Electronics Weekly Electronics Design & Components Tech News
ARM’s vp of investor relations told a meeting in Taipei that ARM’s market share was now less ghan 1% and Intel’s share of the server chip market is 90%.
He said that Intel’s ASP for a server chip is $600 while the cost of an ARM server chip is $100 to $200.
Moor Insights say that, over three years, the cost of buying and running ARM servers could be 35% less than x86 and they use two thirds the number of physical racks.
Canaccord reckon ARM server CPUs will have a 20% share of the server chip market by 2018.
ARM saw server chip revenues for the first time last year.
HP brought out two ARM-based servers – the ProLiant M400 with a 2.4GHz ARMv8-based Applied Micro X-Gene eight-core SoC and M800 which has a 32-bit Texas Instruments quad-core 1GHz ARM Cortex-A15 with eight DSP cores.
The M800 is aimed at “businesses wanting VOIP/LTE, seismic processing and/or real-time video transcoding while saving on energy, space and cost,” says HP.. Customers for it include the US Department of Energy’s Sandia National Labs and the University of Utah. The M400 is for tasks like web caching and PayPal is one customer for it.
Four chip-makers have announced 64-bit ARM v8 chips to go into the server and networking markets.
The Big Beast is the 48 core v8 64-bit 2.5GHz Cavium processor called ThunderX. “ThunderX will enable Cavium to be the first ARM-based vendor to deliver the performance and features required by today’s volume server market at half the power and significantly lower cost compared to competing solutions,” says processor guru Linley Gwennap.
As well as ThunderX, there’s Applied Micro’s eight v8 core X-Gene processor and AMD’s eight v8 core Seattle processor.
In January, Iliad, the French provider of telecommunications services, launched a cloud service based on ARM-based servers which it has built in-house using Marvell chips.
18 servers – each with a quad-core CPU – can fit onto a blade, and 16 blades can fit in a chassis.
Well only one of the people you mentioned worked in in the semiconductor industry, Mike, and I assumed it was the semi industry we were talking about
Well I didn’t notice any of the people I mentioned having MBAs either and their technical judgement skills are exceptional.
Many many years ago I asked Gordon Moore this question Mike and he said that it was then too early to hand the industry over to the MBAs and accountants because of the technical judgment involved. Whether it is still too early I don’t know, but the fact that Intel still can’t make cost-effective mobile SOCs and have to put them out to TSMC indicates that the technical judgments may have been poor.
Why is not having a Ph.D an issue ? I agree all the Intel founders had them but maybe that was just what led to them meeting up with each other. Certainly Bill Hewlett and Dave Packard didn’t too badly despite the ‘handicap’ of only having first degrees, whilst Bill Gates and Steve Jobs never even finished their first degrees before becoming successful. And the guy I think should run Intel, Jen-Hsun Huang, got his Ph.D as an honorary award which seems a better way to do it to me.
Indeed based on this shortlist it seems success is almost inversely proportional to your qualification. What matters is the drive to succeed
All very true, Stooriefit and one might add:
8. Of the 15 top-ranking execs, none has a PhD.
The deep pockets thing is interesting. Intel on the face of it are definitely the 500lb gorilla, but:
1. Debt vs cash they are nothing like as comfortable as e.g. Qualcomm.
2. They have an ISA which, whilst widely used, no one would say is exactly elegant, which carries an implementation overhead in terms of speed, efficiency and area.
3. They are on a competitive treadmill which requires them to innovate on the process and architecture simultaneously, unlike their competitors who pick off one and leave partners to pick up the other.
4. They have made extraordinary efforts to diversify and failed over an over again.
5. No matter how big IOT gets they are not going to be able to fill their fabs on server chips alone, and they only compete on cost in low power if they use outside fab capacity. Something (expensive) will have to be done about this.
6. They have a reputation for reaming industries where they have been able to develop dominance.
7. The founders have gone, but they are competing with companies where the founders still run the shop.
That’s a very interesting point, Pedro, we always think of Intel as having deep pockets but its cash of $14bn is almost matched by its debt of $12bn. More importantly, once Intel starts relinquishing its cherished 60% margin, investors will start getting arsey. This could be a fascinating play.
