That's the problem. Apple and later Google chose ARM because MIPS has no products for cell/smart phones. However, MIPS has been used on "higher power products" and MIPS has been used on 64 bit as well as multicore processors. ARM have issues with scalability and other performance related problems compared to MIPS. We will have to see if MIPS can get out of its comfort zone and see if they can convince the manufacturers that they can make better products compared to ARM.
MIPS has strong roots in TV and STB, but is fast losing the race even in these markets. Most of the LCD TVs produced so far use MIPS cores, and this includes SoCs made with 45nm or even 40nm. The biggest problem for MIPS is to reconcile with the fact that they are not the market leaders ..and cannot afford to model their business(or behavior) on the lines of Intel. So it is heartening to see MIPS start at the bottom - and try to get into commodity mobile / tablet business rather than try flirting with the big guys. (Perhaps they tried the latter and failed?).
Secondly, MIPS needs to get into good books with Google and prove that Android works just as well on MIPS. However they cannot run with the hares and hunt with the hounds at the same time.
Thirdly, they need an IP ecosystem. ARM has Mali in their backyard and the benefit of several independent IP providers dishing out everything from radios to codecs to video improvement blocks to optimized software. If you want to make a SoC, selecting ARM makes your life easy. Selecting MIPS, circa 2012, makes it challenging.
Depends on your application allot, MIPS might not have much going for it in the mobile space now days and did certainly lost out a bit on the mobile gpu front (although many of the current techs where developed for the mips market). They could still very well hold an advantage in specialized applications because the vendor has strong SDK's and platforms built around it which is certainly lacking in the ARM-space still.
That said they need Android and or Google TV and other smart platforms not to be pushed out of the space. MIPS is still powering plenty of CE and communications/enterprise systems. That vendors like Ingenic with Xburst is quite low end doesn't mean any mips implementation needs to be. It's certainly mass market products that drives both MIPS Technologies and ARM Ltd for that matter, which means not all SoC's will be top of the line. Although PowerPC is also strong in the automotive business as is MIPS it's hard to compete against say ARM7 microcontrollers. There are clearly different strengths.
Broadcom is still a major supplier and is moving MIPS-based STB, media, DTV/IPTV, home networking, etc applications along. Unless they switch you won't really get the same tools for ARM there. They might not drive baseband chips but neither is ARM Cortex-A8/9/15. ARM solutions simply doesn't have as strong IP, software and support like MIPS vendors in the STB field. Or other fields. It's not all about network streaming, you also need support for receiving OTA, Cable, Satellite, IPTV channels as well as encryption and so on. They can probably bring more if they succeeds with bringing the smarts of internet streaming, browsing and mobile applications to their platforms. More then ARM, ARM implementors in the field has to start from scratch pretty much. Mips can scale both quite low as well as ultra high-powered 64-bit multicore network processors. Latter is something ARM can't compete in at all. PowerPC can though. Applications processors for Tablets and mobile phones then it gets harder of course. But you have to still keep in mind in some fields companies like Broadcom have heavily invested in their own custom architectures that are MIPS-compatible. It's not likely to disappear or be replace for many years. Obviously no mayor vendor tries to create MIPS application processors for phones and tablets though, so those is pretty much reserved to those Chinese vendors. MIPS aren't actually targeting that market with their own soft and hard-cores.
It seems like not much uses them anymore. Sony did with the Playstation 1, 2, and Portable, but switched to PowerPC and ARM. Tivo switched from PowerPC in the Series 1 to MIPS in the 2, 3, and 4, but I don't really know what else uses them.
It seems like anymore ARM and x86 would be viable replacements for them in most anything? I hate to see another CPU architecture get squeezed out, but I don't know what they offer...
I think a lot of the home routers (e.g. Linksys E-series) use MIPS-based Broadcom SoC's.
The new TiVos also use Broadcom/MIPS, but I've heard many people mistakenly refer to them as using ARM cores. Seems kind of unfair that MIPS can't even get credit for that design win. People just assume anything embedded automatically equals ARM.
Elsewhere in Google, the Native Client people are trying to work around architecture dependence by publishing native code as LLVM bytecode (Portable Native Client, or pNaCl). One bold approach would be for Google to ask for native code in LLVM bytecode format for native code in Android apps. Then the Goog compiles and distributes as x86, ARM, MIPS, PowerPC...
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
8 Comments
Back to Article
iwod - Wednesday, January 18, 2012 - link
MIPS have an advantage for 64bit, I wonder if its power consumption can be as low as ARM.pugster - Wednesday, January 18, 2012 - link
That's the problem. Apple and later Google chose ARM because MIPS has no products for cell/smart phones. However, MIPS has been used on "higher power products" and MIPS has been used on 64 bit as well as multicore processors. ARM have issues with scalability and other performance related problems compared to MIPS. We will have to see if MIPS can get out of its comfort zone and see if they can convince the manufacturers that they can make better products compared to ARM.blue_cheese - Wednesday, January 18, 2012 - link
MIPS has strong roots in TV and STB, but is fast losing the race even in these markets. Most of the LCD TVs produced so far use MIPS cores, and this includes SoCs made with 45nm or even 40nm. The biggest problem for MIPS is to reconcile with the fact that they are not the market leaders ..and cannot afford to model their business(or behavior) on the lines of Intel. So it is heartening to see MIPS start at the bottom - and try to get into commodity mobile / tablet business rather than try flirting with the big guys. (Perhaps they tried the latter and failed?).Secondly, MIPS needs to get into good books with Google and prove that Android works just as well on MIPS. However they cannot run with the hares and hunt with the hounds at the same time.
Thirdly, they need an IP ecosystem. ARM has Mali in their backyard and the benefit of several independent IP providers dishing out everything from radios to codecs to video improvement blocks to optimized software. If you want to make a SoC, selecting ARM makes your life easy. Selecting MIPS, circa 2012, makes it challenging.
Penti - Wednesday, January 18, 2012 - link
Depends on your application allot, MIPS might not have much going for it in the mobile space now days and did certainly lost out a bit on the mobile gpu front (although many of the current techs where developed for the mips market). They could still very well hold an advantage in specialized applications because the vendor has strong SDK's and platforms built around it which is certainly lacking in the ARM-space still.That said they need Android and or Google TV and other smart platforms not to be pushed out of the space. MIPS is still powering plenty of CE and communications/enterprise systems. That vendors like Ingenic with Xburst is quite low end doesn't mean any mips implementation needs to be. It's certainly mass market products that drives both MIPS Technologies and ARM Ltd for that matter, which means not all SoC's will be top of the line. Although PowerPC is also strong in the automotive business as is MIPS it's hard to compete against say ARM7 microcontrollers. There are clearly different strengths.
Broadcom is still a major supplier and is moving MIPS-based STB, media, DTV/IPTV, home networking, etc applications along. Unless they switch you won't really get the same tools for ARM there. They might not drive baseband chips but neither is ARM Cortex-A8/9/15. ARM solutions simply doesn't have as strong IP, software and support like MIPS vendors in the STB field. Or other fields. It's not all about network streaming, you also need support for receiving OTA, Cable, Satellite, IPTV channels as well as encryption and so on. They can probably bring more if they succeeds with bringing the smarts of internet streaming, browsing and mobile applications to their platforms. More then ARM, ARM implementors in the field has to start from scratch pretty much. Mips can scale both quite low as well as ultra high-powered 64-bit multicore network processors. Latter is something ARM can't compete in at all. PowerPC can though. Applications processors for Tablets and mobile phones then it gets harder of course. But you have to still keep in mind in some fields companies like Broadcom have heavily invested in their own custom architectures that are MIPS-compatible. It's not likely to disappear or be replace for many years. Obviously no mayor vendor tries to create MIPS application processors for phones and tablets though, so those is pretty much reserved to those Chinese vendors. MIPS aren't actually targeting that market with their own soft and hard-cores.
Wolfpup - Wednesday, January 18, 2012 - link
It seems like not much uses them anymore. Sony did with the Playstation 1, 2, and Portable, but switched to PowerPC and ARM. Tivo switched from PowerPC in the Series 1 to MIPS in the 2, 3, and 4, but I don't really know what else uses them.It seems like anymore ARM and x86 would be viable replacements for them in most anything? I hate to see another CPU architecture get squeezed out, but I don't know what they offer...
ananduser - Wednesday, January 18, 2012 - link
It might be over time. They're good for low end tablets right now. MIPS might challenge ARM during the long run more than x86.Metaluna - Wednesday, January 18, 2012 - link
I think a lot of the home routers (e.g. Linksys E-series) use MIPS-based Broadcom SoC's.The new TiVos also use Broadcom/MIPS, but I've heard many people mistakenly refer to them as using ARM cores. Seems kind of unfair that MIPS can't even get credit for that design win. People just assume anything embedded automatically equals ARM.
twotwotwo - Wednesday, January 18, 2012 - link
Elsewhere in Google, the Native Client people are trying to work around architecture dependence by publishing native code as LLVM bytecode (Portable Native Client, or pNaCl). One bold approach would be for Google to ask for native code in LLVM bytecode format for native code in Android apps. Then the Goog compiles and distributes as x86, ARM, MIPS, PowerPC...