IBM Microelectronics – the chip designing and wafer baking division of the IT giant – has inked a deal with ARM Holdings to help make sure those who license ARM chips have the processes and fabs to make them. ARM Holdings has been collaborating with IBM Microelectronics since 2008, and it has used IBM's 32 nanometer and 28 …
ARM at 28NM?
The current ARM chip in my Samsung Epic is a Hummingbird at 45nm. It has the Imagination Technologies SGX540 GPU (with GPGPU capabilities) with four graphical pipelines. It runs Unreal Engine 3, a full-scale immersive 3D engine and the video in games like Dungeon Defenders is not to be believed on a phone. It's only a few months old.
28NM brings the same experience at less than half the watts, or photorealistic gaming on a tablet with six times the cores. Think Crysis - on a tablet or phone, all day on a long flight without recharging. Three days of HD video on a slate before the battery runs out, or Citrix if you prefer that.
If they can do this in mobile, they can do photorealistic ray-tracing rather than compositing in high-power desktops.
The third world? They've been wanting to join us online and add their value to the Net, but they don't have Watts. This would do it for them.
And they're embedding this in TV's? Vizio sounds like they're playing that game. A whole bunch of stuff is about to change. This is big. Huge. Holy Cow, the mobile revolution is upon us. The times, they are a-changing.
ARM signing an intel 22 nanometre would be a very good thing to
ARM signing an intel 22 nanometre contract would be a very good thing for al ARM licensees to, after all .....
Intel are now opening up their advanced plant to other's now, such as Intel to fab FPGAs for startup Achronix etc
10/31/2010 9:58 PM EDT
SAN FRANCISCO—Intel Corp. will lend its semiconductor process technology muscle to build FPGAs for programmable logic startup Achronix Semiconductor Corp. at 22-nm and beyond under the terms of a strategic agreement between the two companies announced Monday (Nov. 1).
Executives from Achronix (San Jose, Calif.) said the deal would not only help the company bring 22-nm FPGAs to market faster than programmable logic market leaders Xilinx Inc. and Altera Corp., but also give the startup a leg up in 15-nm and future technology nodes.
"If you look at history, Intel over the last 11 process generations has done a new process successfully every two years," said John Lofton Holt, Achronix founder, chairman and CEO. "We believe that trend will continue, and we are going to benefit from that....."
"Achronix said it would have engineering samples of its 22-nm devices, Speedster22i, by the fourth quarter of 2011, the same time that Intel is set to be in production on its own 22-nm devices. Because of Intel's capacity and process technology, Holt said Speedster22i could be in volume production at the same time.
The Achronix parts will be built with the same 22-nm process technology that Intel will use for its own chips. Intel's 22-nm process technology will include high-k metal gate, a technologically difficult trick which Intel has been employing since the 45-nm node but that other firms have been struggling to perfect......"
talking about "eDRAM takes about 60 per cent less space on a chip than static RAM"
how Exactly does the old Freescale MRAM tech and chips fit into this die shrink etc picture Today ?
an update would be nice on tech IP you can actually buy and combine into a self contained ARM SOC/FPGA etc
Freescale first to market with MRAM chips
By Thomas Ricker posted Jul 10th 2006 8:31AM
Remember those new MRAM (magnetoresistive RAM) chips the industry's been working on for the last decade or so?
Well Freescale stepped away from all that UWB bidness long enough to develop what they're calling the world's first commercially available MRAM chips to combine RAM's endurance with the hard disk's ability to keep data while powered down.
What about Flash memory you ask?
Kicked to the curb son, since MRAM (or universal memory) is faster than flash and doesn't degrade over time.
Only don't expect these new chips to hit your price sensitive consumer electonics just yet.
The first markets for MRAM will most likely be automotive or industrial settings where durability is critical. So hold tight and let the invisible hand work some magic, mkay?"
What about Flash memory you ask? Kicked to the curb son
wont anyone in the know, some 'investigative journalist' reading , or at least an 'elreg hack' ;D type comment on, and give updates as regards were this Freescale MRAM might stand in today's market and core shink etc!
interestingly to some when put into context about those faster 28/22nm coming as regards FPGA above and the latest Intel SOC.
oh that would be http://www.eetimes.com/electronics-news/4210937/Intel-rolls-six-merged-Atom-FPGA-chips
Intel rolls six merged Atom, FPGA chips
apparently "The Intel Atom Processor E6xx and E6x5C Series Hands-On Training Programme is coming to a city near you! " has now started.
perhaps they are going for the big build up the so called ‘reconfigurable computing’ that’s now back in vogue ? as a side spec to the "ARM"'s race about to start this year :S
FPGA co-processors should have been around years ago....
