Forums

ARM Cortex M3 - Who's utilizing it?

Started by diggerdo February 17, 2006
>> Wilco Dijkstra wrote: >>> Interestingly it turns out Atmel's marketing department has been >>> working overtime - their benchmarking figures are obviously bogus. >>> > Yes, because it's quite brazen and they don't even try to hide it. Do > they really think anyone would take them seriously with such wild > claims?
As always, people will have to make their own tests.
> > The 3x speedup over ARM is complete nonsense. The benchmarks > in question are floating point, and it is hardly surprising a CPU with > an FPU outperforms one that uses emulation. With an FPU the ARM > part becomes 4x faster. Where is that revolutionary performance lead > now? > > The documents don't mention floating point anywhere (no FP > instructions either, only mention of an optional FPU in the user > guide), so it looks like they are misleading on purpose. >
The AVR32 has SIMD (Single Issue/Instruction, Multiple Data) similar to Pentium/MMX and this speeds up things vs ARM in multimedia benchmarks and string instructions -- Best Regards, Ulf Samuelsson ulf@a-t-m-e-l.com This message is intended to be my own personal view and it may or may not be shared by my employer Atmel Nordic AB
"Ulf Samuelsson" <ulf@a-t-m-e-l.com> wrote in message 
news:dtcl1j$802$1@nntp.aioe.org...
>>> Wilco Dijkstra wrote: >>>> Interestingly it turns out Atmel's marketing department has been >>>> working overtime - their benchmarking figures are obviously bogus. >>>> >> Yes, because it's quite brazen and they don't even try to hide it. Do >> they really think anyone would take them seriously with such wild >> claims? > > As always, people will have to make their own tests.
Sure, but overstating things means people will be disappointed. In order to get a design win you generally need to be better than the competition in most areas, eg. power, performance, area, codesize, tool support etc. If all of the advantage is based on a bogus claim then people will start looking more closely at the other areas too... Interestingly eventhough power consumption was mentioned as a key advantage, I couldn't find any estimates. Same for area. I'm wondering why that is?
>> The 3x speedup over ARM is complete nonsense. The benchmarks >> in question are floating point, and it is hardly surprising a CPU with >> an FPU outperforms one that uses emulation. With an FPU the ARM >> part becomes 4x faster. Where is that revolutionary performance lead >> now? >> >> The documents don't mention floating point anywhere (no FP >> instructions either, only mention of an optional FPU in the user >> guide), so it looks like they are misleading on purpose. >> > > The AVR32 has SIMD (Single Issue/Instruction, Multiple Data) similar to > Pentium/MMX > and this speeds up things vs ARM in multimedia benchmarks and string > instructions
You mean like the ARM media instructions? They provide a similar speedup on the ARM because the instruction set is almost identical. Indeed, the ARM compiler uses some of them in string functions. EEMBC doesn't contain any benchmarks that are trivially vectorizable, so you typically only get any gains in hand written assembler code. Although the maximum speedup is 4x, most of the time you're lucky to get 1.5-2x. To get bigger speedups you need to have plenty of registers and use very wide SIMD. Wilco
"Ulf Samuelsson" <ulf@a-t-m-e-l.com> wrote in message 
news:dt9k47$467$1@nntp.aioe.org...
> > Wilco Dijkstra wrote: >>> Don't think any one plans to put a Cortex-A8 in a smart card >>> which is one very obvious application for the AVR32... >> >> I don't think anyone sane is going to put the AVR32 there. As I said, >> it is an ARM11 class core, so totally unsuitable for smartcards >> (it doesn't even have a rotate instruction which is essential for >> cryptography). Maybe there will be a smaller low power version >> eventually but that wasn't mentioned. >> > You have ARM7s in the current high end smartcards. > I believe the AVR32 has the JAVA funcitonality precisely for this > application.
Smart cards are very area and cost sensitive. ARM7 is already big for a smart card, AVR32 must be about 10 times bigger even without caches. Small (16KB) caches typically double the area. Cortex-M3 is smaller than ARM7 and so more suitable for smartcards. Wilco
> Smart cards are very area and cost sensitive. ARM7 is already big > for a smart card, AVR32 must be about 10 times bigger even without > caches.
No it does not have to be 10 x size, and the main reason is because it is not. The Smart Cards have to be 25 mm2 or smaller (due to the size of the cavity) and the AVR32 will certainly fit there complete with memory & interfaces. The Java functionality and MMU (not MPU) is what counts...
> Small (16KB) caches typically double the area. Cortex-M3 > is smaller than ARM7 and so more suitable for smartcards. > > Wilco
-- Best Regards, Ulf Samuelsson This is intended to be my personal opinion which may, or may bot be shared by my employer Atmel Nordic AB
> > The Tiny2313 is a 1.8-5.5V process, so I very much doubt it is 180nm, > more likely 0,35um. Ulf will know ? :) > > -jg
Yes, it is 0.35u -- Best Regards, Ulf Samuelsson ulf@a-t-m-e-l.com This message is intended to be my own personal view and it may or may not be shared by my employer Atmel Nordic AB
>> >> As always, people will have to make their own tests. > > Sure, but overstating things means people will be disappointed. In > order to get a design win you generally need to be better than the > competition in most areas, eg. power, performance, area, codesize, > tool support etc. If all of the advantage is based on a bogus claim > then people will start looking more closely at the other areas too... >
I think this will be sorted out soon enough after the silicon is released. -- Best Regards, Ulf Samuelsson ulf@a-t-m-e-l.com This message is intended to be my own personal view and it may or may not be shared by my employer Atmel Nordic AB
Jim Granville wrote:

> Wilco Dijkstra wrote: > >> Cortex-M3 isn't yet produced in volume, this will happen later this year. >> Test sillicon is available, and if you understand some Russian, you can >> become an alpha tester (http://supplier.ru/rus/news/?action=show&id=7). > > > Interesting - often useful info comes via .ru websites... > I did wonder if Luminary Micro was going to use the Cortex Core. > > Since they are a brand new startup, it would seem most of the > established ARM microcontroller vendors did NOT rush to embrace the > Cortex M3. Hmmmm. > > According to the press, they will start the hoopla on March 6th. > http://www.eet.com/news/semi/showArticle.jhtml?articleID=180203864 > [Just remember, hoopla can preceed full commercial release by 12+ months] > > > Still, a green core, plus a green startup, combined has more risks than > many will tolerate. > Be worth watching... 8-64KF, and 2-6KRam is now crowded space, they will > need something in the peripherals to get attention.
More info is leaking out, prior to their much hyped release (where DO they get the "embedded world will change forever" nonsense) ? http://www.mouser.com/catalog/625/1813.pdf Called the STELLARIS family. Seems they have one of the HIGHEST priced development solutions around. Core Speed ? : 20MHz core ? Code size : Same as LPC2101 series, well, _one_ of the series. 8K is your only choice thus far. [ This is a single sourced core, that needs new tools : so pray your first design always stays under 8K ] ADC performance: OOps, sorry, no ADC. Is this the first new uC to release _without_ an ADC ? Package: Strange choice - a SO28 : Thick, with large PCB area ? Models: Seems you can have 2 Comparators, OR 2 PWM, but not both ? Nothing really stellar in that feature lineup, the LPC210x seems to have it outflanked in all areas. More interesting silicon is the multi-cored Propellor chip at Parallax. -jg
Jim Granville wrote:
> More info is leaking out, prior to their much hyped release (where DO > they get the "embedded world will change forever" nonsense) ? > > http://www.mouser.com/catalog/625/1813.pdf > > Called the STELLARIS family. > > Seems they have one of the HIGHEST priced development solutions around.
..and it seems you need one development kit per variant ?! - Mouser show separate catalogue numbers for the 3S102 and 3S101 - surely that is a mistake ? -jg
> ..and it seems you need one development kit per variant ?! - Mouser show > separate catalogue numbers for the 3S102 and 3S101 - surely that is > a mistake ? > -jg
The development kit has a base that is common to both devices, and a daughter board with the device mounted. I believe the daughter board can be purchased separately. Regards, Richard. http://www.FreeRTOS.org *Now for ARM CORTEX M3!*