Post by thread:[PIC] PIC24 On the Radar Screen of Cortex M3 MCU V

Yesterday an ST STM32 (Coretx M3) guy from France visited us promoting STM32. I can clearly see that it beats the TI TMS470 we use right now in almost very aspect (lower current, single supply, 5V I/O tolerant, fast bit toggling, lower cost). Looks nice and it supports 105 degree C operation just like TMS470. The only thing unique to TMS470 is the HET High End Timer, kind of a coprocessor for waveform generator. Interestingly they also list PIC24 along with Renesas M16C and H8S for code density and speed comparison. Of course in their benchmark, STM32 beat all three. Interestingly PIC24 beats M16C and H8S by a big margin and lose not too much to STM32. The PIC24 is cheap, good performance (up to 40MIPS, lose out to STM32 -->up to 72MHz or 80MIPS??). Harold is asking for more performance from PIC24 or its successor. It is very interesting to see what is the response from Microchip. Rumor is that they will develop PIC32, a 32bit PIC. http://forum.microchip.com/tm.aspx?m=291734 Regards, Xiaofan

On 10/31/07, Jan-Erik Soderholm <spam_OUTjan-erik.soderholmTakeThisOuTtelia.com> wrote: > > Rumor is that they will develop PIC32, a 32bit PIC. > > http://forum.microchip.com/tm.aspx?m=291734 > > > > Where does it say that ? No that is the post from Harold asking for more speed. PIC32 is only a rumor I heard some time ago. Xiaofan

OK, I thought that the text and the link was related... :-) Jan-Erik. Xiaofan Chen wrote: > On 10/31/07, Jan-Erik Soderholm <.....jan-erik.soderholmKILLspam@spam@telia.com> wrote: >>> Rumor is that they will develop PIC32, a 32bit PIC. >>> http://forum.microchip.com/tm.aspx?m=291734 >>> >> Where does it say that ? > No that is the post from Harold asking for more speed. PIC32 is only > a rumor I heard some time ago. > > Xiaofan

> Wondering How big a market for 30-60MIPS? > For end user, the higher the MIPS, the better for most of the case. That's where the ARM uC's are strong now. I doubt Microchip could beat the NXP chips in this arena on price and features. There might be a niche left for DIP chips, but I doubt that is large enough. Wouter van Ooijen -- ------------------------------------------- Van Ooijen Technische Informatica: http://www.voti.nl consultancy, development, PICmicro products docent Hogeschool van Utrecht: http://www.voti.nl/hvu

On 10/31/07, wouter van ooijen <wouterKILLspamvoti.nl> wrote: > > Wondering How big a market for 30-60MIPS? > > For end user, the higher the MIPS, the better for most of the case. > > That's where the ARM uC's are strong now. I doubt Microchip could beat > the NXP chips in this arena on price and features. There might be a > niche left for DIP chips, but I doubt that is large enough. > NXP is not really cheap other than the a few low end device. And PIC24H is actually pretty cheap. http://www.gnuarm.com/ArmDevices_frame.html http://www.microchip.com/ParamChartSearch/chart.aspx?branchID=8186&mid=14&lang=en&pageId=75 The Cortex M3 based MCU is supposed to be cheaper than those Arm7 based MCU according to ST. They say their price will beat NXP to death. Let's wait and see. Xiaofan

> > NXP is not really cheap other than the a few low end device. > And PIC24H is actually pretty cheap. > http://www.gnuarm.com/ArmDevices_frame.html > Which chips would you compare? If I take the LPC2103 it seems to match 24HJ32GP204 / 24HJ64GP206. For prices I prefer to compare one supplier, digikey @ 100 is 2.92 for the LPC, 4.05 for the GP206 (but maybe that does not do full justice to microchip-direct). The PICs are 40 MIPS, the LPC's (IIRC) 72 (maybe only 60) MIPS. And the LPC is 32 bits, these PICs are 16 bits, right? Another popular ARM is the LPC2148: 512k flash, 32k ram, USB interface, 64 LQFP, digikey $9.48 @ 100. Which 24HJ would be comparable? The ARM is strongly supported by free (GCC) and commercial (based on either GCC or propriatary compiler) toolchains. Asm, C, C++, pascal, java, download whatever you want. Upward migration is possible to uP mode chips that run Linux. I think the PIC24 support is a bit pale in comparison... One strong point for the 24's might be Microchips reputation for not dropping a chip. It is too soon to say how NXP and the other ARM suppliers will so on this aspect. Wouter van Ooijen -- ------------------------------------------- Van Ooijen Technische Informatica: http://www.voti.nl consultancy, development, PICmicro products docent Hogeschool van Utrecht: http://www.voti.nl/hvu

On 10/31/07, Xiaofan Chen <.....xiaofancKILLspam.....gmail.com> wrote: > Interestingly they also list PIC24 along with Renesas M16C and H8S > for code density and speed comparison. Of course in their benchmark, > STM32 beat all three. Interestingly PIC24 beats M16C and H8S by > a big margin and lose not too much to STM32. > > The PIC24 is cheap, good performance (up to 40MIPS, lose out > to STM32 -->up to 72MHz or 80MIPS??). Harold is asking for > more performance from PIC24 or its successor. It is very > interesting to see what is the response from Microchip. Rumor > is that they will develop PIC32, a 32bit PIC. If they want to develop a 32bit MUC, I think they have sevearl choices. 1) Using an existing 32bit MCU core like those from ARM, espeiclaly the new Cortex M3 where not many vendors are supporting it (Luminary, ST now. TI is said to develop some new ones). Or they can use MIPS since it just announced the new Core targeted toward 32bit MCU http://www.eetimes.com/news/latest/showArticle.jhtml;jsessionid=E0BXFAWUYKDL4QSNDLSCKHA?articleID=202800129 2). Use the existing PIC24 and then extend it to PIC32. 3) Combined 1) and 2) just like Atmel with ARM and AVR32. Take note Atmel is not backing Cortex M3 but choose to use Arm7/Arm9/AVR32. I will tend to guess Microchip will go route 1. Xiaofan

On Oct 31, 2007, at 8:06 AM, wouter van ooijen wrote: > Another popular ARM is the LPC2148: 512k flash, 32k ram, USB > interface, > 64 LQFP, digikey $9.48 @ 100. Which 24HJ would be comparable? I find the prospect of a single architecture spanning the "1 dollar to 1 GHz" (Luminary Micro phrase, I think) to be pretty attractive. Assuming the M3 and ARM9 or whatever are really similar enough to group under the "same architecture" label. Assorted company's 16-bit offerings (including Microchip) look pretty unattractive compared to the low-end ARM chips. I couldn't muster much interest at all in the 16-bit PIC contest that was held recently, for instance... BillW

On 11/1/07, William Chops Westfield <EraseMEwestfwspam_OUTTakeThisOuTmac.com> wrote: > I find the prospect of a single architecture spanning the "1 dollar to > 1 GHz" (Luminary Micro phrase, I think) to be pretty attractive. > Assuming the M3 and ARM9 or whatever are really similar enough > to group under the "same architecture" label. Assorted company's > 16-bit offerings (including Microchip) look pretty unattractive compared > to the low-end ARM chips. > > I couldn't muster much interest at all in the 16-bit PIC contest that > was held recently, for instance... True. I did not use my Explorer 16 Demo board even though I attended the seminar and bought the board. That is why I was supprised to see the relative low price of PIC24. That being said, 8051 is said to be the "multi-vendor" platform and have a lot of vendors and 3rd party tools. However this does not prevent PIC/AVR/MSP430 "prorietory" architetures to be very successful in the market. So believe the same thing will happen. ARM7/9/Cortex M3 core are now the popular "multi-vendor" platform and enjoyed quite some vendors and 3rd party tools. Still I think Renesas/FreeScale/Microchip/Atmel will still do well. Atmel is an interesting case that it used ARM7/ARM9 but choose not use Coretx M3 and uses its own AVR32 UC3 to compete with Cortex M3 core. Freescale is now pushing seamless migration between its 8/32bit with RS08 core and Coldfire V1 core. Renesas is now pushing 16/32bit easy migration path with its various cores. It will be really interesting to see if Microchip will use ARM core for 32bit PIC. My bet is that it will not. Xiaofan

On 10/31/07, wouter van ooijen <wouterspam_OUTvoti.nl> wrote: > > Another popular ARM is the LPC2148: 512k flash, 32k ram, USB interface, > 64 LQFP, digikey $9.48 @ 100. Which 24HJ would be comparable? Microchip is pretty weak in USB. And LPC2148 is way too expensive if compared with the STM32 USB parts. > The ARM is strongly supported by free (GCC) and commercial (based on > either GCC or propriatary compiler) toolchains. Asm, C, C++, pascal, > java, download whatever you want. Upward migration is possible to uP > mode chips that run Linux. I think the PIC24 support is a bit pale in > comparison... Ture. PIC24 is not in the same league. But I think it can compete well with LPC210x or higher end 8051. The ARM9/XScale core are pretty the choice for Linux or Windows CE compatibility. Maybe Intel and AMD will not sit there and will come out some lower power and integrated x86 to get a piece of the big pie. Xiaofan

> > Another popular ARM is the LPC2148: 512k flash, 32k ram, USB > > interface, > > 64 LQFP, digikey $9.48 @ 100. Which 24HJ would be comparable? > > Assorted company's 16-bit offerings > (includin Microchip) look pretty unattractive compared to the low-end > ARM chips. I tend to agree, but I am open to arguments to the contrary. One notable difference: Mirochip has a hughe range of chips, compared to that even the combined ARM offererings look pale. But for the beginner / small-scale professional that is not a real advantage, maybe even a disadvantage. For the professional / large-scale user it is probably an advantage. Wouter van Ooijen -- ------------------------------------------- Van Ooijen Technische Informatica: http://www.voti.nl consultancy, development, PICmicro products docent Hogeschool van Utrecht: http://www.voti.nl/hvu

