• ARM Looking To Build Its Own Hardware?

    From Lawrence =?iso-8859-13?q?D=FFOlivei@21:1/5 to All on Tue Aug 19 22:22:00 2025
    ARM’s business model, up to now, is to license chip designs and
    patents to all comers who want to make their own chips. This has made
    the ARM architecture the most successful in human history.

    But the company has been making some moves that are unnerving its
    licensees, lately. And here’s another one: it now wants to make its
    own hardware. Won’t that put it in direct competition with those
    licensees? Can you say “conflict of interest”?

    <https://www.tomshardware.com/tech-industry/artificial-intelligence/arm-hires-amazons-ai-chip-developer-to-help-create-its-own-processors-rami-sinno-returns-to-the-company-boasts-trainium-and-inferentia-on-resume>

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Richard Kettlewell@21:1/5 to ldo@nz.invalid on Wed Aug 20 08:47:39 2025
    Lawrence D’Oliveiro <ldo@nz.invalid> writes:
    ARM’s business model, up to now, is to license chip designs and
    patents to all comers who want to make their own chips. This has made
    the ARM architecture the most successful in human history.

    But the company has been making some moves that are unnerving its
    licensees, lately. And here’s another one: it now wants to make its
    own hardware. Won’t that put it in direct competition with those
    licensees? Can you say “conflict of interest”?

    <https://www.tomshardware.com/tech-industry/artificial-intelligence/arm-hires-amazons-ai-chip-developer-to-help-create-its-own-processors-rami-sinno-returns-to-the-company-boasts-trainium-and-inferentia-on-resume>

    Something similar was reported in February.

    It wouldn’t be outside historical industry norms. Intel made their own
    8086s but also licensed them to around 10 other manufacturers; the
    Intel/AMD thing today is similar although there may be licensing going
    on in both directions there.

    From a customer point of view: it’s not _that_ hard to change CPU architecture. We’ve migrated between different CPU architectures a
    couple of times in our product history, we could do it again if it was
    really necessary (though don’t know of any reason to think an Arm entry
    into manufacturing would force us in particular to do so).

    --
    https://www.greenend.org.uk/rjk/

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Lawrence =?iso-8859-13?q?D=FFOlivei@21:1/5 to Richard Kettlewell on Wed Aug 20 22:05:32 2025
    On Wed, 20 Aug 2025 08:47:39 +0100, Richard Kettlewell wrote:

    It wouldn’t be outside historical industry norms. Intel made their own 8086s but also licensed them to around 10 other manufacturers ...

    But Intel was a chip manufacturer from the get-go. ARM Ltd was set up
    precisely as a company that made designs (and got patents on them) for licensing to others, not as one that made its own chips. That way other companies paying it money for those licences could feel confident they
    were all competing fairly, on a level field with an honest broker.

    But if ARM makes its own chips, now it is competing with its own
    customers. Can you say “conflict of interest”?

    From a customer point of view: it’s not _that_ hard to change CPU architecture.

    Are you talking embedded? Because in general-purpose computing, the
    proprietary vendors anyway (Microsoft being the obvious example) have a
    great deal of trouble supporting more than one platform.

    The Open Source world certainly can manage it quite successfully. That’s
    how it could be the first to embrace RISC-V.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Richard Kettlewell@21:1/5 to ldo@nz.invalid on Thu Aug 21 08:41:00 2025
    Lawrence D’Oliveiro <ldo@nz.invalid> writes:
    Richard Kettlewell wrote:
    From a customer point of view: it’s not _that_ hard to change CPU
    architecture.

    Are you talking embedded?

    Yes.

    Because in general-purpose computing, the proprietary vendors anyway (Microsoft being the obvious example) have a great deal of trouble
    supporting more than one platform.

    Apple have changed architecture three times now, HP and Sun at least
    twice each. Microsoft support two architectures at present and have
    supported more in the past.

    --
    https://www.greenend.org.uk/rjk/

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Lawrence =?iso-8859-13?q?D=FFOlivei@21:1/5 to Richard Kettlewell on Thu Aug 21 21:41:41 2025
    On Thu, 21 Aug 2025 08:41:00 +0100, Richard Kettlewell wrote:

    Lawrence D’Oliveiro <ldo@nz.invalid> writes:

    Because in general-purpose computing, the proprietary vendors
    anyway (Microsoft being the obvious example) have a great deal of
    trouble supporting more than one platform.

    Apple have changed architecture three times now ...

    But never more than main one at a time -- not for longer than
    absolutely needed to make the transition.

    ... HP and Sun at least twice each.

    All the old HP minis and RISC architectures have gone. Itanium has
    gone. HP only does x86 now. And Sun has gone.

    Microsoft support two architectures at present and have supported
    more in the past.

    Windows-on-ARM is struggling. And Windows NT on non-x86 architectures
    never lasted long. Fun fact: even on x86, which is now 64-bit, Windows
    is still a 32-bit architecture at heart.

    You see what I mean, don’t you?

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Richard Kettlewell@21:1/5 to ldo@nz.invalid on Fri Aug 22 09:10:07 2025
    Lawrence D’Oliveiro <ldo@nz.invalid> writes:
    Richard Kettlewell wrote:
    Lawrence D’Oliveiro <ldo@nz.invalid> writes:

    Because in general-purpose computing, the proprietary vendors
    anyway (Microsoft being the obvious example) have a great deal of
    trouble supporting more than one platform.

    Apple have changed architecture three times now ...

    But never more than main one at a time -- not for longer than
    absolutely needed to make the transition.

    So what? The claim I made was about an architecture migration, not about supporting multiple architectures concurrently.

    ... HP and Sun at least twice each.

    All the old HP minis and RISC architectures have gone. Itanium has
    gone. HP only does x86 now. And Sun has gone.

    Again, so what?

    Microsoft support two architectures at present and have supported
    more in the past.

    Windows-on-ARM is struggling. And Windows NT on non-x86 architectures
    never lasted long. Fun fact: even on x86, which is now 64-bit, Windows
    is still a 32-bit architecture at heart.

    You see what I mean, don’t you?

    You’re obviously not addressing the claim I actually made, so I’m going back to assuming you’re not arguing in good faith.

    --
    https://www.greenend.org.uk/rjk/

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Theo@21:1/5 to ldo@nz.invalid on Fri Aug 22 19:12:17 2025
    Lawrence D’Oliveiro <ldo@nz.invalid> wrote:
    On Wed, 20 Aug 2025 08:47:39 +0100, Richard Kettlewell wrote:

    It wouldn’t be outside historical industry norms. Intel made their own 8086s but also licensed them to around 10 other manufacturers ...

    But Intel was a chip manufacturer from the get-go. ARM Ltd was set up precisely as a company that made designs (and got patents on them) for licensing to others, not as one that made its own chips. That way other companies paying it money for those licences could feel confident they
    were all competing fairly, on a level field with an honest broker.

    But if ARM makes its own chips, now it is competing with its own
    customers. Can you say “conflict of interest”?

    Microsoft sold Windows to PC manufacturers, until they got fed up that the
    PC manufacturers weren't making PCs that made good use of their OS. So they released the Surface line. Microsoft make a decent income from that, but PC manufacturers did not go bankrupt - in fact they incorporated many of the features from the Surface to make their products more attractive to
    customers (in the face of competition from in particular Apple). Everybody won.

    Theo

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Lawrence =?iso-8859-13?q?D=FFOlivei@21:1/5 to Theo on Fri Aug 22 22:19:29 2025
    On 22 Aug 2025 19:12:17 +0100 (BST), Theo wrote:

    Microsoft sold Windows to PC manufacturers, until they got fed up
    that the PC manufacturers weren't making PCs that made good use of
    their OS. So they released the Surface line. Microsoft make a decent
    income from that ...

