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.
Intel execs lowered the bar for the Data Center Group's growth this year to the "high-single digits."
When you utterly dominate a market it is hard to grow faster than the market itself.
http://www.idc.com/getdoc.jsp?containerId=prUS41735516
vendor revenue in the worldwide server market declined 0.4% year over year to $13.4 billion in the second quarter of 2016
[..]
Demand for x86 servers improved in 2Q16 with revenues increasing 7.3% year over year in the quarter to $11.6 billion worldwide while unit shipments increased 2.7% to 2.3 million servers
Growing high single digits in servers is a heck of lot better than a certain other
company that makes server MPUs and systems just released results of shrinking
revenue by low double digits in servers.
IBM's q2 report reflects non-x86 server sales shrinking fast
http://www.ibm.com/investor/att/pdf/IBM-2Q16-Earnings-Press-Release.pdf
Systems (includes systems hardware and operating systems software) -- revenues of $2.0 billion,
down 23.2 percent (down 23.3 percent adjusting for currency). Revenue reflects z Systems product
cycle dynamics; gross profit margin improved in both z Systems and Power.
Mainframes in down portion of upgrade cycle, POWER is simply dying
away like the rest of the non-x86 servers, especially UNIX based.
Sorry but I don't have much insight. Where I work now hires
infrequently (very low turnover) and then only experienced
individuals).
I hear there is healthy hiring in the valley, don't know about
new grads though.
Perhaps there's a limit to what can be done in a mobile?
Ultimately there is somewhere down the road but the Note 7 fiasco
has nothing to do with IC or semi technology but rather unsafe battery
design and/or manufacturing.
More energy dense batteries means bigger and hotter fireworks when
things go bad.
Torvalds has an affinity for x86 because of the infrastructure and ecosystem
LOL, Linus schools the more ardent ARM pompom shakers about this a
few times a week at RWT forum.
I spec'd out all the components I wanted and got the computer shop
to assemble it and install the OS. Costs a bit more than DIY but it
saves me aggravation and possibly a broken connector or improperly
installed heatsink.
Posting this from my new 6700k system with GTX 1070, 16 GB DRAM,
500 GB SSD, 3 TB HDD, and ... wait for it... Windows 7 of course!
Scale, basically. Intel is also the only one in the mobile market who can develop really differentiated products instead foranother me-too ARM SoC.
No, no, no!
You can't race speed boats around a slalom course of buoys in a
500,000 ton displacement supertanker.
Intel is really good at designing super high performance MPUs. But it
uses intensively engineered uarch, circuit, and physical design to
achieve it. This is a long, cumbersome, and expensive development
pipeline. It only makes sense when the market:
1) is large, monolithic, and valuable (high ASP and high volume) for
individual masksets.
2) slow moving (designs starting now mostly hit market opportunities
that will exist 5 years from now)
3) predominantly values the general purpose computing characteristics
of the product (i.e. its all about the CPUs baby!)
Intel excels mightily in this realm as dozens of dead computer and
MPU companies can attest.
The market ARM swims in is virtually the opposite - slap together
any old craptastic core with all sorts of other things in thousands of
combinations and permutations and get it to market niches ASAP to
enjoy its 6 to 12 month window of sales. Rinse and repeat and live
on bare bones profits or die and have a dozen competitors fight over
your former customers.
Intel failed in mobile because it isn't all about the CPU core, It is a
question of having the right checklist of features at the right time
not promises for next year. To do that you have to turn on a dime,
churn out good enough crap fast, and hope you guessed right.
I suspected that I got something wrong when I typed it. It should have read XScale (which came out of Intel's purchase of StrongARM from DEC).
OK. XScale is very distinct from StrongARM in microarchitecture and
design methodology.
The problem with Xscale was 1) it was designed by "B" and "C" teams
who didn't get support like the guys behind the flagship x86 core design
teams, 2) Xscale was made in Intel's high performance semi processes
which were not appropriate (high power/cost) for most ARM applications,
and 3) didn't offer the customizability of alternative paths to ARM.
