due to exceptional software tool ecosystem and process leadership.
Apple will begin to slide until they concede that x86 fabbed by Intel is their only option. It's happened before...
ARM's valuation bubble will pop...
gb
With 100K employees and the existing BOD having gone through the "foundry" business model, no doubt, dozens of times at SLRP, I'd be astounded if an outsider came in "RR style" and declared foundry as the shiny new thing.
Furthermore the same analysts who keep declaring that foundry esp Apple foundry brings exciting possibilities will bemoan the gross margin hit that would no doubt come with it.
gb
Saw a post somewhere that speculated that Intel foundry wins which are focused on those that need the highest performance have the effect of removing process driver accounts from existing foundries.
Without those accounts the payback for new process development becomes longer or those new wafers become much more expensive for a time.
gb
Dunno. But I suspect HSW will ship sooner than June since they have significant inventory already per January CC.
gb
IIRC Haswell supports self refresh displays which could help system level power consumption. It also has an on package high resolution VRM which makes for much finer grained power management.
The 15w chip number doesn't really tell the whole story.
gb
Zackly!
gb
New BOD member at AMD. Interesting CV.
http://m.yahoo.com/w/legobpengine/finance/news/martin-l-edelman-appointed-amd-211500879.html?.intl=us&.lang=en-us
gb