Christopher Frost wrote:
>From the information I am told, the 'Cell' contains a normal PowerPC64 (same
> ISA), and it will not run a 32 bit kernel, but will run 32 bit code. So if
> the kernel ran on the PowerMac G5, it would also run on the PS3.
The current port is not being written to run on the PowerMac G5; there's
nothing
explicitly preventing it from doing so, but the port is very definitely being
written for a 32-bit PPC interface.
> The Cell information, is also released. Yet the kernel runs out of the
> hypervisor, Sony documents this in their "otheros" information. To also be
> quite honest, there is a lot of published papers on the cell structure and
> framework.. Yet only a base 64 bit kernel is needed to have it up and
> running, as the SPUFS is extra, and isn't a necessary to run the kernel.
I haven't looked at it, but having the kernel run under a hypervisor may be yet
another hurdle, depending upon how Sony implements their support.
> I appreciate the feedback, comments and concerns, but if there is no future
> for a 64bit kernel... What exactly is the point of this project again?
> Embedded is nice and all.. but that sure makes our community quite a bit
> smaller. I'd have trouble believing that Sun/Toshiba/IBM would be against
> increasing the benefits of their new, joint, venture. If it would be helpful
> I contact them, I'd be able to do that also. Yet, I of course wouldn't
> proceed unless this project actually had a solid roadmap.
I'm not saying a 64-bit kernel is impossible, just that it has not even been a
consideration in development of the current port. What this means is that it
would take a second port effort to port the 32-bit PPC port, once it's
completed, to a 64-bit system.
It would be great to have a 64-bit Solaris kernel running on PPC, but IMHO we
need to learn to walk (get the 32-bit port up) before we attempt to fly.
William Kucharski
