PIC vs AVR vs ARM
in [Embedded]
|
From: An Schwob in the USA on 7 Oct 2006 00:49 On Oct 3, 1:56 am, Joseph <joseph....(a)somewhere-in-arm.com> wrote: > Ulf Samuelsson wrote: > > > LMI make Cortex chips which are incompatible with most of the others. > > Apparently they are financed by ARM themselves. > > I guess that is one reason why the uptake is not dramatic.Hi Ulf, > > You might have been seriously misinformed :-) > LMI is not financed by ARM. We are two different companies, and LMI > is a ARM partner. Sure different companies but almost as sure ARM puts in a lot of money into Luminary. How could they otherwise survive? Having devices that use the lowest power ARM CPU yet they are much higher current than SAM7 or LPC2000. THe marketing gimick of the 99 cent device which I yet have to see available anywhere is a 20 MHz max 28-pin device. You get an LPC2101 for a $1.47 @ 10k that runs 3.5 times faster, uses a lot less current at the same speed, offer more peripherals and a nice upgrade path that does not multiply the price just because the next device runs a whopping 25 MHz. This company (Luminary) can not really make any money with the devices that lack performance, are high power and can not keep up with the wide range of available ARM7 devices, so they must be sponsored and who but ARM would be interested to finance it? Talking about ARM partner, the only thing that I see Luminary attacking are other ARM partners (canibalism!?) They should compare themselves to the PIC24, dsPIC, AVRs (sorry Ulf), ST10, HC12, S12, MSP430..... you name them. An Schwob
From: rickman on 7 Oct 2006 12:39 An Schwob in the USA wrote: > On Oct 3, 1:56 am, Joseph <joseph....(a)somewhere-in-arm.com> wrote: > > Ulf Samuelsson wrote: > > > > > LMI make Cortex chips which are incompatible with most of the others. > > > Apparently they are financed by ARM themselves. > > > I guess that is one reason why the uptake is not dramatic.Hi Ulf, > > > > You might have been seriously misinformed :-) > > LMI is not financed by ARM. We are two different companies, and LMI > > is a ARM partner. > Sure different companies but almost as sure ARM puts in a lot of money > into Luminary. How could they otherwise survive? Having devices that > use the lowest power ARM CPU yet they are much higher current than > SAM7 or LPC2000. You are assuming facts not in evidence. If you read their web site you will see that they just completed a second round of financing... "Luminary Micro closed its first private funding of $5M in February 2005 with lead investor EXA Ventures. Luminary Micro also closed a Series B of $14 million as announced on June 12, 2006. " I am sure they are getting a ton of support from ARM and they may have gotten their initial funding in an indirect way from ARM. I have seen startups get free office space, free use of facilities and even the loan of personel from interested parties without showing it as "financing". But I have no doubt that at this point LM is self sufficient and will be turning a profit in the next year or so. > THe marketing gimick of the 99 cent device which I yet have to see > available anywhere is a 20 MHz max 28-pin device. You get an LPC2101 > for a $1.47 @ 10k that runs 3.5 times faster, uses a lot less current > at the same speed, offer more peripherals and a nice upgrade path that > does not multiply the price just because the next device runs a > whopping 25 MHz. I don't agree that the $.99 LM3S101 is a "gimick". At that price point there are any number of sockets this device can steal from the 8 bit world. I seem to recall a PIC clone that ran at 50 MHz and emulated a lot of peripherals in software. Clearly there is a place for higher horsepower at the low price end of the market. These smallest parts are clearly aimed at the automotive market. If you don't think this market is significant, ask TI (whose ARM parts were only for automotive customers until a year or so ago) and Micronas who makes several ARM parts that they target to the automotive world. Neither TI nor Micronas make low power devices since this is not the factor in this segment that it is in other markets. > This company (Luminary) can not really make any money with the devices > that lack performance, are high power and can not keep up with the wide > range of available ARM7 devices, so they must be sponsored and who but > ARM would be interested to finance it? Of course they can make money. The claim of lacking performance is not in any way justified. Their core can run at 50 MHz from Flash with no wait states regardless of the code being executed which no other ARM part can do. The Philips parts use a lookahead buffer, but when a jump is executed you get a multi clock cycle delay while a new Flash access is started. The high power claim is a bit exaggerated. LM parts use more power than the Atmel or Philips parts, but they are in the same ballpark as the ADI, ST and TI product lines. Are you suggesting that none of these four companies have a chance of competing in the ARM market place? > Talking about ARM partner, the only thing that I see Luminary attacking > are other ARM partners (canibalism!?) I'm not sure what that is supposed to mean. All ARM vendors compete against one another. > They should compare themselves to the PIC24, dsPIC, AVRs (sorry Ulf), > ST10, HC12, S12, MSP430..... you name them. I guess you are referring to their marketing. My guess is that currently there is a lot of market focus on moving to the 32 bit world. There are any number of articles in the trade journals about how users are skipping 16 bits and moving directly from 8 bit parts to the 32 bit devices. In that context the smart marketing is to use that momentum as leverage and show your advantages over the other 32 bit devices. I guess your requirements exclude the LM ARM devices. But that does not mean they are making inferior products. Different customers have different requirements.
