Hi Michel, Thanks for your reply. Michel Dänzer <michel at daenzer.net> writes: > Not sure what kind of compatibility you mean. I was referring to the gfx_v6 driver, which I believed was already BE compatiple (although this may have been a misunderstanding), using a different mechanism, which would then clash with the "byteswap in amdgpu_ring_write" method, since the amdgpu_ring_write function is shared and not virtualized via the ring ops. > Anyway, AFAIK all GPUs supported by amdgpu simply don't have these byte > swapping capabilities anymore. This will pose a much bigger challenge in > the shader related code in Mesa. I find it strange that they would still have the bits in the registers then. In v8 they were even moved to a different bit offset; if they were not supported why were they not simply removed? But if the capability is definitely not there, then I guess I'll go for the software swapping route. I'll consider it a success if I get /dev/fb, Mesa is an adventure for a different day. :-) > My advice would be to reconsider whether whatever reasons you have for > running in big endian mode are worth the pain. :) I'm reconsidering if RX 550 is really worth the trouble when I could just use an R7 which has a known working driver (including 3D), but it also feels like a bit of a cop out to give up after having good progress so far... // Marcus -- -------------------------------------------------------------------------- Marcus Comstedt Lysator Academic Computer Society marcus at lysator.liu.se Linköping University, Sweden // ----------------------------------------------------------------------\X/-
