Disable float_to_int16_interleave_sse2 in favor of
float_to_int16_interleave_sse, since it crashes for reasons I would have no clue about. Fixes a few duplicated tickets, which I'll have to sort out later. git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@38636 a95241bf-73f2-0310-859d-f6bbb57e9c96
This commit is contained in:
@@ -2878,7 +2878,9 @@ void dsputil_init_mmx(DSPContext* c, AVCodecContext *avctx)
|
||||
if(mm_flags & FF_MM_SSE2){
|
||||
c->int32_to_float_fmul_scalar = int32_to_float_fmul_scalar_sse2;
|
||||
c->float_to_int16 = float_to_int16_sse2;
|
||||
c->float_to_int16_interleave = float_to_int16_interleave_sse2;
|
||||
// NOTE (Haiku modification): Disabled because it crashes.
|
||||
// c->float_to_int16_interleave = float_to_int16_interleave_sse2;
|
||||
c->float_to_int16_interleave = float_to_int16_interleave_sse;
|
||||
#if HAVE_YASM
|
||||
c->scalarproduct_int16 = ff_scalarproduct_int16_sse2;
|
||||
c->scalarproduct_and_madd_int16 = ff_scalarproduct_and_madd_int16_sse2;
|
||||
|
||||
Reference in New Issue
Block a user