memcpy performance (fwd)

[Joel Sherrill]

On Tue, 9 Dec 1997, Jonathan Larmour wrote:

> Joel Sherrill wrote:
> > 
> > I thought I would pass this on.  Does the new version of memcpy do much
> > better than this?
> > 
> > ---------- Forwarded message ----------
> > Date: Tue,  9 Dec 97 12:03:28 -0600
> > From: Eric Norum <eric@skatter.USask.Ca>
> > To: rtems-list@oarcorp.com
> > Subject: Re: memcpy performance
> > 
> > It's even worse than just a byte-by-byte copy!
> > 
> > On the 971024 snapshot (gen68360 BSP) a call to memcpy produces:
> [...a lot of unnecessary wastefulness]
> 
> Is he compiling with -O? If so, then gcc should include its own builtin
> memcpy instead of the newlib one AIUI.
> 
> This may not be true for that target possibly. Anyone know?

This complaint is about code generated for a call inside RTEMS.  We always
compile at the highest optimization level which does not cause gcc to core
dump or generate incorrect code. :)

For the target board Eric uses (68360), that means RTEMS is compiled with:

-m68020 -mnobitfield -msoft-float -O4 -fomit-frame-pointer

I will verify the toolset and args just in case though.


--joel