> I find the prospect of a single architecture spanning the > "1 dollar to > 1 GHz" (Luminary Micro phrase, I think) to be pretty > attractive. > Assuming > the M3 and ARM9 or whatever are really similar enough to > group under the > "same architecture" label. Assorted company's 16-bit > offerings > (including > Microchip) look pretty unattractive compared to the > low-end ARM chips. > I couldn't muster much interest at all in the 16-bit PIC > contest that > was held recently, for instance... The area that ARM are going to have most problem invading is Microchip's traditional homeland - the ultra low power applications where low clock speed is acceptable as a tradeoff for energy use. ARM is inherently more 'lossy'. A friend who carries a bright burning ARM torch at all times (only figuratively most of the time (you know who you are, Ken) says that uChip will adopt ARM or die. Failing that they may have to head for the hills and remain the bastion of the cable TV and smart card and ... applications. Russell

On 11/1/07, Xiaofan Chen <@spam@xiaofancKILLspamgmail.com> wrote: > On 10/31/07, wouter van ooijen <KILLspamwouterKILLspamvoti.nl> wrote: > > > > Another popular ARM is the LPC2148: 512k flash, 32k ram, USB interface, > > 64 LQFP, digikey $9.48 @ 100. Which 24HJ would be comparable? > > Microchip is pretty weak in USB. And LPC2148 is way too expensive > if compared with the STM32 USB parts. By the way, I bought the Olimex LPC-P2148 1.5 years ago and I still do not have much time to play with it. It is a nice chip. The LPC2888 is even better (high speed USB). Xiaofan

> Microchip is pretty weak in USB. And LPC2148 is way too > expensive if compared with the STM32 USB parts. One more source to check, thanks. Wouter van Ooijen -- ------------------------------------------- Van Ooijen Technische Informatica: http://www.voti.nl consultancy, development, PICmicro products docent Hogeschool van Utrecht: http://www.voti.nl/hvu

> By the way, I bought the Olimex LPC-P2148 1.5 years ago and I > still do not have much time to play with it. It is a nice > chip. The LPC2888 is even better (high speed USB). IIRC the were lots of issues with this chip (late delivery, bugs). Almost like Mirochip ;) Wouter van Ooijen -- ------------------------------------------- Van Ooijen Technische Informatica: http://www.voti.nl consultancy, development, PICmicro products docent Hogeschool van Utrecht: http://www.voti.nl/hvu

> The area that ARM are going to have most problem invading is > Microchip's traditional homeland - the ultra low power > applications where low clock speed is acceptable as a > tradeoff for energy use. IMO the PICs still rule in the word of the very small chips. My worry is that uChip will devote too much of its attention to the higher end market. Wouter van Ooijen -- ------------------------------------------- Van Ooijen Technische Informatica: http://www.voti.nl consultancy, development, PICmicro products docent Hogeschool van Utrecht: http://www.voti.nl/hvu

On Nov 1, 2007, at 2:44 AM, wouter van ooijen wrote: >> The LPC2888 is even better (high speed USB). > > IIRC the were lots of issues with this chip I was a little disturbed to recently receive advertising from NXP on their small ARM chips like the 2103, all pointing to documentation that is still tagged "preliminary", though it was released in Jan 2006. BillW

> I was a little disturbed to recently receive advertising from > NXP on their small ARM chips like the 2103, all pointing to > documentation that is still tagged "preliminary", though it > was released in Jan 2006. Welcom to the ugly world. A lot of datasheets from Microchip, Atmel and NXP are marked "preliminary" forever. I mention only those 3 because I have not checked any other ;) Wouter van Ooijen -- ------------------------------------------- Van Ooijen Technische Informatica: http://www.voti.nl consultancy, development, PICmicro products docent Hogeschool van Utrecht: http://www.voti.nl/hvu

Another rumor...did I hear it here or? is they may license the ARM core as well for the 32 bit Jan-Erik Soderholm <RemoveMEjan-erik.soderholmTakeThisOuTtelia.com> wrote: Xiaofan Chen wrote: > Rumor > is that they will develop PIC32, a 32bit PIC. > http://forum.microchip.com/tm.aspx?m=291734 > > Regards, > Xiaofan Where does it say that ? Jan-Erik.

On 11/1/07, wouter van ooijen <spamBeGonewouterspamBeGonevoti.nl> wrote: > > Microchip is pretty weak in USB. And LPC2148 is way too > > expensive if compared with the STM32 USB parts. > > One more source to check, thanks. It is interesting that NXP does not go the route of Cortex M3 which apparently have better performance for MCU application than ARM7TDMI which is not designed with MCU application in mind. Xiaofan