Skip to content

GitLab

Commit 97946622 authored by James Zern's avatar James Zern
Browse files
Options

Revert "mips msa vp9 subpel variance optimization" 

This reverts commit a42df86c.

this change causes MSA/VP9SubpelVarianceTest.Ref and
MSA/VP9SubpelVarianceTest.ExtremeRef failures under
mips32r5el-msa-linux-gnu and mips64r6el-msa-linux-gnu

Change-Id: I40b71a0b774eaeb31f66f795733f95cf360909f7
parent ced98264
Expand all Hide whitespace changes
Inline Side-by-side
Showing with 13 additions and 828 deletions
test/variance_test.cc
View file @ 97946622
......@@ -2058,43 +2058,5 @@ INSTANTIATE_TEST_CASE_P(
make_tuple(3, 2, variance8x4_msa, 0),
make_tuple(2, 3, variance4x8_msa, 0),
make_tuple(2, 2, variance4x4_msa, 0)));
#if CONFIG_VP9_ENCODER
const SubpixVarMxNFunc subpel_variance4x4_msa = vp9_sub_pixel_variance4x4_msa;
const SubpixVarMxNFunc subpel_variance4x8_msa = vp9_sub_pixel_variance4x8_msa;
const SubpixVarMxNFunc subpel_variance8x4_msa = vp9_sub_pixel_variance8x4_msa;
const SubpixVarMxNFunc subpel_variance8x8_msa = vp9_sub_pixel_variance8x8_msa;
const SubpixVarMxNFunc subpel_variance8x16_msa = vp9_sub_pixel_variance8x16_msa;
const SubpixVarMxNFunc subpel_variance16x8_msa = vp9_sub_pixel_variance16x8_msa;
const SubpixVarMxNFunc subpel_variance16x16_msa =
vp9_sub_pixel_variance16x16_msa;
const SubpixVarMxNFunc subpel_variance16x32_msa =
vp9_sub_pixel_variance16x32_msa;
const SubpixVarMxNFunc subpel_variance32x16_msa =
vp9_sub_pixel_variance32x16_msa;
const SubpixVarMxNFunc subpel_variance32x32_msa =
vp9_sub_pixel_variance32x32_msa;
const SubpixVarMxNFunc subpel_variance32x64_msa =
vp9_sub_pixel_variance32x64_msa;
const SubpixVarMxNFunc subpel_variance64x32_msa =
vp9_sub_pixel_variance64x32_msa;
const SubpixVarMxNFunc subpel_variance64x64_msa =
vp9_sub_pixel_variance64x64_msa;
INSTANTIATE_TEST_CASE_P(
MSA, VP9SubpelVarianceTest,
::testing::Values(make_tuple(2, 2, subpel_variance4x4_msa, 0),
make_tuple(2, 3, subpel_variance4x8_msa, 0),
make_tuple(3, 2, subpel_variance8x4_msa, 0),
make_tuple(3, 3, subpel_variance8x8_msa, 0),
make_tuple(3, 4, subpel_variance8x16_msa, 0),
make_tuple(4, 3, subpel_variance16x8_msa, 0),
make_tuple(4, 4, subpel_variance16x16_msa, 0),
make_tuple(4, 5, subpel_variance16x32_msa, 0),
make_tuple(5, 4, subpel_variance32x16_msa, 0),
make_tuple(5, 5, subpel_variance32x32_msa, 0),
make_tuple(5, 6, subpel_variance32x64_msa, 0),
make_tuple(6, 5, subpel_variance64x32_msa, 0),
make_tuple(6, 6, subpel_variance64x64_msa, 0)));
#endif // CONFIG_VP9_ENCODER
#endif // HAVE_MSA
} // namespace
vp9/common/vp9_rtcd_defs.pl
View file @ 97946622
......@@ -805,80 +805,80 @@ if (vpx_config("CONFIG_VP9_ENCODER") eq "yes") {
# variance
add_proto qw/unsigned int vp9_sub_pixel_variance64x64/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
specialize qw/vp9_sub_pixel_variance64x64 avx2 neon msa/, "$sse2_x86inc", "$ssse3_x86inc";
specialize qw/vp9_sub_pixel_variance64x64 avx2 neon/, "$sse2_x86inc", "$ssse3_x86inc";
add_proto qw/unsigned int vp9_sub_pixel_avg_variance64x64/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, const uint8_t *second_pred";