David, If that price-war starts. Who do you think will run out of money first ???
One of the ARM Server bit-players like Cavium or Applied Materials or Intel ??
And do you think their venture-capital investors signed up for a price-war ?? When do you think the penny will drop there and “Calxeda” repeats itself ??
Do you think, when you search your soul, that AMD’s powerpoints bear any relation to reality ?? Do you?? Really ??
So the future of “ARM Server” is in the hands of people with deep pockets – Qualcomm and the French Government, don’t you think.
The only ‘definite’ I can see here is a “plague of lawyers” in about 3 years time 🙂
Well of course they can go in for a server chip price war, Pedro, but with a loss-making mobile business to support and a shrinking PC business, they’re going to crater their share price.
David , “The Play” is still in progress (as evidenced by the recent release of new Atoms on TSMC’s process in order to ensure the modems work and are certified) So the jury hasn’t yet been sent to consider what will happen when those modems are on 14nm and integrated with Core-M and “the empty fab” in Arizona gets its tools.
And do you really think that Intel (an outfit that will sell you a Bay Trail Celeron notebook part for $15 with no “contra revenue” in order to kill the rump of AMD) won’t stamp on the first ARM server part with a $15 “Denverton” ?
Trust me: They will!! They will “eat grass” to combat the pincer movement from “Open IBM” and “ARM server”.
Well Yes Mike but, at that time, Apple had 0% of the phone industry and Nokia had 50%+ and so Nokia called the shots on which processor the mobile industry would use. And Nokia did not want the mobile industry to be screwed like the PC industry was screwed.
You’re absolutely right, Pedro, Intel’s marketeers have gone to extreme lengths – $4bn a year lengths – to get mobile manufacturers to want to have it done to them but there’s little sign of them wanting it. And I wonder if hell will freeze over before Intel sells a server chip at cost.
David, The job of Intel’s marketeers is to make the device manufacturers WANT to “have it done to them”.
I worked at Intel from 1983 to 2003 and I saw how it was done in PCs. Compaq and their brothers fought very hard to avoid it but were eventually overwhelmed by Intel’s “Systems Division” with its 4 PC and motherboard factories and their army of tame ODMs (A little “too tame” if you believe the EU justice system).
The same play is being made in mobile at present except the factories are external and the new Intel management are deliberately using the term “contra revenue” and its obvious costs to punish the BOD and remind them of the penalties of their lack of “strategic vision”.
I believe the endless stream of “ARM Server” propaganda is intended to deliver false reassurance to investors who clearly don’t have a snowball’s chance in hell of getting their money back if Intel decides to sell parts like THIS (The world’s most efficient server chip) at cost.
http://ark.intel.com/products/77985/Intel-Atom-Processor-C2730-4M-Cache-1_70-GHz
Cost being about $10 (it being 100sqm of fully depreciated silicon)
Not quite – Apple went to Intel for their first mobile processor and Intel turned them down because volumes would be too low !
One of the worst decisions in business history.
I”m not sure it was so much that Intel didn’t see it coming, Pedro, I think it was more because the mobile phone manufacturers did not want done to them what Intel did to the PC manuacturers.
My point being, David, that ARM ate Intel’s lunch in mobile because Intel didn’t see it coming.
Whereas now :
http://www.anandtech.com/show/9070/intel-xeon-d-launched-14nm-broadwell-soc-for-enterprise
My faith would be restored if Linley Gwenapp would release a projection of ThunderX performance per watt that didn’t simply extrapolate from AMD’s Seattle claims and use some arbitrary fudge factor that he pulled from some hidden orifice 🙂
http://www.cavium.com/pdfFiles/ThunderX_Rattles_Server_Market.pdf
Page 3 halfway down on the left.
O ye of little faith, Pedro.
This would be the same ARM servers that will have “20 percent of the market by 2016 or 17”, right??
http://www.pcworld.idg.com.au/article/465171/amd_reboots_server_strategy_first_arm_chips/
1/3 down the page.