The conventional short-term business-school wisdom teaches that the correct response to an economic downturn is to batten down the hatches, quickly jettison large amounts of staff, cut costs to the bare bone, apply whatever financial magic/distortions possible, focus on the core profit centers and do absolutely anything necessary to keep Wall Street on your side. This is actually pretty common in mature low-tech industries – all nice and straightforward in theory.
Other folks with real vision in more dynamic industries also see it as a time to focus and execute, but just as importantly as a simultaneous opportunity to invest for the future so as to come out the other side stronger. Economic downturns tend to weed out the marginal competition and are an opportunity to cement growth into new areas and applications. The paybacks are greatly magnified on the other side. Very smart.
I believe the latter scenario is what we are seeing in this ARM press release from today. Absolutely fantastic!
This development is about enabling our broad partnership to bring our demonstrated model of diversity, creativity, and innovation into brand new high-margin and high-performance opportunities previously occupied by just a few proprietary silicon venders. It is about shattering the barriers to entry at 2GHz, for example. It perfectly complements our existing partner’s strengths and efforts, while sparking imagination and opportunity big time.
What do you think? What opportunities are now open? Comments are always welcome and tend to drive future content, but I should say in advance that I am way too close to this one to do Q&A, so will be mostly lurking.
However, I will collect some press clippings down below over the next few days, so do check to see the latest going-ons.
James Hamilton’s Perspectives Blog: ARM Cortex-A9 SMP Design Announced
The Guardian: ARM Targets Intel with 2GHz Osprey Chip Built for Speed
ZDnet: ARM Targets Intel with 2GHz Multicore Chips
PC World: Arm Flaunts Performance by Boosting Processor Speed
ITProPortal: ARM Unveils 2GHz Dual Core Cortex-A9 Processor
ITPro: New ARM chip promises $100 netbooks
Hexus: ARM takes the fight to Intel
EE Times: ARM announces ‘Osprey’ A9 core as Atom-beater
New Electronics: ARM targets systems with Cortex-A9 hard macros
Electronics Weekly: ARM produces hard Cortex A9 for high performance
Techwatch: ARM sets sights on Intel with new 2GHz chips
TechRadar: 3D gaming capable netbooks due early 2010
ITProPortal: Could ARM’s Cortex A9 Defeat x86 CPUs?
EETimes: Commentary: ARM’s Osprey broadens the battle front with Intel
SiliconStrategies: ARM announces ‘Osprey’ A9 core as Atom-beater
PC Magazine: New 2-GHz ARM Chip Heads for Linux Devices
TechSpot: ARM announces 2GHz dual-core low power CPUs
Maximum PC: ARM Announces Development of 2GHz Cortex-A9 Dual-Core Chip
TweakTown: ARM shows off Dual Core Cortex-A9 running at 2GHz
Mac Rumors: ARM Announces Dual-Core Cortex-A9 Processor Designs
Electronista: ARM develops 2GHz dual-core Cortex A9 chip
[...] ARM at 2GHz! « Eric Schorn’s Processors for People Blog ericschorn.com/2009/09/16/arm-at-2ghz – view page – cached The conventional short-term business-school wisdom teaches that the correct response to an economic downturn is to batten down the hatches, quickly jettison large amounts of staff, cut costs to the bare bone, apply whatever financial magic/distortions possible, focus on the core profit centers and do absolutely anything necessary to keep Wall Street on your side. This is actually pretty common in mature low-tech industries – all nice and straightforward in theory. — From the page [...]
By: Twitter Trackbacks for ARM at 2GHz! « Eric Schorn’s Processors for People Blog [ericschorn.com] on Topsy.com on September 16, 2009
at 11:21 AM
I think that if ARM9 is truly capable of matching x86 Core2Duo/ Core2Quad in raw performance terms, but at much lower power consumption rates, then the next question is how well does it work in a multi-processor environment?
I think that the advent of OpenCL/TESLA is showing a whole new avenue of High-Performance Computing at the desktop/desk-side. As a long term SGI user, I can remember the excitement generated by the idea of a portable SGI workstation (based on mockups of an Indy laptop for a couple of films, including Twister), as these would offer workstation class performance and stability for field scientists, etc. I could forsee a real market for a deskside HPC box running mulitple ARM9 multicore chips and an Open CL setup to add grunt for iterative ciomputation – Cray are using x86 as essentially a switching unit for the multiple NVidia GPUs used as compute cores in its new machines, etc.
As someone who owns a few SGIs, I can say that they suck power at a huge rate, as does the average top-spec PC nowadays (gaming machines are racking up 1Kilowatt PSUs, etc), so if this could run Linux natively and Windows under WINE, etc, then I can’t see why this wouldn’t offer a real boon to the HPC market and the top-end creatives – fast machines that are cheap to run and don’t overheat…what’s not to like? Of course, the advantage is that this would make for a great alternative to x86 in desktop-replacement laptops – Imagine a Dell XPS, etc, that could last for more than an hour between charges!
By: Chance Hooper on September 16, 2009
at 2:12 PM
Interesting that your pitch is on almost exclusively on frequency and you are not highlighting the hard macro deliverable (vs. traditional soft core).
Is that because most analysts are not sophisticated enough to comprehend the implications of am RTL solution vs. GDSII?
By: Werner on September 16, 2009
at 3:06 PM
I’m sure I remember the Archimedes having software that emulated x86 (I think it was 80286) and ran faster than then-current PCs. It could have been speeded up further by re-implementing the OS (I forget whether that was DOS or Windows) in native code, but I don’t think that was ever done.
