| From: Evan Leibovitch via talk <[email protected]>
| On Fri, Feb 2, 2024 at 11:38 AM D. Hugh Redelmeier via talk <[email protected]>
wrote:
|
| For serious applications, the openness RISC-V helps but doesn't make
everything you
| need open and free. Or even available. You actually need chip designs
-- what SiFive
| sells.
|
|
| If the RISC-V design is open source, what is SiFive selling? Something easily
copyable? Support
| and documentation?
Scott has answered this, but I will expand on it.
All important processor architectures are well documented: so we know what
the processor is supposed to do.
To implement it (in a chip, for example) and sell it, you need to have the
right to do so. For commercially important architectures, these rights
cost real money or are unavailable. ARM is the only one that sells rights
at a price that is worth paying. RISC-V is becoming commercially
important and those rights are free.
Once you have the right to implement it, you then have the hard work of
actually designing and producing a system (including a processor) that
will actually run programs for that architecture.
You've seen photomicrographs of processor chips. All those blobs are not
random: they are the product of a lot of work. Not unlike writing a
program. And to make such a design that meets performance goals
multiplies the work and experience necessary. There are not many
teams with that experience.
All that work needs eventually to be paid for by a sufficiently promising
market, one that doesn't exist now. Chicken-and-egg. So a successful
process involves a step-by-step development of capabilities and markets.
We depend on something like venture capitalists or governments to prime
that pump.
| ARM has a vibrant ecosystem with all these things available for
licensing. And ARM
| doesn't seem to be too greedy. Even so, it has taken a long time to
get ARM
| processors that match x86 at the high end.
|
|
| Assuming that ARM's time to implement was not a matter of laziness, doesn't
its experience suggest
| that RISC-V will take similarly long -- or longer -- to evolve from low-level
SBCs to high-end
| computing?
|
| So: if you want a short time to delivery, ARM is way ahead.
|
|
| Plus, ARM has clients such as Qualcomm and Samsung and Apple that have
decades of experience in
| implementation of the architecture and lots of high-quality fabs.
Not really.
High-end processor fabrication is now a specialized business. Only
Samsung, TSMC, and Intel have the capability as far as I know. There are
stories out of China suggesting that they are trying really hard to get
there (SMIC, for example).
Most ARM processor designs are done by ARM in-house and are licensed at
rates that seem to be good enough for the market. Interestingly, ARM
charges considerably more for you to use their architecture without using
one of their designs. The few companies with that second license include
Apple (who got their license by being a fo-founder of ARM as we know it)
and Qualcomm (I think).
This second license requires conviction that your processor design team
can do better than an off-the-shelf design for your application.
A few companies have tried and failed (eg. AMD). Of the shelf ARM designs
have not been good enough to break into the datacentre but a couple of
non-ARM designs are doing OK (Amazon's Graviton, for example).
Note: users of the second license could just use RISC-V instead, for free.
In my opinion, the software support for RISC-V is mature. Linux supports
it as a first class architecture and that's all I need. Android supports
it or soon will (I don't remember which). Microsoft does not support it,
as far as I know.
| So the definition of "short" here could be an understatement.
Yes. ARM has taken a lot longer than I expected.
The NetWinder was a credible machine for desktop or server Linux something
like 25 years ago. It failed for several reasons but one was that there
was no ARM implementation that got near x86 performance.
Remember: that same fact killed off all the RISC desktops (HP, MIPS,
SPARC, Power) and 68k, NS32032, etc.
| If you think more strategically, RISC-V has some advantages.
|
|
| Not sure I gather this conclusion from the rationale.
|
| The US used a foot-gun on Huawei by banning ARM from dealing with
Huawei. The largest
| damage is to ARM: China can no longer think of ARM as a reliable
partner. So China
| will switch to RISC-V (there really isn't
| a better choice).
|
|
| Sure, this means that Chinese R&D will focus on RISC-V. As will that of its
military clients and
| other lesser-aligned countries such as Pakistan, Russia and Brazil. OTOH, US
allies won't be
| sinking much into RISC-V for fear of running afoul of the same embargoes and
sanctions that have
| hit AMD, TSMC etc. As a result you have Western R&D entrenching around ARM,
such as the recent
| move by Nvidia to start making PC-speed ARM designs.
|
| China is wary of buying from a US company like SiFive.
NVidia is hamstrung selling to PRC. Long term, this will likely hurt the
US.
NVidia tried to buy ARM. That would have even further aided RISC-V
because NVidia competitors would be a bit wary of buying ARM designs.
| That wariness cuts both ways, given that China's ARM subsidiary unilaterally
declared
| "independence" from its parent in what I would call a blatant act of IP
theft. Why would anyone
| want to invest into an environment in which even the threat of that exists?
So now we have
| websites whose single purpose is to track the "re-shoring" of chip production
to US and
| US-friendly countries.
Right. The ARM China thing was crazy (we don't understand the importance
of corporate seals in China). It has been resolved (that's an old
article).
RISC-V can be used by both sides of this divide. ARM cannot.
China rightly views the capability of access to fast processor as
strategically important. You can be sure that they have more money than
venture capitalists. They will make it happen.
The US is trying to block PRC and they have somewhat clumsy defence in
depth. They block AMSL from shipping recent lithographic systems to PRC.
They block ARM. They are blocking chemicals for lithography from
Japanese companies. This buys time but it won't work long-term.
This will prime the RISC-V pump. But there are rumblings in the US of
kind of black-listing RISC V which is really sad.
| So, yeah, geopolitics have intervened to retard the progress of perhaps the
best-ever shot at
| major open hardware goodness. Shame.
Actually, it cuts both ways.
---
Post to this mailing list [email protected]
Unsubscribe from this mailing list https://gtalug.org/mailman/listinfo/talk
