On Wed, 2012-10-24 at 18:50 +0530, Arun Raghavan wrote: > On Tue, 2012-10-23 at 17:54 +0200, Peter Meerwald wrote: > > From: Peter Meerwald <p.meerwald at bct-electronic.com> > > > > v4: > > * fix for sample length < 4 > > v3: > > * convert from intrinsics to inline assembly > > v2: > > * load and store data with vld1/vld1q and vst1/vst1q, resp., to work > > around alignment issues of compiler-generated vldmia instruction > > * remove redundant check for NEON flags > [...] > > conversion may be off by one for some samples due to rounding issues > > I'm not sure about the impact of this. Is there no way to avoid the > rounding error? Thinking about it some more, I think having sconv be exactly the same across implementations is important. Is this some sort of NEON quirk? >From what I can tell, the result of the vmul is off-by-one. -- Arun
