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.
you realize of course that rambus makes identically zero products. they make ip only.
"using fd-soi" means nothing in that case.
gb
There seems to be a lot of denial in the semi industry of late. The HiK/MG breakout by Intel has never been dealt with and has now been compounded by tri-gate.
How long will this go on before something snaps?
gb
It's almost like they're presuming Intel is going on a safari somewhere. I realize that they weren't successful in their previous attempts to compete with Intel but this flailing attempt for a new direction seems pathetic.
gb
That's why I think AMD will get smaller in all metrics including x86 headcount.
They've given up on big core x86 and heading for small core in a somewhat agnostic fashion. Meanwhile the Atom and Core lines on leading process tech are likely going to wipe the floor with even their lowend x86 ambitions leaving them with ARM, "fabric" and ATI.
I don't think that will amount to much.
I suspect that one of the "levers" mentioned in the Q1CC will be on the trapdoor of the gallows for many.
gb
I can't wait for the 14nm ramp at Intel to lay waste to all the dreams of others...
gb
My prediction: AMD is about to get much smaller.
gb
Intel revealed details of its second generation Intel Xeon Phi products aimed to further increase their supercomputing capabilities. Codenamed "Knights Landing," the next generation of Intel MIC Architecture-based products will be available as a coprocessor or a host processor (CPU) and manufactured using Intel's 14nm process technology featuring second generation 3-D tri-gate transistors.
As a PCIe card-based coprocessor, "Knights Landing" will handle offload workloads from the system's Intel Xeon processors and provide an upgrade path for users of current generation of coprocessors, much like it does today. However, as a host processor directly installed in the motherboard socket, it will function as a CPU and enable the next leap in compute density and performance per watt, handling all the duties of the primary processor and the specialized coprocessor at the same time. When used as a CPU, "Knights Landing" will also remove programming complexities of data transfer over PCIe, common in accelerators today.
I expected this CPU role eventually. This should be interesting.
gb
I just browsed through the list going back to 2009. It looks like through at least that time frame that only one "big system" gets introduced each 6 month period to push down the others. Seldom is there more than one new entrant to the top5.
This latest Milky Way tops the previous #1 by 2x. I wonder if something other than a Phi based system will knock it from its slot. Anyone know of other systems being designed at this level and what their makeup looks like?
gb
Top500 win
http://finance.yahoo.com/news/milky-way-2-wins-1-080000816.html
As expected. I'll be interested to hear about other Phi news this week at ISC.
gb
I wonder what OS support is needed for C8-C10 states?
gb
Not clear what socket HSW-EP would use. So upgrade from IVB-EP is unclear as well.
gb
I thought the design was interesting. I was surprised to see them spend time on something they couldn't sell immediately to a bunch of developers however.
I think a common cooling tower applied to a workstation, combined with a Tbolt link to optional expansion seems reasonable. I have no idea on price however given the likely smallish volumes. Lots of custom engineering and small volumes mean high prices to achieve ROI.
It will probably be fast and if they're right be rather quiet so that may float a few boats.
I doubt it will trickle down to PC desktops however.
gb
yawn...
gb
Every design has a lot of trade offs. I LOL when some overclocker gets all twisted about how Haswell doesn't float his boat. SemiA is full of them.
HSW was obviously targeted at serious platform power reduction while doing useful work. Some OEMs will do it right, some won't.
If intel wants to proliferate GT3e they will. Anyone remember the PII configuration?
I think the intel plan could be described by "non legitimus carborundum".
HSW on 22nm is a start. 14nm will be 60 grit.
gb
BPPC? Bletchley Park PC?
gb
4 "British PC manufacturers".
wouldn't that be a null set?
gb
I suspect this is datamode only rather than voice data.
gb
BTW, pure speculation on my part as to the motive. I'm not remotely a finance whiz. But in the past when these things moved from private to public the next shoe to drop was another private placement, often to those who were previous private placement customers. By freeing up their cash by giving the opportunity to sell their current holdings they are potentially in a position to get some even riskier bets from AMD.
gb
The new ones are registered and therefore publicly tradeable.
gb
This appears to make these privately placed obligations more easily traded thus freeing up opportunity for more private placements.
gb
Looks like AMD is starting to pull some of the unnamed "levers" mentioned in the last CC.
http://m.yahoo.com/w/legobpengine/finance/news/amd-announces-extension-exchange-offer-120000617.html?.intl=us&.lang=en-us
gb
Zackly!
No one designing a product will give those slides a moments notice. They're strictly intended to support a hyper-inflated share price.
gb
From what has been launched so far it seems to me that intel has focused on mobile performance and battery life. Probably a good call by them.
