3 ms·

Wave Computing Closes Its MIPS Open Initiative with Immediate Effect

32 points by johlo 7y ago

skissane 7y ago
> Thus far, the company has not given a reason for the closure of the MIPS Open Initiative - nor explained why it has opted to do so with immediate effect, rather than giving those who had begun building around the architecture fair warning of its plans It is disappointing when a company won't publicly explain the reasons behind its decisions. It doesn't impress.
tyingq 7y ago
Lasted less than a year. https://news.ycombinator.com/item?id=18701145 https://news.ycombinator.com/item?id=18701145
happycube 7y ago
We're still relevant, da... no, we're not.
- tyingq 7y ago
  I believe lots of home routers still use them.
  - tenebrisalietum 7y ago
    The most recent router I bought was a dual core ARM. I suspect anything with a Broadcom chipset is going to be ARM these days.
  - zmix 7y ago
    My Ubiquiti EdgeRouter-Lite does. The MIPS CPUs can have hardware offloading for special network stuff, or so I have been told.
    - klagermkii 7y ago
      Yeah, MIPS was nice with allowing custom instructions to be added to their cores, but with the relatively recent Arm licensing change for Armv8-M that allows something similar I wonder if more will just switch to Arm. https://hackaday.com/2019/10/23/arm-allows-custom-instructions/ https://hackaday.com/2019/10/23/arm-allows-custom-instructio...
      - zmix 7y ago
        Interesting. But, oh, the comments to that post...!
baybal2 7y ago
They were a decade too late with their initiative. And Mips-Imagination as a commercial entity was braindead. Look for stories of them responding to sales calls after 1+ month. Execution was below the sea level. I can say the same of many other "Big Semi" that went nowhere in the last 5 years. They all had underperforming C-suites stuffed with overfed big name executives.
- CalChris 7y ago
  > They were a decade too late with their initiative. I don't think MIPS, open source or not, is competitive in a RISC-V world, and I'm not even much of a RISC-V fan (I'm more of an ARMv8 fan). Yeah, if MIPS had done this a decade ago and forestalled the rise of a RISC-V, then they may have had a chance.
- exikyut 7y ago
  My horribly underinformed understanding is that MIPS was ultimately a, uh, baseline architecture design in terms of performance to power consumption, particularly with respect to ARM's exponential progress post-iPhone. The going-nowhere bureaucracy-heavy structure resonates with the impression I got from reading this - I can't help but wonder if: - they brought in outside consultants to "strategize" how to respond to the OpenPOWER / RISC-V "threat" (postulation: RISC-V put the frozen deer in front of the headlights, while OpenPOWER was the foghorn that got them scrambling at the last moment - in any direction) - said consultant wanted to ensure they'd get paid (translation: knew they were talking to a board that expected a yes-team), so simply coughed up "modern/relevant industry standard" advice they (probably correctly) assumed was expected of them: go open - the entire industry went "all riiiiight, free MIPS!", and wasted no time commercially optimizing around royalty-free licencing - MIPS' "overfed" (as you astutely put it) board suddenly saw all their holidays and etcetera flying away - either this happened a while ago and the consultant said to hang on and it backfired, or this just happened recently - Exceptionally loud meetings (probably involving death threats) were had, possibly angry and/or irrational "NO, WE ARE STOPPING THIS NOW" decisions were made, and... ...welp. Given MIPS' execution being "below sea level" (again, as you put it), I get the impression the management had no resonance with the technical architecture of what it was they were managing. So, they couldn't move the innovation forward, only industrialize the process of perpetuating what already existed. Hence, going open without considering how to pull that off in a way that is both financially strategic and brings the best out of the architecture. And because they didn't understand the strengths of their architecture they had no confidence/competence to execute properly in the first place, anticipate the disaster in this approach, or handle it correctly. </naive_anecdotal_theory>
justinjlynn 7y ago
In addition to the fully open ISAs listed in the article (OpenSPARC and RISC-V) the POWER ISA (of PowerPC origin) is also fully open. This is a somewhat recent development.
duskwuff 7y ago
Good lord. I couldn't come up with a better way of permanently destroying trust in a company if I tried.

32 points by johlo 7y ago

> Thus far, the company has not given a reason for the closure of the MIPS Open Initiative - nor explained why it has opted to do so with immediate effect, rather than giving those who had begun building around the architecture fair warning of its plans It is disappointing when a company won't publicly explain the reasons behind its decisions. It doesn't impress.

tyingq 7y ago

Lasted less than a year. https://news.ycombinator.com/item?id=18701145 https://news.ycombinator.com/item?id=18701145

happycube 7y ago

We're still relevant, da... no, we're not.

tyingq 7y ago
I believe lots of home routers still use them.
- tenebrisalietum 7y ago
  The most recent router I bought was a dual core ARM. I suspect anything with a Broadcom chipset is going to be ARM these days.
- zmix 7y ago
  My Ubiquiti EdgeRouter-Lite does. The MIPS CPUs can have hardware offloading for special network stuff, or so I have been told.
  - klagermkii 7y ago
    Yeah, MIPS was nice with allowing custom instructions to be added to their cores, but with the relatively recent Arm licensing change for Armv8-M that allows something similar I wonder if more will just switch to Arm. https://hackaday.com/2019/10/23/arm-allows-custom-instructions/ https://hackaday.com/2019/10/23/arm-allows-custom-instructio...
    - zmix 7y ago
      Interesting. But, oh, the comments to that post...!

baybal2 7y ago

They were a decade too late with their initiative. And Mips-Imagination as a commercial entity was braindead. Look for stories of them responding to sales calls after 1+ month. Execution was below the sea level. I can say the same of many other "Big Semi" that went nowhere in the last 5 years. They all had underperforming C-suites stuffed with overfed big name executives.

CalChris 7y ago
> They were a decade too late with their initiative. I don't think MIPS, open source or not, is competitive in a RISC-V world, and I'm not even much of a RISC-V fan (I'm more of an ARMv8 fan). Yeah, if MIPS had done this a decade ago and forestalled the rise of a RISC-V, then they may have had a chance.
exikyut 7y ago
My horribly underinformed understanding is that MIPS was ultimately a, uh, baseline architecture design in terms of performance to power consumption, particularly with respect to ARM's exponential progress post-iPhone. The going-nowhere bureaucracy-heavy structure resonates with the impression I got from reading this - I can't help but wonder if: - they brought in outside consultants to "strategize" how to respond to the OpenPOWER / RISC-V "threat" (postulation: RISC-V put the frozen deer in front of the headlights, while OpenPOWER was the foghorn that got them scrambling at the last moment - in any direction) - said consultant wanted to ensure they'd get paid (translation: knew they were talking to a board that expected a yes-team), so simply coughed up "modern/relevant industry standard" advice they (probably correctly) assumed was expected of them: go open - the entire industry went "all riiiiight, free MIPS!", and wasted no time commercially optimizing around royalty-free licencing - MIPS' "overfed" (as you astutely put it) board suddenly saw all their holidays and etcetera flying away - either this happened a while ago and the consultant said to hang on and it backfired, or this just happened recently - Exceptionally loud meetings (probably involving death threats) were had, possibly angry and/or irrational "NO, WE ARE STOPPING THIS NOW" decisions were made, and... ...welp. Given MIPS' execution being "below sea level" (again, as you put it), I get the impression the management had no resonance with the technical architecture of what it was they were managing. So, they couldn't move the innovation forward, only industrialize the process of perpetuating what already existed. Hence, going open without considering how to pull that off in a way that is both financially strategic and brings the best out of the architecture. And because they didn't understand the strengths of their architecture they had no confidence/competence to execute properly in the first place, anticipate the disaster in this approach, or handle it correctly. </naive_anecdotal_theory>

justinjlynn 7y ago

In addition to the fully open ISAs listed in the article (OpenSPARC and RISC-V) the POWER ISA (of PowerPC origin) is also fully open. This is a somewhat recent development.

duskwuff 7y ago

Good lord. I couldn't come up with a better way of permanently destroying trust in a company if I tried.