For instance for hardware decoding. You have an H264 video to decode?, load the FPGA with a hardware decoder for H264. Next video is a WebM, re-program the FPGA with a WebM decoder. Running a specific app that needs a certain type of hardware acceleration? All you would need is an FPGA image for each task, and load the FPGA at run time (takes a few hundred ms for a large FPGA). Each image has a footprint of 10MB for a small FPGA to 100MB plus for a large one, so a few GB of FPGA images could do a lot.
have you seen the price of a truly worthwhile truly fast FPGA?
Unless your application happens to fit conveniently on a (current) low end FPGA, we're talking hundreds of pounds for a decent sized decent speed FPGA before adding any bits needed to integrate it in your system (PCB, programming tools, whatever). If your app is a good fit for FPGA, FPGAs are great, otherwise today's processors are generally so fast that FPGAs simply aren't cost effective. Looks that way from here anyway.
Yes, I do know the cost
There are low cost versions in the low 10's of dollars range that are suitable for most tasks, including video decoding (Xilinx Spartan devices for instance). You do not necessarily need a 'fast' FPGA when you can do the processing in parallel, rather than instruction by instruction.
Even these low end devices come with IO that can cope with PCI express etc so connectivity is not a problem either.
blah blah BOM cost, blah blah lower bulk cost ,blah blah
see this is the problem with elreg, your users non clickable url's dont make it easy for people to click and go there so you ether include all the url data (not really an option usually) or people dont see the data before posting
"The high–end chip uses a 1.3 GHz Atom with a 400 MHz graphics block consuming 3.6W and costing $106. The low-end device runs at 600 MHz with a 320 MHz graphics block, consumes 2.7W and costs $61."
OC perhaps your "FPGA, we're talking hundreds of pounds" is including all the board support package, and the per seat costs shite that currently comes with it .....
OC in the old school low volume unit's BOM costs thinking, perhaps have a point.
this is not the old school ‘reconfigurable computing’ thinking though , if people actually read all 3 gages of that "Intel to fab FPGAs for start-up Achronix" URL posted;
you might see that they say large cost savings, consumer markets, and here's the real BIG hint
"Possible game changer For Achronix, the deal with Intel could be very significant. By beating Xilinx and Altera to the punch at 22-nm, executives say the Speedster22i family could put the company in position to compete for many more design sockets that currently go to standard cell ASICs."
that's " 'compete' for many more design sockets that currently go to standard cell 'ASIC's." to super highlight the point .
its all about the costs, but the end user pays the real costs in the end, and as proven time and again, mass market world volumes bring down these real costs for the OEM making the PCB.
if the old school FPGA companies like Xilinx and Altera cant or wont do this FPGA for mass market ‘reconfigurable computing’ deal at a good bulk price then perhaps this new Achronix company with their far faster and far cheaper 28nm/22-nm products will!
"Achronix said the Speedster22i family would establish new standards for FPGAs in terms of size, performance, cost and power consumption.
The family will offer devices with up to 2.5 million look up tables (LUTs), twice as many as any other FPGA, Achronix said. Because of Intel's process technology, the devices will also offer a cost advantage of more than 40 percent, according to the company. Like the previous generation of Speedster, the devices will also offer a peak performance of 1.5 GHz, 300 percent faster than other FPGAs, and consume 50 percent lower power, Achronix said. "
who knows, but dont discount an option just because of the old school BOM cost mindset.
blah blah BOM , blah blah lower bulk cost ,blah blah, i cant do it so no one ever will....profit.
"video decoding (Xilinx Spartan ...)
I'm familiar enough with Spartan (and Virtex too) thank you. But back to video decoding.
If the processor is fast enough, you do the decode in software, regardless.
If the processor is not fast enough and the volume is worthwhile you put the decode in an ASIC, ideally in the SoC. If the volume is not worthwhile for ASIC/SoC, or the ASIC/SoC is not flexible enough, then it's FPGA time, and that applies whether the application is video decoding or any other readily hardware-acceleratable application.
FPGAs are inevitably in a niche in the middle, squeezed every time processors get faster and every time ASIC/SoCs get cheaper. FPGAs are amazing technology, amazing power for the price, and there are some places where they are an obvious fit. But not, I submit, in mass market consumer electronics manufacture. Prototyping or moderate volume manufacture where the one off costs of going ASIC are not so easily recoverable is a different thing.
Just my 2c.
My EEEPC901 netbook can struggle with video playback, depending on the video encoding type. Since there lots of types of video encoding, are you going to put a decoder for each type of encoding into an ASIC or GPU? What about when a new type of encoding comes around, like WebM? If you have an FPGA co-processor you write/synthesize/place and route a new decoder for WebM, stick it in a repository, download it from a package manager just like you would with Linux apps, and hey presto, you have WebM hardware accelleration. From a $20 FPGA.
So FPGA's are not used in mass market consumer apps then? Have you had a look inside a set top box? Are they mass market? Are there no FPGA's in cars? are they mass market? Plenty more examples too.
With the greatest possible respect
"EEEPC901 netbook can struggle with video playback"
Of course it does. My 486 struggles too. So what. They're both obsolete; mass market consumer electronics is a throwaway market (maybe it shouldn't be, but it is).
"are you going to put a decoder for each type of encoding into an ASIC or GPU? "
That's exactly what the manufacturer may do, if the cost saving per unit over hundreds of thousands (millions?) of units makes it worthwhile. Or if it's a closed market (DVB-S, DVB-T, etc) they ignore the new encoders. How many kinds of HD encoder do we need anyway; sooner or later someone will notice that in high action high detail stuff there's no value in compression, you need bandwidth, and if there's enough bandwidth who needs more than basic compression anyway.
Who cares if the punter's original kit is obsolete, some other sucker will be along in a few minutes to buy a new upgraded one.
"From a $20 FPGA."
Or a respin of the $5 ASIC/SoC? The ASIC/SoC is a fraction of the FPGA price but a respin has a high one off cost. So at low volumes FPGA wins. At high volumes the ASIC wins. In the middle, it may be more interesting.
"So FPGA's are not used in mass market consumer apps then? Have you had a look inside a set top box?"
Have YOU had a look inside a *volume production* set top box? FPGAs let a manufacturer get an early prototype out the door to take round the cablecos, telcos, etc, tweaking it as they go along, but if the aim is profitable high volume production on a sustained run, the design goes to ASIC/SoC as soon as it makes financial sense. Feel free to show examples to the contrary.
There will be cases where FPGA wins, there will be cases where SoC wins. One size does not fit all. The numbers tell all, otherwise you're more expensive than the competition (and how many people are *really* prepared to pay extra for future-proofing?)
This is the exact same cost-driven logic that led Linksys to switch from Linux to VxWorks in some of their routers and stuff. VxWorks costs money, and also costs up front development to port existing code. Linux needs more ROM and RAM than VxWorks so the chips cost more per box. If the volumes are high enough, the one off costs of VxWorks are exceeded by the saving per box on ROM and RAM.
we take your points about the old world order and low volume
we take your points about the old world order and low volume and cost etc
however are you ignoring the clear message above that
the Achronix executives say they intend their new style FPGA
with 3 times the speed,
2.5 million look up tables (LUTs), twice as many as any other FPGA,
50 percent lower power usage, and
will also offer a cost advantage of more than 40 percent,
Are/will be going after the Massive ASIC markets as a priority , so they most feel they can meet the ASIC price points today and that's before you take into account the mass production cost savings as they grow their markets.
that's the key point really , How do you grow your low volume market share and world profits! without new thinking and feeding new markets you can grow with...
Achronix executives seem to imply they can and will do exactly that in 2011/12 and beyond as part of their Intel plant plan's
also its seems clear that most of the world's fab plant are looking find new ways to fill their various future production capability's to near 100% , hence Intel opening up etc
"Good news, bad news
This year, the IC equipment business is expected to hit $46.3 billion in terms of sales, down 5 percent over 2010, according to the firm. In 2010, the business hit $48.8 billion, up a whopping 99.7 percent over 2009, according to the firm."
I wouldn't say I'm ignoring Achronix
I'm just bearing in mind that it's an Intel-backed PR puff (even though it's not funded directly by Intel Capital).
If even 50% of Intel-backed futures stuff delivered 50% of what was promised, Intel wouldn't still be 99.3% reliant on x86, with a record of non-x86 failures that goes back a decade or more.
If Achronix/Intel do deliver what the Stellarton (?) concept promises, everybody wins (probably even today's FPGA vendors and customers).