Hello Robert,
I've got question about two ptxconfigs:
For OSELAS.Toolchain-1.99.3.1 there are :
1).
arm-v5te-linux-gnueabi_gcc-4.3.2_glibc-2.8_binutils-2.18_kernel-2.6.27-sanitized.ptxconfig
(which I'm using now)
and very similar one:
2).
arm-v5te_vfp-linux-gnueabi_gcc-4.3.2_glibc-2.8_binutils-2.18_kernel-2.6.27-sanitized.ptxconfig.
As I said before, I'm using first toolchain (1.).
It looks that it isn't working as it should. It sometimes happens that
very heavy computational operations hangs (for instance Kalman filtering
with a lot of doubles) or calculations aren't performed (no filtered
data on the output).
My linux kernel is configured to only support NwFPE floating point
emulation (It is mandatory to use at least one emulation). I don't want
to use kernel trap/exception to handle floating point. I'd prefer to
use floating point handling as library calls. Toolchain which I'm using
now has PTXCONF_GLIBC_CONFIG_EXTRA="--without-fp" , so maybe this is
the problem?
I need full software support for floating point operations (since my
arm-v5 hasn't got FPU) as efficient as possible.
I'm compiling my code with -msoft-float option.
Could you give me advice and tell which toolchain I should use?
Regards,
Lukasz
--
ptxdist mailing list
[email protected]
