On 05/09/2011 00:55, Travis Geiselbrecht wrote:
Anyway, my personal suggestion is to start with an ARMv7-a class cpu (cortex-a8 or cortex-a9). The mmu, cache operations, and SMP is well standardized in ARMv7. ARMv6 is a bit more loose with some of that, and requires a bit more per-core code. A Raspberry Pi appears to be an ARM11 (probably an arm1136jf-s) which is an ARMv6 implementation.
It's an ARM1176JZF-S, although I'm not particularly well-versed in the differences. It's annoying Broadcom didn't go with ARMv7, as their "GPU" (which is more just a general-purpose multimedia chip that runs GL state trackers and stuff itself AFAIK) is pretty much best in class.
I asked on the Raspberry-Pi forum about setting up just a 2D framebuffer which is apparently just a case of writing a pointer and width/height to a magic memory location, and the GPU will grab from there and stick it on the screen, so graphics should at least be reasonably straight-forward.
I'll probably get one when it's released and have a play, but I have no kernel experience at all, so I doubt I'll get very far.
However, as long as you start with most ARMv6 and ARMv7 based designs you wont have to deal with the real monsters: TLB and cache being non coherent.
You say above that ARMv6 is a bit more loose with some of the specs - do you think it would be significantly more work on the ARM11 then the Cortex-A8-based SoCs?
Simon
