UltimatePPC and everything else

[AJCopland]

To be honest i am reluctant to show enthusism with the NatAmi recently being cancelled. I was so enthusiastic about the NatAmi but now i'm gutted. I don't want another emotional rollercoaster ride. What kind of guarantees can they give this will be released?

Cancelled? Thomas is off working on it alone a bit more but he's been doing that every few months thoughout the project anyway. It's no more real but certainly not cancelled.

All the bitching and whining around it is just that the other project members were hoping to be able to contribute to the AHDL side of building up the (S)AGA side of things and they aren't.

None of that affects it development though, that's always been 100% Thomas driven and subject to whatever else is happening in his work/love/personal/hobby/etc life.

When you say cancelled there's still 7 working developer boards with 68060s, some of them clocked to within an inch of 100Mhz, that people are using to write test cases and PCI/usb/network drivers. That's far from cancelled, it's just a huge job and no-one has a clue when it will become commercially available.

Andy

[AJCopland]

I've just been reading up on the included CPU (QoriQ P1013 e500v2). It does have an FPU, but it's incompatible to pretty much any other PowerPC FPU out there (Power ISA).

TL;DR: Every binary using FPU needs to be recompiled to run on this CPU.

This will definitely create huge challenges if you ever attempt to run any existing PowerUP, WarpUP, MorphOS or OS4 binaries.

Challenges yes but not huge challenges, unless the entire program/library is written in PPC assembler. I wonder why they went with that particular CPU though if they knew that it has these issues. Maybe they only found as out they were developing it or they just don't view it as worth the tradeoff to use another another CPU. If a compatible CPU costs more, or doesn't run as fast then it might not be worth using it.

[AJCopland]

Yes, a *huge* challenge! Adapting and porting the OS and all applications for this new platform (yes, you could consider this a platform change) and then *keep* this fork (of some 100 users or so) supported in parallel in the future, will be a big undertaking. And if you are about to consider this ISA-change anyway, you should as well consifering changing to some better, faster, future-proof ISA like the x86 (or ARM even).

Changing the PPC CPU in the design would be a far less of a challange...

I was responding to Piru's comment about having to recompile software. Since I've developed projects across 4 platforms simultaneously I know that it really is pretty simple even where you're having to mix 4 different compilation environments and different instruction sets.

A huge challenge would be to get existing software running that was targeted at an incompatible ISA, that ISN'T what I was referring too.

Andy

[AJCopland]

I was talking about existing binaries of which you don't have source code for (for whatever reason) and thus cannot recompile.

Ah yeah, then we're screwed would patching be possible for the PowerPC side? I know people have tried with 68k code but that has complications due to it's mixing of code and data, not to mention self modifying code.

Think I got your original post mixed up about the two things you were talking about.