Re: glibc - capaibility control mechanism
At Sat, 2 Apr 2005 10:31:32 +0200,
Bastian Blank wrote:
> On Sat, Apr 02, 2005 at 03:44:22PM +0900, GOTO Masanori wrote:
> > However it just changes HWCAP value - so we need to make ld.so to
> > search library path using some library search path modifying
> > environment variable (ex: LD_LIBRARY_PATH).
>
> No, this will be fucking slow, thats why my proposal only uses them with
> entries in the cache file.
Could you explain me why it'll be too slow?
> > I think the simple shell script wrapper controlling HWCAP_MASK can
> > achieve the original request by Bastian.
>
> No, it does not. You have to replace any binary with this wrapper, even
> /sbin/init.
Hmm, I have not understood why you want to use such mechanism. What's
the merit of your proposal? Which application are you focusing in
this proposal?
> > Sometimes speed-crazy developers want to
> > play with optimized libraries only changing gcc optimized options.
>
> SSE gains between 50 and 70% in this case, don't call that speed-crazy.
>
> > So we should be care to enable SSE everytime as well.
>
> You have to ask the kernel/cpu for available SSE support.
BTW, I expect you to read my point - if you have experience to modify
the code for specific architecture's capability to get performance
improvement like HPC application, you would nod some parts of my
previous mail.
Regards,
-- gotom
Reply to: