Hi all, I've been curious about the relative merits of ARM and MIPS and just stumbled across a thread from March in which this was discussed a bit (http://groups.google.com/group/comp.arch.embedded/browse_frm/thread/732d062b040a5039/6261b3e0103cb19b?lnk=st&q=arm+processor&rnum=1#6261b3e0103cb19b). I'm a physics student and have gotten interested in RISC processors, since taking a course on Verilog/FPGA. I've been playing around with the open-source MIPS cores uCore and YACC (from opencores.org), and it's pretty fun to get a 100 MHz MIPS processor running on my Altera development board. I was under the impression that ARM's instruction set allows for denser code (even without Thumb), since it allows barrel shifting in arithmetic instructions and post-increment of index registers. It seems like a very clever design to me. On the other hand, MIPS has twice as many registers, 32 vs. 16. And all (I think?) licensed MIPS processors include an MMU, while many ARM cores do not have an MMU. Then, ARM is more popular it seems, but has very high licensing costs. And ARM's assembly language might be easier to work with, but which is better for compilers? I'm wondering what others think... what are the relative merits of the two architectures? Am I correct to think that MIPS might allow faster clock speeds and smaller dies, because of its simpler instruction set? Anyway, I'd love to know what others think. Myself, I have only coded assembly language for x86 and AVR (8-bit RISC microcontroller), so I don't have any low-level experience with 32-bit RISC. Thanks! Dan Lenski
MIPS vs ARM architectures
Started by ●June 29, 2007
Reply by ●June 29, 20072007-06-29
Daniel Lenski wrote:> Hi all, > I've been curious about the relative merits of ARM and MIPS and just > stumbled across a thread from March in which this was discussed a bit >(http://groups.google.com/group/comp.arch.embedded/browse_frm/thread/732d062b040a5039/6261b3e0103cb19b?lnk=st&q=arm+processor&rnum=1#6261b3e0103cb19b).> > I'm a physics student and have gotten interested in RISC processors, since > taking a course on Verilog/FPGA. I've been playing around with the > open-source MIPS cores uCore and YACC (from opencores.org), and it's > pretty fun to get a 100 MHz MIPS processor running on my Altera > development board. > > I was under the impression that ARM's instruction set allows for denser > code (even without Thumb), since it allows barrel shifting in arithmetic > instructions and post-increment of index registers. It seems like a very > clever design to me. On the other hand, MIPS has twice as many registers, > 32 vs. 16. And all (I think?) licensed MIPS processors include an MMU, > while many ARM cores do not have an MMU. Then, ARM is more popular it > seems, but has very high licensing costs. And ARM's assembly language > might be easier to work with, but which is better for compilers? > > I'm wondering what others think... what are the relative merits of the two > architectures? Am I correct to think that MIPS might allow faster clock > speeds and smaller dies, because of its simpler instruction set? Anyway, > I'd love to know what others think. Myself, I have only coded assembly > language for x86 and AVR (8-bit RISC microcontroller), so I don't have any > low-level experience with 32-bit RISC.Phil Koopman wrote an interesting book that compared processor architecture and provided some notions to examine the different styles. The book is now downloadable from http://www.ece.cmu.edu/~koopman/stack_computers/index.html -- ******************************************************************** Paul E. Bennett ....................<email://peb@amleth.demon.co.uk> Forth based HIDECS Consultancy .....<http://www.amleth.demon.co.uk/> Mob: +44 (0)7811-639972 Tel: +44 (0)1235-811095 Going Forth Safely ..... EBA. www.electric-boat-association.org.uk.. ********************************************************************
Reply by ●June 29, 20072007-06-29