Who knows what future SKUs hold for DT or server.
gb
Product rulz, metal pitch chestbeating drulz
gb
Well I was being a bit conservative but there's clearly a lot of volume between $100 and $200 that is a big chunk of the ROI. Perhaps the dominant portion.
Any way you slice it Tom Piazza and company seem to have clear field running with a great process roadmap to execute on and apparently are "sanctioned."
gb
Makes no sense. Why do you love a dGPU rather than what can do?
gb
I didn't say kill. I said kill anything below $1K. That's effectively the same thing however. Of course AMD/Nvidia can chose to price their "$1K" cards lower if they like to lure some customers but the lack of volume is really going to kill the ROI on these cards.
I see now you mentioned laptops. I think that's completely game over. Nvidia may win a few nutcase units but gamers don't use laptops and still buy those monster case deskside units.
gb
What's funny is that Chuckles gets all worked up over Linux drivers and that's the basis for the echo chamber comments in the forum. Yet the forum members are all Windows gamers and don't give a crap about Linux or Linux drivers for the most part. Yet they use his rants as the basis for their convictions.
There's a lot of denial over there. Haswell and the next tick and tock are going to complete the kill of PC discrete graphics below $1K boards. There's going to be a lot of wailing about that.
For a preview of that type of wailing watch what happens if the Chinese MIC system rumor turns out to be true. Pretty sure the black helicopters don't have ability to reach Chinese airspace.
gb
What purpose would these serve other than chestbeating? Who will code for them with their market share?
gb
AMD fanbois won't make AMD products successful. OEMs building and successfully selling AMD based products can.
If AMD's products don't meet the end product needs, especially battery life and fanless operation, most OEMs won't bother. A few will, it always happens and no doubt these will be touted by the fanbois.
But the forums are filled with those same folks complaining that AMD based products are not on the shelves and they can't figure out why. They never will figure it out either...
Product volume is what matters, audio volume matters not at all.
gb
I also see that he's bringing the divisions currently reporting to Dadi to report directly to him in the iA space.
He's going to be busy. Perhaps Dadi will go focus on some of the mobile/new device stuff. He's an intense guy that would be good at that.
gb
Before gesture acceleration can be economically integrated it needs to be understood sufficiently. Usually that takes a few generations of shipping products used widely enough to find the big payoff areas.
To do otherwise requires some with a lot of money, developer clout and potential volumes.
gb
Those desktop users don't configure their PCs. They buy ones preconfigured with crap by OEMs.
gb
I build my own PCs and I love the IGP. It works and doesn't waste power and money. I build my own not because it saves money but it allows me to equip it with SSDs, quiet well built cases etc. it gives me control.
I'll be building a Haswell by the end of the year to replace this noisy Dell.
gb
After being in the industry for nearly 20 years and now 14 years retired I've concluded there will always be an "ARM."
There's a lot of money at stake. There's always some risk investment money available. There's always some investment advisors willing to spout off publicly. If a public offering there's always some analysts with an agenda willing to pontificate.
There are always some Rubes who can be taken.
This will never end but will likely always "end" the same way for the latest contender as long as Intel stays sharp.
I'd bet on Intel...
gb
IIRC it was Facebook that made a big stink about ARM in their disaggregated server thingy. I wonder if Avoton or other Silvermont incarnations has them rethinking their plans.
gb
The reaction of most forum members on S/A is predictably negative on Intel's prospects in spite of Charlie's Silvermont articles. I've said many times in the past that as Intel competes with ARM, whatever is now left of AMD will be clobbered as collateral damage. I seriously doubt any game machine revenue will come quick enough to save them from the double punch of Haswell and Silvermont.
And with fab utilization down Intel will ramp the old stuff down and the new stuff up even harder which means "priced to move."
AMD's CFO mentioned having "levers" to pull to manage cash flow and expecting to need them in coming quarters. It will be interesting to see what those are.
gb
None of these guys are remotely technical. They can only regurgitate data from "independent" sources who they believe are competent or from non-independents who join them in cheering for one side or the other.
Break out the clue bat for a wood shampoo.
gb
Haswell IGP supports self refreshing displays which should be a nice power reduction. It will be interesting to see if any such displays materialize in a volume fashion.
Not sure it will happen soon given all the display issues.
gb
aeassa: it will be interesting to see if AMD's Kabini does as well as you think. Personally I see it squeezed hard between Atom and Haswell and even price reduced IvyBridges leaving little volume for it.
Without volume to make up for ASPs AMD will hurt as they have little to offer beyond Jaguar based products.
I'm skeptical of all of the console "buzz". Not clear that any of them will create the splash necessary to drive the volume AMD needs much less all of them.
gb