Sign in / up

back to article MIPS enters Android Honeycomb tablet race

Amid all the brouhaha about the low power–chip tussle between Intel and ARM, another processor architecture has been quietly advancing into the same tablet and smartphone battleground: MIPS Technologies, which has announced a partnership with Beijing's Ingenic Semiconductor to port Android 3.0, aka Honeycomb, to the Chinese …

COMMENTS

House rules Send corrections
This topic is closed for new posts.
  1. Joe Montana
    WTF?

    Low power servers..

    But when will we actually be able to buy hardware based on this? I want to buy a small low power mips or arm based server, but there doesnt seem to be much on the market.

    MIPS might make a better choice for servers than ARM, there is already a well specified tried and tested MIPS64 variant which should work well for servers.

    3 1
    1. Anonymous Coward
      Reply Icon
      Devil

      The article said 32

      It is not MIPS32 by the look of it - 32. That still makes a more than reasonable server out of it. You do not need more than 1G for email, file serving, minor web and minor apps.

      For example, my house server sits at 1.5G used memory (out of 2) while serving xterms for log-ins, doing all of my mail, serving diskless clients and providing NFS access to a few TB worth of storage. If I chuck out the content checking for junior's web access and the x-terms it will be content with 1G.

      0 0
  2. asdf

    will be amazed

    Anxious to see if they can put out a worldwide product that survives IP lawyer scrutiny as the Chinese aren't immune to reading IP laws a bit liberally. Wouldn't be surprised if they end up only able to sell the chip domestically and trying to export it through the knock off goods black market. The do have lots of resources including human capital so who knows. The more competitors the better for the customer.

    0 5
    1. MacroRodent
      Reply Icon

      Loongson

      The Chinese already export some stuff containing a MIPS64 clone processor called Loongson. I guess the CPU discussed in this article is different.

      Interesting this MIPS renaissance. It was the original pure RISC CPU, with an uncompromisingly Spartan instruction set.

      0 0
      1. Anonymous Coward
        Reply Icon
        Boffin

        Loongson

        Yes, very much a different beast.

        The Loongson 2E/2F are MIPS-III ISA, not a MIPS64 clone... Loongson 2E re-used some op-codes defined in MIPS64 for some custom functions, which makes it incompatible with MIPS64. Loongson 2F changed the op-codes for those functions so they no longer clash.

        The Loongson 3 series will possibly be MIPS64, but I'm unsure there... I think ICT were also working on a multicore chip, which would be very nice indeed.

        Been pretty happy with the Lemote Yeeloong (Loongson 2F) and the two older Fulong systems (Loongson 2E). Admittedly I'm running full blown GNU/Linux on mine, Android didn't exist when I bought the Yeeloong.

        0 0
    2. P. Lee
      Reply Icon
      Trollface

      re: Wouldn't be surprised if they end up only able to sell the chip domestically

      ... and thus the Chinese will be the only ones who spend their income with companies which make physical things, rather than the West with its ludicrous web of inter-corporate IP agreements which serve only to prevent smaller companies from challenging the incumbents.

      0 0
  3. Matt Bucknall

    Used in microcontrollers too...

    Microchip Technology's PIC32 microcontrollers are MIPS based apparently. Never used them though, I'm happy sticking with ARM.

    1 0
  4. Torben Mogensen

    Code density

    The original MIPS32 ISA has horrible code density, so I would choose ARM any day over that. Even the original 32-bit-only ARM ISA has a much better code density. MIPS, however, made a Thumb-like extension that has decent code density (though not quite as good as Thumb2), so if the planned processor implements this, it might be O.K.

    Still, MIPS and ARM were designed with 80's technology in mind, and it is not a clear fit to modern technology, though both have evolved somewhat with extensions and coprocessors (ARM more than MIPS). It would be refreshing if someone would make a new ISA that was a better fit for modern technology. It would take quite an effort to get a good code base, though. This could be partly remedied by supporting LLVM and JVM, though.

    1 0
  5. ThomH

    Slight problem versus the NDK, presumably?

    And I was under the impression that the only way to compile C code is via NDK, bypassing Android's virtual machine, giving a lazy option to EA, Epic, etc when porting their engines. I guess it'll be fine though — I'll bet that 99% of applications are purely Dalvik based.

    0 0
  6. Ken Hagan Gold badge
    Unhappy

    How depressing

    They describe a system that would have blown away top-end workstations of a decade ago, and suggest it might be usable in phones and e-readers.

    It seems that Gates' law trumps Moore's every time.

    3 0
    1. Anonymous Coward
      Reply Icon
      Unhappy

      I had similar thoughts..

      The poster above saying that; 1GB should be find for a small web server and such, or the article a few days back complaining that a phone "only had 512MB" to run in.

      At the other extreme I clearly remember using VAX/VMS at work with as little as 20MB Ram and tiny hard disks, that supported:

      - 10-20 concurrent software developers editing, compiling and testing code (via dumb terminals).

      - a decent file system, with native indexed files, version numbers and more.

      - a built in mail client.

      - the equivalent of MSN (called Phone)

      - printer server

      - Word processing (WordPerfect was available for VMS)

      - File serving via FTP and DECnet

      - and much, much more.

      I acquired a couple of VAX's a few years back and I gave one to a friend who has used it the run his personal web site for nearly 10 years.

      It has never been hacked (though the logs show the script kiddies were trying within 40 minutes), and the only down time has been due to him moving house and a few power failures.

      Not bad for a microVAX 3100-95 (~38VUPS or the speed of a 100Mhz Pentium) with the maximum 256MB and a GB or so of hard disk.

      1 0
      1. Hieronymous Kitsch
        Reply Icon

        VAX VMS

        Indeed. VMS allowed for a serious load - as long as you sorted the UAF parameters versus the Sysgen parameters. Tie them together, then you were rolling. But watch out for the MPW stuff; aka, swapper; Get you Hi an Lo limits wrong, you could seriously stuff it.

        Base it on 8, use a pen and pencil to craft it then quick stuff for quantum (2) if real, (30) if batch.

        I did 1500 users on 6220s cluster at the end of the day. Started on PDP RSX11, but who knows how these days.

        Got a pet Shoggoth now - who's giving a monkey's

        HK

        0 0
      2. Anonymous Coward
        Reply Icon
        Meh

        Welcome to the world of bloat

        Indeed, we've upsized our web server here... not because the PIII 550MHz was too slow, it was doing quite well... but because SCSI HDDs became too expensive compared to SATA drives... and the power bill was getting a bit much. Computing requirements aren't high, but we've got a lot of email and a photo gallery with a lot of content online.

        We have a small Intel Atom D510 server now with two "enterprise"-grade (that's how they were advertised; so far so good) 500GB HDDs in mirror RAID.

        The OS of course gets bigger each year (running Gentoo Linux)... thankfully not nearly as big as some of the commercial platforms out there... but not as Spartan as operating systems of years gone by which would otherwise serve us well and run with a mere fraction of the system overhead.

        0 0
This topic is closed for new posts.

Other stories you might like