By: John Grant on September 16, 2009
at 3:43 PM
I recall you innocently asking whether ARM chips will ever pass the 2Ghz mark in a previous post – you young scallywag!
I’m not a designer, I’m not a software engineer – I’m simply an ARM enthusiast. I’ve been waiting patiently since 1988 for ARM to take the battle to Intel. You’ve made me a very happy person today. I am so excited.
I hope ARM will release lots of specs, comparisons against Intel chips, and the allike so you maximise the amount of coverage from this annoucement.
I have a news site dedicated to ARM technology at http://www.nytrex.com. Feel free to drop by sometime. I’d love to do an interview with you in the future sometime.
By: ARMnews on September 16, 2009
at 6:10 PM
Eric, you’ve fallen into the classic marketing philosophy that the vast majority of technology companies fall into: that “speeds and feeds” or “bits and bytes” or “name your technical feature” are what really matters. Nowhere in your press release or messaging do you talk about the real problem to be solved, the benefit to the end consumer or how this will help your licensees increase market share and accelerate their time to revenue. In other words, how does ARM at 2GHz help my customers solve their problems, sell more product in the market and, in turn, make more money for ARM?
As a former product manager for a number of ARM based SoC’s, I can tell you that the biggest obstacle to success in certain markets is the ultimate user experience which is ultimately influenced much more by the software than it is the hardware. If your user experience is inferior to that of your competitor (i.e. x86) then I don’t care how fast you run, or how long your battery life is. Apple is the classic example of a company that does not promote the technical aspects of their offering, instead they create value by developing the world’s best user experience and marketing the heck out of it.
So, while I agree with your initial premise that it is critical to invest in the future, you need to make sure that you invest in the appropriate place. Only time will tell if ARM was better off investing in perceived hardware benefits or if an equivalent investment in enhancing the user experience on ARM platforms would have a greater benefit to your customers.
By: Brad Hale on September 16, 2009
at 9:24 PM
Brad, May be the “real problem to be solved” is not well defined yet. May be the increased processing power will lead designers to approach problems that they could not think of in the past, or may be revisit something they shelved for lack of enough processing power.
Hardware improvements abet software innovation.
An example: The idea for a PMP was reportedly in the works long before Apple released the first iPod. They had to wait until Toshiba came up with a small form factor hard disk to solve the capacity constraint.
You point out that ” Apple … does not promote the technical aspects of their offering”. The virtues of this approach is debatable. Apple however shouts at the top of its voice that iPhone 3GS is twice as fast as previous version. The speed improvement is, partly, courtesy of a faster ARM core based processor.
Even Apple couldn’t have made an iPhone 3/4/5 years ago. The hardware was not fast enough at such low power levels.
By: Nalini Kumar Muppala on September 17, 2009
at 1:08 PM
Brad’s right to point out that processor speed does not map directly onto user QoE. I remember some 30 years ago putting together a chart of clock speed and execution speed of some dozen different microprocessors; the one with the fastest clock executed fewest instructions per second, and vice versa. And then, of course, different numbers of instructions are needed to perform the same task on different processors. So a 2GHz ARM might be faster or slower than a 2GHz x86, or indeed faster for some apps and slower for others.
Increasing processor power may or may not drive software innovation. It may make possible a task (such as, for instance, decompression of video in real time) that couldn’t be done on the slower processor. On the other hand, it allows programmers to be lazy, letting the increased speed improve QoE instead of looking for more efficient algorithms.
By: John Grant on September 18, 2009
at 5:45 PM
I just want ARM to deliver enough speed to run a full browser with unlimited amounts of tabs, full flash, HD video, full multimedia, and be able to run cloud applications in ways that all these applications need to feel at least as smooth as using Intel.
And the most important thing would be to achieve that “performance” level at the lowest possible cost to the end consumer.
Then I’d like to see “Moore’s law” applied on ARM processors, to simply lower the costs and improve battery life further every 18 months and keep that same performance level.
I want the “performance” of tomorrow to really come from the broadband Internet. For example I believe HD video editing, HD video encoding, HD photo editing, sound editing, 3D animating all these things, even playing advanced 3D games, all those things can be best powered by the cloud. And I just want a thin client that really works, even offline.
Make 18-tabs Google Chrome work on a $100 ARM powered Laptop with just 256mb or 512mb RAM (don’t need to call it Smartbooks/netbooks), full Flash, Java, and other scripts support. Thanks for your work to make this happen!
By: Charbax on September 20, 2009
at 12:11 AM
[...] [Many folks seem to be arriving here while looking for the ARM at 2GHz post. If that's you click here, then quickly come [...]
By: IDF questions « Eric Schorn’s Processors for People Blog on September 21, 2009
at 6:27 AM
[...] I am hiring! [It appears many folks are arriving here while looking for the earlier ARM at 2GHz post. If so, click here] [...]
By: I am hiring! « Eric Schorn’s Processors for People Blog on September 27, 2009
at 6:36 AM
Hi Eric,
Loads of questions for you.
what is the power consumption of the core when it is running at 2 GHz? Were these fast parts from the fab? did they use dynamic logic? Was it just 1 part coming back from the fab working at 2 GHz out of 40.
when arm sells soft IP, whats preventing the rest of the world (TI, Samsung) from getting here in 45nm technology? why are they struggling? Does ARM have plans of becoming an SOC company soon?
These are huge companies with loads of engineers and a huge talent pool. why are they still @ 1GHz? may be I need to go ask them, but it looks obvious that they havent been able to push the frequency up very much.
what good is just a standalone 2GHz processor core if it cant go into an SOC used in a handheld?
By: Nick on October 23, 2009
at 1:36 PM