I still maintain that the decision to sell XScale to Marvell was an absolute disaster and Intel has lost a whole decade due to that fateful decision.
How? It was a minor side line business at best. Marvell isn't exactly
getting rich selling XScale derived products.
Lost a decade of what? A low ROI rump business that distracts from
the x86 cash cows?
Let me ask anyone who thinks doing ARM chips is some magic bullet
for Intel:
What is the value Intel can bring to a low ASP commodity that is
sold in thousands of highly customized variants? There is no magic
Apple A10 high runner merchant ARM processor socket out there for
Intel to aim for. The A10 brings value because its Apple's in-house
processor for a product whose entire platform from silicon to hardware
to OS is a captive proprietary luxury product that sells billions upons
billions of dollars worth of flash memory to consumers at incredible
markup. Everything else ARM is a commodity with dozens of sharks
fighting over every single scrap and crumb.
There is no "there" there for Intel.
The decision to sell StrongARM was an absolute disaster and IMO, set Intel back by 10 years!
Dude your hindsight is severely broken.
The StrongARM MPU was a technically excellent design - it ran at 3-5 W
and outperformed PowerPC and MIPS processors consuming may times
that. But is was an ARM chip at a time when the market for ARM cores/
chips was for devices in the mW range. The utter vast majority of ARM
customers back then wanted cheap chips and didn't care about the
performance of StrongARM and just wouldn't pay for it. At most they
wanted XX MIPS at $Y dollars, not XXX MIPS at $YY dollars. It was also
an non-integrated MPU with no chipset when ARM customers wanted
(and still do) ASICs and SoCs.
The StrongARM was a technical achievement as a CPU core but an
utter business failure as a product. It was a failure under DEC. It was
a failure under Intel.
Your clouded rose coloured memories of StrongARM as some huge
success just waiting to happen but fumbled by Intel is just a bunch
of crap. Wake up and give your head a shake.
"The former European commissioner Neelie Kroes, who is now a paid adviser to Bank of America and Uber, failed to declare her directorship of an offshore firm in the Bahamas while she was the most powerful corporate enforcer in Brussels."
Vive la difference.
Seriously, this witch should be perp walked straight to jail.
US companies withdraw from European markets. It's an interesting idea though I think that a lot of companies really don't want to leave money on the table.
The E.U. is a pretty big and wealthy market so its definitely not
inconsequential to pull out.
The bigger problem is the vacuum left behind will either be filled
by the Chinese or home grown companies that could grow up to
be strong competitors. Neither outcome is desirable.
However if the E.U. starts imposing fines larger than the benefit
of having a presence in the E.U. then U.S. companies may choose
to leave. It is also clear that the U.S. will have no choice but to
strongly retaliate against E.U. companies selling in the U.S., a
path that quickly spirals out of control globally as trade barriers
go up everywhere and nearly everyone gets much poorer.
To paraphrase Archer, "Do you want a great depression? Because
this is how you get great depressions!"
The strongest take whatever they want from the weakest. To me that sums up our history as mankind - so far at least. I hope for something better.
Judging by the banking bail out after the 2008 mortgage collapse I
would say there is no danger of "better" happening any time soon.
At least these days this basic function involves the pen more than
the sword so there is more of a veneer of civilisation over it.
Lots more Euro banks to go after too.
I think tit-for-tat retaliation between the U.S. and E.U. on each
other's corporations for overblown transgressions and ridiculous
penalties can easily get out control and lead to a major trade war
that could cripple the global economy. So far the U.S. has let
EU crap slide with Microsoft and Intel but seriously going after
google and Apple for huge bucks may be the red line. Really bad
time for Volkswagen to be caught red handed.
Like Hans Solo said, I have a bad feeling about this.
Looks like the U.S. is turning to EU style revenue generation tricks
http://www.economist.com/news/business-and-finance/21707337-fortunes-germanys-largest-lender-go-bad-worse-americas-department
On September 16th Germany’s largest bank confirmed reports that America’s Department of Justice had asked for $14 billion to settle possible claims connected with the underwriting and sale of residential-mortgage-backed securities (RMBSs) between 2005 and 2007.
