[J-core] J-core roadmap (was Re: Google Fucsia and Toybox)

Rob Landley rob at landley.net
Tue Aug 16 17:44:59 EDT 2016


On 08/16/2016 03:39 PM, Roberto A. Foglietta wrote:
> 2016-08-16 20:57 GMT+01:00 Rob Landley <rob at landley.net
>> On 08/16/2016 02:03 PM, Roberto A. Foglietta wrote:
>> > 2016-08-16 18:36 GMT+01:00 Rob Landley <rob at landley.net>:
>> >
>> >     (I'm also tempted to try to get it running on the superh architecture,
>> >     since I'm helping with http://j-core.org and *j3 should be out with mmu
>> >     * next year. Porting go to that seemed like a can of worms (last time I
>> >     looked it was full of hardwired glibc and architecture assumptions, but
>> >     that was a while ago), but if I don't need go...)
>> >
>> > Considering this
>> >
>> > /The j2 is a nommu processor because sh2 (the processor in the Sega
>> > Saturn game console) was, and the last sh2 patent expired in October 2014. /
>> > */The sh4 processor (dreamcast) has an mmu, but the last sh4 patents
>> > don't exire until 2016./*
>> >
>> > I suggest to call it J4 and use sh4 instruction set
>> > ;-)
>>
>> There's some debate over that. :)
>>
>> There's a http://j-core.org/roadmap.html still subject to revision.
>> Right now they're working on some products using J2, and when they catch
>> up on that the next thing up is J1 (arduino country) which is actually
>> mostly toolchain issues, because they want to use a fully open source
>> VHDL toolchain to build it instead of Xilinx's "free download"
>> toolchain, and the closest they've gotten is an open source simulator
>> that won't QUITE connect with the open source code generator (abstract
>> syntax tree conversion's still missing some pieces).
>>
>> The next _bigger_ version is j3/sh3, adding mmu, and grabbing some
>> features from the Big Bucket 'o Todo Items.
>>
>> j4 gives us a target to work on multi-issue, which is a big engineering
>> thing that only shows up to users as a performance increase. so we
>> _could_ make j3 use the sh4 instruction set instead of sh3, but it would
>> be confusing. :)
>>
>> From an instruction set perspective, j3 and j4 are pretty close, and
>> we'll probably go ahead and do the full thing. But that's a decision for
>> next year.
>>
>> It's possible that we'll jump straight to an sh4-compatible j4 and bump
>> multi-issue to j64, but sh4 _did_ do multi-issue, so....
>>
>> The roadmap of features to implement is reasonably stable, but sequence
>> points where we decide to cut releases are a bit fuzzy this far out.
>>
>> Rob
>>
>> P.S. Mind if I cc: this reply to the j-core mailing list?
> 
> It is your reply, so you can CC to everyone you like
> :-)

And lo...

Rob


More information about the J-core mailing list