specialize qw/vp9_sub_pixel_avg_variance64x64 avx2/, "$sse2_x86inc", "$ssse3_x86inc";
add_proto qw/unsigned int vp9_sub_pixel_variance32x64/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
specialize qw/vp9_sub_pixel_variance32x64 msa/, "$sse2_x86inc", "$ssse3_x86inc";
specialize qw/vp9_sub_pixel_variance32x64/, "$sse2_x86inc", "$ssse3_x86inc";
add_proto qw/unsigned int vp9_sub_pixel_avg_variance32x64/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, const uint8_t *second_pred";
specialize qw/vp9_sub_pixel_avg_variance32x64/, "$sse2_x86inc", "$ssse3_x86inc";
add_proto qw/unsigned int vp9_sub_pixel_variance64x32/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
specialize qw/vp9_sub_pixel_variance64x32 msa/, "$sse2_x86inc", "$ssse3_x86inc";
specialize qw/vp9_sub_pixel_variance64x32/, "$sse2_x86inc", "$ssse3_x86inc";
add_proto qw/unsigned int vp9_sub_pixel_avg_variance64x32/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, const uint8_t *second_pred";
specialize qw/vp9_sub_pixel_avg_variance64x32/, "$sse2_x86inc", "$ssse3_x86inc";
add_proto qw/unsigned int vp9_sub_pixel_variance32x16/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
specialize qw/vp9_sub_pixel_variance32x16 msa/, "$sse2_x86inc", "$ssse3_x86inc";
specialize qw/vp9_sub_pixel_variance32x16/, "$sse2_x86inc", "$ssse3_x86inc";
add_proto qw/unsigned int vp9_sub_pixel_avg_variance32x16/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, const uint8_t *second_pred";
specialize qw/vp9_sub_pixel_avg_variance32x16/, "$sse2_x86inc", "$ssse3_x86inc";
add_proto qw/unsigned int vp9_sub_pixel_variance16x32/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
specialize qw/vp9_sub_pixel_variance16x32 msa/, "$sse2_x86inc", "$ssse3_x86inc";
specialize qw/vp9_sub_pixel_variance16x32/, "$sse2_x86inc", "$ssse3_x86inc";
add_proto qw/unsigned int vp9_sub_pixel_avg_variance16x32/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, const uint8_t *second_pred";
specialize qw/vp9_sub_pixel_avg_variance16x32/, "$sse2_x86inc", "$ssse3_x86inc";
add_proto qw/unsigned int vp9_sub_pixel_variance32x32/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
specialize qw/vp9_sub_pixel_variance32x32 avx2 neon msa/, "$sse2_x86inc", "$ssse3_x86inc";
specialize qw/vp9_sub_pixel_variance32x32 avx2 neon/, "$sse2_x86inc", "$ssse3_x86inc";
add_proto qw/unsigned int vp9_sub_pixel_avg_variance32x32/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, const uint8_t *second_pred";
specialize qw/vp9_sub_pixel_avg_variance32x32 avx2/, "$sse2_x86inc", "$ssse3_x86inc";
add_proto qw/unsigned int vp9_sub_pixel_variance16x16/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
specialize qw/vp9_sub_pixel_variance16x16 neon msa/, "$sse2_x86inc", "$ssse3_x86inc";
specialize qw/vp9_sub_pixel_variance16x16 neon/, "$sse2_x86inc", "$ssse3_x86inc";
add_proto qw/unsigned int vp9_sub_pixel_avg_variance16x16/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, const uint8_t *second_pred";
specialize qw/vp9_sub_pixel_avg_variance16x16/, "$sse2_x86inc", "$ssse3_x86inc";
add_proto qw/unsigned int vp9_sub_pixel_variance8x16/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