    Somehow I doubt that. Given their so-so reviews, I would suspect they
    are a loss-making venture -- a corporate vanity project, no more.

    ... but PC manufacturers did not go bankrupt - in fact they
    incorporated many of the features from the Surface to make their
    products more attractive to customers ...

    Actually, Surface is an example to PC vendors of how *not* to make a
    successful product.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Lawrence =?iso-8859-13?q?D=FFOlivei@21:1/5 to Richard Kettlewell on Fri Aug 22 22:17:00 2025
    On Fri, 22 Aug 2025 09:10:07 +0100, Richard Kettlewell wrote:

    Lawrence D’Oliveiro <ldo@nz.invalid> writes:

    Richard Kettlewell wrote:

    Lawrence D’Oliveiro <ldo@nz.invalid> writes:

    Because in general-purpose computing, the proprietary vendors anyway
    (Microsoft being the obvious example) have a great deal of trouble
    supporting more than one platform.

    Apple have changed architecture three times now ...

    But never more than main one at a time -- not for longer than
    absolutely needed to make the transition.

    So what?

    Because I said they “have a great deal of trouble supporting more than one platform”.

    All the old HP minis and RISC architectures have gone. Itanium has
    gone. HP only does x86 now. And Sun has gone.

    Again, so what?

    Because I said they “have a great deal of trouble supporting more than one platform”.

    Windows-on-ARM is struggling. And Windows NT on non-x86 architectures
    never lasted long. Fun fact: even on x86, which is now 64-bit, Windows
    is still a 32-bit architecture at heart.

    You see what I mean, don’t you?

    You’re obviously not addressing the claim I actually made ...

    You were trying to refute what I said, which was they “have a great deal
    of trouble supporting more than one platform”.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Scott Dorsey@21:1/5 to invalid@invalid.invalid on Sat Aug 23 09:55:19 2025
    Richard Kettlewell <invalid@invalid.invalid> wrote:

    Apple have changed architecture three times now, HP and Sun at least
    twice each. Microsoft support two architectures at present and have >supported more in the past.

    In the case of HP and Sun, customers often used the architecture change as
    a reason to move to a competing product. If we have to recompile the code anyway, we might as well recompile it for a faster machine from a different company.

    Microsoft has over the years tried very hard to promote Windows on a variety
    of architectures from Alpha to ARM and in spite of spending a lot of money
    they have never really been able to get customers off of the x86.

    Apple is weird... Apple is selling the UI more than anything else, and they have actually changed architectures four times.... 68k->ppc->x86->m4 but
    each time they have made sure they had backwards compatibility for a few versions of the code on the old architecture. People want Apple because
    of the UI and they will put up with a lot of grief to have it.
    --scott

    --
    "C'est un Nagra. C'est suisse, et tres, tres precis."

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Lawrence =?iso-8859-13?q?D=FFOlivei@21:1/5 to All on Sat Aug 23 23:01:01 2025
    On Sat, 23 Aug 2025 15:07:19 +0100, Sn!pe wrote:

    ... decent interoperability within its own ecosystem.

    An oxymoron, if ever there was one ...

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Lawrence =?iso-8859-13?q?D=FFOlivei@21:1/5 to All on Sun Aug 24 21:53:35 2025
    On Sun, 24 Aug 2025 00:34:54 +0100, Sn!pe wrote:

    Lawrence D'Oliveiro <ldo@nz.invalid> wrote:

    On Sat, 23 Aug 2025 15:07:19 +0100, Sn!pe wrote:

    ... decent interoperability within its own ecosystem.

    An oxymoron, if ever there was one ...

    How else would you describe the family of iDevices that all talk to each other with ease straight from the box? [rhet.]

    Words like “proprietary”, “closed system” and “walled garden” come to
    mind ... [fact.]

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)