Daniel Lenski wrote:> Hi all, > I've been curious about the relative merits of ARM and MIPS and just > stumbled across a thread from March in which this was discussed a bit > (http://groups.google.com/group/comp.arch.embedded/browse_frm/thread/732d062b040a5039/6261b3e0103cb19b?lnk=st&q=arm+processor&rnum=1#6261b3e0103cb19b). > > I'm a physics student and have gotten interested in RISC processors, since > taking a course on Verilog/FPGA. I've been playing around with the > open-source MIPS cores uCore and YACC (from opencores.org), and it's > pretty fun to get a 100 MHz MIPS processor running on my Altera > development board.Then you should also look a FPGA specific 32 bit cores too.> > I was under the impression that ARM's instruction set allows for denser > code (even without Thumb), since it allows barrel shifting in arithmetic > instructions and post-increment of index registers. It seems like a very > clever design to me. On the other hand, MIPS has twice as many registers, > 32 vs. 16.So, that means BOTH vendors can select benchmarks, that make their cores "look good". And all (I think?) licensed MIPS processors include an MMU,> while many ARM cores do not have an MMU. Then, ARM is more popular it > seems, but has very high licensing costs. And ARM's assembly language > might be easier to work with, but which is better for compilers? > > I'm wondering what others think... what are the relative merits of the two > architectures? Am I correct to think that MIPS might allow faster clock > speeds and smaller dies, because of its simpler instruction set? Anyway, > I'd love to know what others think. Myself, I have only coded assembly > language for x86 and AVR (8-bit RISC microcontroller), so I don't have any > low-level experience with 32-bit RISC.The Core vendors will pitch that a) the core is everything and b) Theirs is the best! (and remember, there are MANY cores from MIPS and ARM) The reality is the core matters less and less, and Memory bandwidth, Power and peripherals will determine how your system works. Then there is pure availability. Try and buy a MIPS Flash based Microcontroller ? There is more and more choice in the 32 bit MCU area: Soft CPUs, that are optimised for FPGAs, Freescale have a new V1 Coldfire, Atmel have a new AVR32 range (as well as ARM7 and ARM9), Zilog have Z16F, Infineon have the XC2200, and there are many vendors offering ARM uC..... Few these days will care about assembler level code, so choose the best uC for the task you have. Debug support, and tools, will start to matter in many design starts. -jg
Reply by ●June 30, 20072007-06-30
On Sat, 30 Jun 2007 12:15:01 +1200, Jim Granville wrote:> Then you should also look a FPGA specific 32 bit cores too.I should clarify that I don't actually have any product I want to produce at this point :-) I'm basically trying to learn more about processor architecture and how to prototype and tinker with these things.>> I was under the impression that ARM's instruction set allows for denser >> code (even without Thumb), since it allows barrel shifting in >> arithmetic instructions and post-increment of index registers. It >> seems like a very clever design to me. On the other hand, MIPS has >> twice as many registers, 32 vs. 16. > > So, that means BOTH vendors can select benchmarks, that make their cores > "look good".So, I'm guessing ARM would look good in a benchmark with arrays and data structures that require a lot of addressing modes? While MIPS would look better in a benchmark with limited memory bandwidth, making the registers more valuable?>> And all (I think?) licensed MIPS processors include an MMU, >> while many ARM cores do not have an MMU. Then, ARM is more popular it >> seems, but has very high licensing costs. And ARM's assembly language >> might be easier to work with, but which is better for compilers? >> >> I'm wondering what others think... what are the relative merits of the >> two architectures? Am I correct to think that MIPS might allow faster >> clock speeds and smaller dies, because of its simpler instruction set? >> Anyway, I'd love to know what others think. Myself, I have only coded >> assembly language for x86 and AVR (8-bit RISC microcontroller), so I >> don't have any low-level experience with 32-bit RISC. > > The Core vendors will pitch that a) the core is everything and b) Theirs > is the best! (and remember, there are MANY cores from MIPS and ARM) > > The reality is the core matters less and less, and Memory bandwidth, > Power and peripherals will determine how your system works. Then there > is pure availability. > Try and buy a MIPS Flash based Microcontroller ?Good point. I know that there's a lot more to a successful device than just the core. The most complex hardware project I've done used an 8-bit AVR microcontroller... and it was a pleasure to use, since it had on-chip DAC, PWM, and GPIO, and in-system-programmable flash memory.