[linux-cirrus] Re: Yet another MaverickCrunch hardware bug?

Martin Guy wrote:
> Got it. The old workaround code to avoid placing maverick instructions
> in the two slots following a branch was buggy (well, actually, just
> wrong!). Now paranoia compiled with -mieee says:
>
> -----
> No failures, defects nor flaws have been discovered.
> Rounding appears to conform to the proposed IEEE standard P754.
> The arithmetic diagnosed appears to be Excellent!
> ----
> 
> \O/
> 
> I've also changed it so it always enables the workaround for revision
> D1 E1 E2 E3 silicon when generating code for Maverick, since all the
> silicon produced needs them.
> The -mcirrus-fix-invalid-insns flag is now only necessary when
> compiling for the rare d0 revision (like it always said in the manual)
> 
> Patches and new tarballs are at http://simplemachines.it/tools and
> thanks to them for funding this work.

Great work. What is the status of Maverick Crunch using the patches? ie,
are all floating operations handled in hardware, or are some done in
software? What sort of performance gain is involved in using your
toolchain for floating point operations over software emulation?

~Ryan

-- 
Bluewater Systems Ltd - ARM Technology Solution Centre

       Ryan Mallon                              Unit 5, Amuri Park
       Phone: +64 3 3779127                     404 Barbadoes St
       Fax:   +64 3 3779135                     PO Box 13 889
       Email: ryan@xxxxxxxxxxxxxxxx             Christchurch, 8013
       Web:   http://www.bluewatersys.com       New Zealand
       Freecall Australia  1800 148 751         USA 1800 261 2934

Other related posts: