Re: [gccsdk] Unaligned data access vs pack(push...

In message <2b97a3c252.beeb@ron1954.woosh.co.nz>
Ron <beeb@woosh.co.nz> wrote:

> I am attempting to compile LZ4 compression, and the working binary
> doesn't get it's arithmetic right, decompressing declares corruption.
> The warnings
>
> warning: #pragma pack(push[, id], <n>) is not supported on this target
> warning: #pragma pack(pop[, id], <n>) is not supported on this target
>
> can be avoided by using -D__ARM_FEATURE_UNALIGNED but according to
> http://infocenter.arm.com/help/topic/com.arm.doc.dui0472c/BABFDBCJ.html
> this could be a problem for earlier than ARMv6 also.
>
> I'm not keen on starting on C++ stuff, This is a fairly small project,
> but requires manually configuring.
>
> One other config question is wether we support hardware bit counting,
> in case that means something to anyone?
>

I have found an arm patch for an older release (r33) that is working
on my Iyonix. I haven't tested it for speed yet.
It doesn't touch on the requirements that stops the trunk release.


The ARM_FEATURE_UNALIGNED reportedly will be on the latest gcc and
can increase the compression speed by 50% on processors that support
it (arm6 and higher I think)

LZ4 is supposed to be the fastest kid on the block, it is in
new linux kernels and can be enabled with some new file systems.

I noticed a wiki on Endian with a list of OS's doesn't include
RISC OS, perhaps someone knowledgible could add something there?

I found some endian tests. The Iyonix output the same as X86 for
short int but not long int. The test writer's arm (non RISCOS)
was actually different on short int's to the X86 also.
I'm sort of surprised I found a working solution so easily.

Thanks, Ron M.

_______________________________________________
GCCSDK mailing list gcc@gccsdk.riscos.info
Bugzilla: http://www.riscos.info/bugzilla/index.cgi
List Info: http://www.riscos.info/mailman/listinfo/gcc
Main Page: http://www.riscos.info/index.php/GCCSDK