Ouch.
Or maybe a not too subtle hint to the EU to back off on Apple
retro taxation OR ELSE.
Yeah just 8 years ago the non-x86 server market was about $6B/qtr made
up of ~$5B in Unix servers and ~$1B in mainframes.
Mainframe sales have been fairly steady (although varying with product
release cycles) so it appears that Unix servers sales are on the cusp of
dropping below $1B/qtr with no floor in sight.
I wonder how long IBM and Oracle can continue to throw so much NRE
at RISC Unix processor development for a dying market.
x86 continues to crush all opposition in servers
http://www.idc.com/getdoc.jsp?containerId=prUS41735516
the worldwide server market declined 0.4% year over year to $13.4 billion in the second quarter of 2016 (2Q16)
but
Demand for x86 servers improved in 2Q16 with revenues increasing 7.3% year over year in the quarter to $11.6 billion worldwide while unit shipments increased 2.7% to 2.3 million servers.
for everyone else the crumbs got smaller
Non-x86 servers experienced a revenue decline of 31.1% year over year to $1.9 billion, representing 13.9% of quarterly server revenue.
and now something for the ARM pompom shakers
IDC also continued to track minimal revenue from ARM-based server sales in 2Q16; ARM sales have yet to make an impact on the server market.
I am looking forward to seeing the discussion shift back to INTC rather than other companies.
I am afraid that with the release of Zen based products in the
near future that mention of AMD strategy and competitiveness is
just beginning up again.
To an astounding extent AMD has just been a complete non factor
in the sales and pricing of all Intel products across all markets
for many years. To a large extent ARM had replaced AMD as the
"other" when talking about Intel plans and then only in terms of
attempted penetration into new markets. Intel's main business
was effectively competition-less and dictated entirely by the
health of the x86 server and PC markets themselves.
To some extent this will change over the next year or so if Zen
is half way competitive and is aggressively priced.
Expect a lot more talk about AMD and its products here, not less.
IMO that will be a welcome change, back to the technology dog
fights rather than existential angst and philosophical crap of
bored people.
You mean like your post here?
http://investorshub.advfn.com/boards/read_msg.aspx?message_id=125009056
When the sheep gather it is time for AMD execs to bring out the shears.
This company survives on the effective shortness of the human memory
and inherent laziness to avoid research.
So change the laws and close the loopholes going forward.
Retro-active seizure is the realm of banana republic revolutionaries.
Speaking of Apple... the EU is trying to hit them for $14.5B in "back
taxes".
http://www.investors.com/news/technology/eu-apple-must-pay-ireland-14-5-billion-in-taxes/
The European Union's executive arm on Tuesday ordered Apple (AAPL) to repay 13 billion euros ($14.5 billion) in back taxes plus interest to Ireland, part of a broader crackdown into EU member nations' special tax treatments for big multinationals.
The European Commission ruled that Ireland's tax benefits to Apple amounted to illegal state aid under EU rules.
The executive arm of the European Union Tuesday concluded that Ireland granted undue tax benefits to the U.S. tech giant - which is illegal under EU state aid rules.
yada yada yada big Yankee tech guy pay up big time or else.
Nice to see nothing ever changes in the socialist paradise. I guess
setting up millions of illegal economic migrants in guest housing
isn't cheap. Microsoft, Intel, Google, Apple... who is next?
I wouldn't be suprised if they give up a bit of frequency vs. older designs.
Given the nature of the major architectural changes I would expect Zen to
clock noticeably slower than previous AMD processors despite the shrink.
Zen uarch is much closer to current Intel designs. It will be interesting
to see how close to Intel they can get in frequency and power for the
same number of cores on the chip
Lisa Su: "this is one of those once-in-a-lifetime projects."
It will be for AMD if Zen turns out to be a third steamer in a row.
AMD had to downclock its competitor from 3.2GHz to make the fight fair, but the benchmark as still a notable win
ROFL, if AMD arbitrarily lowers the clock rate of an Intel chip it can even
make Bulldozer seem like a winner in comparison.
Given the huge discrepancy between pre-release hype and post release
harsh look in the sunlight of independent scrutiny for AMD's last couple
of processor architectures I don't think these Zen marketing stunts are
worth a cup of warm spit.
The only reality disclosed is a delay to Zen's release. That is never
a good sign - respin for either a late bug or to desperately tune up
critical paths to squeeze out another frequency bin or drop voltage
and power another 50 mV and 5W to be slightly less uncompetitive.
Isn't the GPU in Kaby much better than the one in Sky?
I think that this agreement is a sign that while Intel has given up on
pushing x86 into smart phones it still wants to get some revenue from
phones while diverting some leading edge process foundry revenue
and profits from TSMC and Samsung. Given the absolute $$$ involved
will be tiny at the top line and invisible on the bottom line I suspect
the latter effect is much more important than the former.
It is a smart move IMO and doesn't impact x86 at all. To the extent
that ARM or any other architecture can encroach on x86 in PCs and
servers (very little would be my guess) that would happen regardless
of Intel fabbing ARM phone chips or not.
I wouldn't expect Intel to ever fab PC or server class ARM chips.
So I guess all we can say with reasonable certainty right now
is a 14 nm Zen core soundly beats a 32 nm bulldozer core even
clocking at lower frequency.
So that is two quad core Intel chips being compared to two octa core
AMD chips right?
I doubt comparisons have any meaning anymore.
Key performance and density metrics are always worth comparing
although it is sometimes hard to get directly comparable numbers
when you parse through the fine print and unstated conditions and
assumptions.
The feature size in the name of processes, particularly from
the big foundry guys, is completely devoid of physical meaning
these days and is basically made up marketing BS.
IBM's Q2
https://www.ibm.com/investor/att/pdf/IBM-2Q16-Earnings-Press-Release.pdf
overall revenue: down 3% YoY
system revenue: down 23% YoY
It's hard to be a non-x86 system seller in an x86 system world.
Intel is pretty much done in mobile SoCs. They flushed their entire pipeline of products there and are reportedly laying off a bunch of people related to the mobile effort.
Why is the overpaid QCOM guy still there?
With no presence in mobile all he can do is f*** up Intel's existing
highly profitable x86 business about which he knows nothing.
If the overall market holds up, I think INTC would hit $40 this year.
Probably too optimistic but not unreasonably so.
Crossing $40 would likely push me to sell off the rest.
Sold half of my INTC holding @$34.71
Gartner: PCs down 5.2% globally, up 1.4% in N.A.
http://www.channelregister.co.uk/2016/07/12/gartner_global_pc_market_q2_2016/?_ga=1.59351726.2085610573.1430316211
Apple sales down 4.9%. Maybe they could actually refresh their Macs
to the latest Intel processors once in a while.
Itanium sales had already started going down in 2009
Hey genius, remember the late 2008 global recession?
All Unix server sales were badly hit - Power, SPARC, and Itanium.
Before this recession Itanium sales were growing faster than Power
and had overtaken SPARC sales for second place behind Power.
Seems like Oracle made the right decision in dropping Itanium support
I strongly disagree.
Itanium was doing fine (as far as any Unix player was doing fine)
until Oracle pulled the critical software rug out from under it in a
flurry of anti-Itanium FUD PR campaign. Once the confidence of
users was undermined it didn't matter that support was reluctantly
reinstated under legal threat, the severe permanent damage was
done.
That being said, once the affair happened HP seemed to want to
maximise the damage to Itanium sales do enhance the award
from its law suit. It did absolutely nothing to promote the huge
benefits of the Poulson processor upgrade of its systems. It
seemed to have done the calculation that with the Unix market in
a secular, sustained decline and buyer confidence undermined by
Oracle it was better to let Itanium sales drop as fast as possible
to be able to demonstrate the largest possible damages to sales
in its law suit.