From: Wilco Dijkstra on 7 Oct 2006 15:38 "An Schwob in the USA" <schwobus(a)aol.com> wrote in message news:1160196565.981433.178410(a)i42g2000cwa.googlegroups.com... > > On Oct 3, 1:56 am, Joseph <joseph....(a)somewhere-in-arm.com> wrote: >> LMI is not financed by ARM. We are two different companies, and LMI >> is a ARM partner. > Sure different companies but almost as sure ARM puts in a lot of money > into Luminary. How could they otherwise survive? ARM did invest a small amount, but only a tiny amount of what a startup needs. They can survive because they will make money once they get to volume production. Even with a tiny profit per chip selling millions of them means lots of money. > Having devices that > use the lowest power ARM CPU yet they are much higher current than > SAM7 or LPC2000. The initial devices are on an older process than the Atmel and Philips devices. Even so, since the M3 is a lot faster, it doesn't need to run at the same frequency as other MCUs so it may actually use less power. > THe marketing gimick of the 99 cent device which I yet have to see > available anywhere is a 20 MHz max 28-pin device. You get an LPC2101 > for a $1.47 @ 10k that runs 3.5 times faster, uses a lot less current You're falling into the MHz = performance trap. An M3 is about twice as fast as ARM7 MCUs running Thumb code (it runs Thumb-2 code at the same efficiency of an ARM9 running ARM code). That means you don't need to run at a high frequency to get the same performance, and you use less power due to the lower frequency. A 20MHz Cortex-M3 is faster than just about all existing 8/16-bit CPUs, including for example a 100MHz single cycle 8051. And the 50MHz version outperforms current ARM7 MCUs by a huge margin. > This company (Luminary) can not really make any money with the devices > that lack performance, are high power and can not keep up with the wide > range of available ARM7 devices, so they must be sponsored and who but > ARM would be interested to finance it? Maybe it's a conspiracy? > Talking about ARM partner, the only thing that I see Luminary attacking > are other ARM partners (canibalism!?) On the contrary, Luminary is aiming their chips primarily to people who upgrade from 8-bit and 16-bit chips when they run out of steam. Other manufacturers can and do make cheap ARM MCUs, competition is good. What evidence do you have Luminary is "attacking" ARM partners? > They should compare themselves to the PIC24, dsPIC, AVRs (sorry Ulf), > ST10, HC12, S12, MSP430..... you name them. Existing ARMs outperform most of those already by a large margin. In cases where they don't, the M3 does now - the M3 exists solely to beat the existing 8/16-bitters on every aspect. As it happens it does beat ARM7 as well, but then again ARM7 is over 10 years old... Wilco
From: rickman on 7 Oct 2006 17:21 Wilco Dijkstra wrote: > "An Schwob in the USA" <schwobus(a)aol.com> wrote in message > news:1160196565.981433.178410(a)i42g2000cwa.googlegroups.com... > > > > On Oct 3, 1:56 am, Joseph <joseph....(a)somewhere-in-arm.com> wrote: > > >> LMI is not financed by ARM. We are two different companies, and LMI > >> is a ARM partner. > > Sure different companies but almost as sure ARM puts in a lot of money > > into Luminary. How could they otherwise survive? > > ARM did invest a small amount, but only a tiny amount of what a startup > needs. > > They can survive because they will make money once they get to volume > production. Even with a tiny profit per chip selling millions of them means > lots of money. But they have to get to the millions level pretty quickly to make that work. Meanwhile they have completed two rounds of financing. > > Having devices that > > use the lowest power ARM CPU yet they are much higher current than > > SAM7 or LPC2000. > > The initial devices are on an older process than the Atmel and Philips > devices. Even so, since the M3 is a lot faster, it doesn't need to run at > the same frequency as other MCUs so it may actually use less power. That is not at all clear. "May" is a word that has been used a lot with these parts and I am still waiting for the evidence. At a 2:1 power ratio for the max speed and with the other parts running at higher clock rates, the LM parts need to do a lot of catching up. > > THe marketing gimick of the 99 cent device which I yet have to see > > available anywhere is a 20 MHz max 28-pin device. You get an LPC2101 > > for a $1.47 @ 10k that runs 3.5 times faster, uses a lot less current > > You're falling into the MHz = performance trap. An M3 is about twice > as fast as ARM7 MCUs running Thumb code (it runs Thumb-2 > code at the same efficiency of an ARM9 running ARM code). That > means you don't need to run at a high frequency to get the same > performance, and you use less power due to the lower frequency. Again, where are the benchmarks to show this? I'm not talking about stuff from simulations of an instruction set, I mean comparisons of an ARM7 chip and a Cortex chip. > A 20MHz Cortex-M3 is faster than just about all existing 8/16-bit CPUs, > including for example a 100MHz single cycle 8051. And the 50MHz > version outperforms current ARM7 MCUs by a huge margin. Are there benchmarks to show this? I would think LM would be posting these all over their web site and I don't see any.
From: Wilco Dijkstra on 8 Oct 2006 07:00
"rickman" <gnuarm(a)gmail.com> wrote in message news:1160256085.857608.36770(a)e3g2000cwe.googlegroups.com... > Wilco Dijkstra wrote: >> They can survive because they will make money once they get to volume >> production. Even with a tiny profit per chip selling millions of them >> means >> lots of money. > > But they have to get to the millions level pretty quickly to make that > work. Meanwhile they have completed two rounds of financing. Absolutely, but they seem to be pushing hard for it. It will take a while before volumes are high enough they become profitable, but that is normal for a startup. >> The initial devices are on an older process than the Atmel and Philips >> devices. Even so, since the M3 is a lot faster, it doesn't need to run at >> the same frequency as other MCUs so it may actually use less power. > > That is not at all clear. "May" is a word that has been used a lot > with these parts and I am still waiting for the evidence. At a 2:1 > power ratio for the max speed and with the other parts running at > higher clock rates, the LM parts need to do a lot of catching up. Sure, the current devices use a lot of power. We'll have to wait till production, they might fix the issues and hopefully move to a better process. However they are ahead on other aspects, I like the 50MHz zero-waitstate flash - nobody has flash that fast. >> You're falling into the MHz = performance trap. An M3 is about twice >> as fast as ARM7 MCUs running Thumb code (it runs Thumb-2 >> code at the same efficiency of an ARM9 running ARM code). That >> means you don't need to run at a high frequency to get the same >> performance, and you use less power due to the lower frequency. > > Again, where are the benchmarks to show this? I'm not talking about > stuff from simulations of an instruction set, I mean comparisons of an > ARM7 chip and a Cortex chip. Dhrystone numbers are available everywhere, including from Luminary: ARM7: ARM 0.9, Thumb 0.7 DMIPS/MHz ARM9: ARM 1.1, Thumb 0.9 DMIPS/MHz Cortex-M3 1.25 DMIPS/MHz These are comparisons of ARM7/ARM9/M3 hardware using the ARM compiler running from 32-bit zero-waitstate memory. As you know speeds will differ depending on the flash implementation. So if you want to know the speed of a particular MCU, you have to benchmark it yourself (manufacturers typically quote max speed of ARM code running from SRAM). I always recommend you benchmark your own application on your MCU using your toolset rather than rely on micro benchmarks done by others. As I've explained before, Dhrystone is not the best possible benchmark, it underestimates the difference between ARM and Thumb and overestimates micro architecture features like fast branches. >> A 20MHz Cortex-M3 is faster than just about all existing 8/16-bit CPUs, >> including for example a 100MHz single cycle 8051. And the 50MHz >> version outperforms current ARM7 MCUs by a huge margin. > > Are there benchmarks to show this? I would think LM would be posting > these all over their web site and I don't see any. It is a well-known fact that 32-bit RISC CPUs are many times faster than 8/16-bit (mostly CISC) chips - around 20 times in the case of 8051... Yes, 1 ARM instruction can do the work of around 20 8-bit instructions! There aren't many comparisons available because they are hard to do and mostly pointless as you know who is going to win. One glance at the cycle timings does it for me :-) Wilco |