> There is more and more choice in the 32 bit MCU area: Soft CPUs, that > are optimised for FPGAs, Freescale have a new V1 Coldfire, Atmel have a > new AVR32 range (as well as ARM7 and ARM9), Zilog have Z16F, Infineon > have the XC2200, and there are many vendors offering ARM uC.....Coldfire looks not so exciting, since it's based on the CISC m68k, though I gather it's got a cleaned-up instruction set. The AVR32 looks really cool, like it is designed to be a do-everything architecture... good code density, DSP capabilities, SIMD capabilities, MMU, Java, wow. It looks like Atmel is aiming to really shake up the 32-bit embedded market.> Few these days will care about assembler level code, so choose the best > uC for the task you have. > > Debug support, and tools, will start to matter in many design starts.Yeah, the thing that's appealing about MIPS and ARM to me is that they are extremely well-supported by Linux and the GCC toolchain. Also, MIPS is even cooler cause there are free Verilog implementations to play around with! Dan
Reply by ●June 30, 20072007-06-30
On Fri, 29 Jun 2007 22:59:49 +0100, Paul E. Bennett wrote:> Phil Koopman wrote an interesting book that compared processor architecture > and provided some notions to examine the different styles. The book is now > downloadable from > > http://www.ece.cmu.edu/~koopman/stack_computers/index.html >Thanks! I skimmed a few chapters, some very interesting stuff. Although, my understanding is that stack-based computers have fallen out of favor due to the increasing disparity between CPU speed and memory bandwidth. Are there any other books or articles that compare register-based architectures to one another? Dan
Reply by ●June 30, 20072007-06-30
--------------8D36130813D245AF2C903A55 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Daniel Lenski wrote:> Hi all, > I've been curious about the relative merits of ARM and MIPS and just > stumbled across a thread from March in which this was discussed a bit > (http://groups.google.com/group/comp.arch.embedded/browse_frm/thread/732d062b040a5039/6261b3e0103cb19b?lnk=st&q=arm+processor&rnum=1#6261b3e0103cb19b). > > I'm a physics student and have gotten interested in RISC processors, since > taking a course on Verilog/FPGA. I've been playing around with the > open-source MIPS cores uCore and YACC (from opencores.org), and it's > pretty fun to get a 100 MHz MIPS processor running on my Altera > development board.For pure architecture curiosity look at the 16 bit Freescale xgate RISC This processor can achieve code density close to that of a CISC. Walter Banks -- Byte Craft Limited Tel. (519) 888-6911 http://www.bytecraft.com email walter@bytecraft.com --------------8D36130813D245AF2C903A55 Content-Type: text/html; charset=us-ascii Content-Transfer-Encoding: 7bit <!doctype html public "-//w3c//dtd html 4.0 transitional//en"> <html> <p>Daniel Lenski wrote: <blockquote TYPE=CITE>Hi all, <br>I've been curious about the relative merits of ARM and MIPS and just <br>stumbled across a thread from March in which this was discussed a bit <br>(<a href="http://groups.google.com/group/comp.arch.embedded/browse_frm/thread/732d062b040a5039/6261b3e0103cb19b?lnk=st&q=arm+processor&rnum=1#6261b3e0103cb19b">http://groups.google.com/group/comp.arch.embedded/browse_frm/thread/732d062b040a5039/6261b3e0103cb19b?lnk=st&q=arm+processor&rnum=1#6261b3e0103cb19b</a>). <p>I'm a physics student and have gotten interested in RISC processors, since <br>taking a course on Verilog/FPGA. I've been playing around with the <br>open-source MIPS cores uCore and YACC (from opencores.org), and it's <br>pretty fun to get a 100 MHz MIPS processor running on my Altera <br>development board.</blockquote> For pure architecture curiosity look at the 16 bit Freescale xgate RISC <br>This processor can achieve code density close to that of a CISC. <p>Walter Banks <br>-- <br>Byte Craft Limited <br>Tel. (519) 888-6911 <br> <A HREF="http://www.bytecraft.com">http://www.bytecraft.com</A> <br> email walter@bytecraft.com</html> --------------8D36130813D245AF2C903A55--
Reply by ●June 30, 20072007-06-30
Dan Lenski wrote:> On Fri, 29 Jun 2007 22:59:49 +0100, Paul E. Bennett wrote: >> Phil Koopman wrote an interesting book that compared processor >> architecture and provided some notions to examine the different styles. >> The book is now downloadable from >> >> http://www.ece.cmu.edu/~koopman/stack_computers/index.html >> > > Thanks! I skimmed a few chapters, some very interesting stuff. Although, > my understanding is that stack-based computers have fallen out of favor > due to the increasing disparity between CPU speed and memory bandwidth.Although eschewed by the mainstream companies, there is still a lot to be said in favour of stack machine architectures. There is, of course, renewed interest in stack machines from the Java generation. Stack machines are simple enough to be able to put several in the current families of FPGA and still have enough logic spare to accomplish a multi-processor application in one chip.> Are there any other books or articles that compare register-based > architectures to one another?These other links may provide some useful material. <http://www.springerlink.com/content/u0uw5gu5q4p3bl9n/> <http://ieeexplore.ieee.org/Xplore/login.jsp?url=/iel5/7018/18934/00874655.pdf?arnumber=874655> <http://ce.et.tudelft.nl/publicationfiles/933_509_shahbahrami.pdf> I would persevere with Phils book a bit further because the bit on how he arranged his comparisons might be useful when you come to do your own comparison work. Someone following on from Phil's work with the up-to date available processors might make for another good book on the topic, especially if you included some of the multi-processor chips as well. -- ******************************************************************** Paul E. Bennett ....................<email://peb@amleth.demon.co.uk> Forth based HIDECS Consultancy .....<http://www.amleth.demon.co.uk/> Mob: +44 (0)7811-639972 Tel: +44 (0)1235-811095 Going Forth Safely ..... EBA. www.electric-boat-association.org.uk.. ********************************************************************
Reply by ●June 30, 20072007-06-30
On Sat, 30 Jun 2007 04:25:18 GMT, Dan Lenski <dlenski@gmail.com> wrote:>Thanks! I skimmed a few chapters, some very interesting stuff. Although, >my understanding is that stack-based computers have fallen out of favor >due to the increasing disparity between CPU speed and memory bandwidth.For the latest stuff, but radically different, see: www.intellasys.net>Are there any other books or articles that compare register-based >architectures to one another?My bookshelf had a clear-out a few months ago, and I've forgotten the titles and authors! Otherwise, the most fascinating is "Interpretation and Instruction Path Coprocessing", Debaere and Campenhout, MIT Press, 1989 (probably out of print), ISBN 0-262-04107-3. This includes an excellent analysis of CPU/memory bandwidth issues. The only downside of this book is that code generation technology for stack machines has moved on a great deal since then, but the parts about the hardware are still valid. Anyone fighting secondary instruction set patents, e.g. Thumb-1 and MIPS16, should also read this book! :-} Stephen -- Stephen Pelc, stephenXXX@mpeforth.com MicroProcessor Engineering Ltd - More Real, Less Time 133 Hill Lane, Southampton SO15 5AF, England tel: +44 (0)23 8063 1441, fax: +44 (0)23 8033 9691 web: http://www.mpeforth.com - free VFX Forth downloads
Reply by ●July 2, 20072007-07-02
Thanks for all the suggestions! For anyone else interested, I've found an excellent 2005 book comparing RISC architectures: Dandamudi, S. P. "Guide to RISC Processors." Springer 2005. 387 pages. It has an overview of RISC, then separate chapters on MIPS, SPARC, PowerPC, Itanium, and ARM, and then a lengthy section on MIPS programming and architecture. Dan
