2008/10/8  <[EMAIL PROTECTED]>:
> Why?  Because they've given the reason why they MUST tackle it in
> another context:  they've refused to implement the latest Javascript
> features on the grounds that Chrome must reach the widest possible
> audience.  Well, by the same logic, they cannot afford to ignore a
> very significant CPU architecture.

Chrome needs to support exactly 1 architecture to reach the widest
possible audience, and that is x86. PowerPC is currently not used in
environments where a desktop browser is relevant, so it is extremely
unlikely that the Chrome team has this as any kind of priority. At the
same time, all operating systems that support x86-64 support a 32-bit
compatibility mode without significant performance impact.

The PowerPC architecture is currently only significant in the gaming
console market, and this is not a market where Chrome is relevant at
all.

> Also, I rather doubt that it's a major task to port to PowerPC, given
> that V8 already supports two architectures.  It may be a major task to
> do it as efficiently as for x86, sure, but colossal speed is not a
> mandatory requirement when the main goal is to reach the widest
> possible audience.  Even a noddy third architecture, or fourth after
> x86_64, will do.

The internal structures needed to support multiple compilation targets
are of course in place, but it is _NOT_ trivial to implement codegen
for any high level language, for any architecture! So no, this will
not happen by itself. :-)
It's not really a question of speed, as much as generating code that
doesn't segfault half the time.

- Simon

--~--~---------~--~----~------------~-------~--~----~
v8-users mailing list
[email protected]
http://groups.google.com/group/v8-users
-~----------~----~----~----~------~----~------~--~---

Reply via email to