The bottom line is Unix servers are in terminal decline. The
action by Oracle basically yanked Itanium from an increasingly
strong second place to Power to a distant third place just behind
SPARC with Power way out in front of both. The server market
continues to become more and more dominated by x86 running
Linux, Windows (and minor OSes originated on RISC, IPF, and
mainframes).
This award to HP is a fair and just legal reply to a despicable
act of betrayal and anti-competitive action by Oracle to kill off
a vulnerable architecture that went from strong ally to competitor
after Oracle bought Sun and acquired its SPARC business. It also
appears that there was a strong element of personal revenge
regarding Mark Hurd's ignominious exit from HP to join Oracle.
I hope Oracle shareholders launch a suit against Larry and friends
for such a gross disregard for business ethics and fiduciary duties
in the course of their actions.
What is not clear is Intel's role in the initial action by Oracle.
But that is a story for another day. :-/
Oracle ordered to pay HP $3 billion in Itanium case
https://ca.finance.yahoo.com/news/oracle-ordered-pay-hp-3-011157549.html
SAN FRANCISCO (Reuters) - A California jury ordered Oracle Corp (NYSE:ORCL - News) to pay Hewlett-Packard Enterprise Co (NYSE:HPE - News) $3 billion in damages in a case over HP's Itanium servers, an Oracle spokeswoman said on Thursday.
Oracle said it would appeal the verdict.
The Itaniuum processor is made by Intel Inc (NasdaqGS:INTC - News).
The legal shot Hurd around the world. I guess leisure suit Larry won't
be buying a new yacht this year.
http://www.bloomberg.com/news/articles/2016-06-30/oracle-ordered-to-pay-hp-3-billion-by-jury-for-itanium-damages
The jury saw the case as clear cut and were swayed by what they saw as efforts by Ellison and former HP Chief Executive Officer Mark Hurd to hurt his previous employer, said juror Kyra Knaver, a student at Loyola Marymount University in Los Angeles. Ellison lured Hurd to Oracle in 2010 shortly after Hurd resigned amid an investigation into his relationship with a contractor.
Jurors in state court in San Jose, California, reached the unanimous verdict in less than five hours of deliberations Thursday, concluding that Oracle’s decision to bail out of an agreement to develop software for the Itanium chip hurt HP’s revenue. Oracle said it will appeal the verdict, which is the largest in the U.S. this year according to data compiled by Bloomberg News.
“Oracle’s decision to stop future software development on the Itanium server platform in March of 2011 was a clear breach of contract that caused serious damage to HP and our customers,” Hewlett Packard Enterprise Co. said in an e-mailed statement.
The decision is the second trial defeat for Oracle in about a month after the company lost a $9 billion verdict to Google while trying to stake a claim to the search giant’s Android phone business. The $3 billion award, if ultimately paid, would account for 5.4 percent of Oracle’s $56 billion cash on hand, according to data compiled by Bloomberg.
Forget malware, well-meaning but crap software could prematurely EOL
3DXP in a system that is wired up like DRAM.
You need the chipset and OS to make sure that the only write access
to 3DXP, even if it is on DIMM wired up like a DRAM DIMM, is through
a driver that understands wear leveling etc. Applications which do get
write access have to update it as rarely as possible with write usage
spread out evenly over time. A 100m write cycle cell lifetime sounds
like a lot but it isn't. Reads on the other hand are fine, let them fly just
like DRAM.
No products on the market yet that I'm aware of. This can't be going as planned...
I wouldn't rush to that conclusion.
3DXP represents a new class of beast that is neither fish nor fowl.
If you use it like flash then you are throwing away all its speed.
If you use it like DRAM then you can easily wear holes in it within
seconds.
This means to properly use it you need 3DXP aware chipsets and
OSes to be in place that work well together. Once this is in place
then flagship applications like DBMS likely need new configuration
models and tuning to really take advantage of this new class of thing
that is half way between memory and storage.
This all takes time (especially the software side which will be done
mostly outside of Intel).