[ELDK] correct/best ELDK architecture for MPC8545E?

[Anthony Foiani]

Greetings, Wolfgang --

>>>>> "WD" == Wolfgang Denk <wd at denx.de> writes:

WD> You should use ppc_85xxDP for optimal performance.

Thank you very much for the quick answer!

(Of course, having done my research and bothered everyone on the list,
they might be changing chips for the final product; the replacement
chip is in the same family, though, so it should also have the e500v2
core.)

WD> What ["DP" is] talking about here is the instruction set of the
WD> Signal Processing Engine (SPE). The MPC8545E has a rev. 2 SPE,
WD> which includes support for double precision, while older 85xx
WD> (say, MPC8560) use a rev. 1 SPE which supports only single
WD> precision (use ppc_85xx for these).

Ok, that makes much more sense.

WD> Of course it makes for nice marketing to be able to say: yes, we
WD> do have FP support.  Marketing does not want to put a straight
WD> "but we do not have any real Floating Point Unit with things like
WD> FP registers and the like" on their nice presentations.

*laugh*

Point taken.  Thanks for cutting through the marketing double-speak
for me!

WD> There is no FPU in your processor, just a SPEv2, which can be used
WD> for FP operations, including double precision). But keep in mind
WD> that the SPE does not provide features like FP registers as you
WD> would get with a FPU, so all parameter passing has to be done
WD> using general purpose registers and the stack, which has impact on
WD> the performance.

Understood, and thank you for the detailed explanation!

WD> See also http://www.denx.de/wiki/DULG/ELDKCheckFloatingPointSupport

Right.

Thanks very much once again.

Regards,
Tony