specialize qw/vp9_sub_pixel_variance8x16 msa/, "$sse2_x86inc", "$ssse3_x86inc";
specialize qw/vp9_sub_pixel_variance8x16/, "$sse2_x86inc", "$ssse3_x86inc";
add_proto qw/unsigned int vp9_sub_pixel_avg_variance8x16/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, const uint8_t *second_pred";
specialize qw/vp9_sub_pixel_avg_variance8x16/, "$sse2_x86inc", "$ssse3_x86inc";
add_proto qw/unsigned int vp9_sub_pixel_variance16x8/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
specialize qw/vp9_sub_pixel_variance16x8 msa/, "$sse2_x86inc", "$ssse3_x86inc";
specialize qw/vp9_sub_pixel_variance16x8/, "$sse2_x86inc", "$ssse3_x86inc";
add_proto qw/unsigned int vp9_sub_pixel_avg_variance16x8/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, const uint8_t *second_pred";
specialize qw/vp9_sub_pixel_avg_variance16x8/, "$sse2_x86inc", "$ssse3_x86inc";
add_proto qw/unsigned int vp9_sub_pixel_variance8x8/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
specialize qw/vp9_sub_pixel_variance8x8 neon msa/, "$sse2_x86inc", "$ssse3_x86inc";
specialize qw/vp9_sub_pixel_variance8x8 neon/, "$sse2_x86inc", "$ssse3_x86inc";
add_proto qw/unsigned int vp9_sub_pixel_avg_variance8x8/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, const uint8_t *second_pred";
specialize qw/vp9_sub_pixel_avg_variance8x8/, "$sse2_x86inc", "$ssse3_x86inc";
# TODO(jingning): need to convert 8x4/4x8 functions into mmx/sse form
add_proto qw/unsigned int vp9_sub_pixel_variance8x4/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
specialize qw/vp9_sub_pixel_variance8x4 msa/, "$sse2_x86inc", "$ssse3_x86inc";
specialize qw/vp9_sub_pixel_variance8x4/, "$sse2_x86inc", "$ssse3_x86inc";
add_proto qw/unsigned int vp9_sub_pixel_avg_variance8x4/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, const uint8_t *second_pred";
specialize qw/vp9_sub_pixel_avg_variance8x4/, "$sse2_x86inc", "$ssse3_x86inc";
add_proto qw/unsigned int vp9_sub_pixel_variance4x8/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
specialize qw/vp9_sub_pixel_variance4x8 msa/, "$sse_x86inc", "$ssse3_x86inc";
specialize qw/vp9_sub_pixel_variance4x8/, "$sse_x86inc", "$ssse3_x86inc";
add_proto qw/unsigned int vp9_sub_pixel_avg_variance4x8/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, const uint8_t *second_pred";
specialize qw/vp9_sub_pixel_avg_variance4x8/, "$sse_x86inc", "$ssse3_x86inc";
add_proto qw/unsigned int vp9_sub_pixel_variance4x4/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
specialize qw/vp9_sub_pixel_variance4x4 msa/, "$sse_x86inc", "$ssse3_x86inc";
specialize qw/vp9_sub_pixel_variance4x4/, "$sse_x86inc", "$ssse3_x86inc";
#vp9_sub_pixel_variance4x4_sse2=vp9_sub_pixel_variance4x4_wmt
add_proto qw/unsigned int vp9_sub_pixel_avg_variance4x4/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, const uint8_t *second_pred";
......
vp9/encoder/mips/msa/vp9_variance_msa.c deleted 100644 → 0
View file @ ced98264
This diff is collapsed.
vp9/vp9cx.mk
View file @ 97946622
......@@ -163,6 +163,5 @@ VP9_CX_SRCS-$(HAVE_MSA) += encoder/mips/msa/vp9_fdct32x32_msa.c
VP9_CX_SRCS-$(HAVE_MSA) += encoder/mips/msa/vp9_fdct_msa.h
VP9_CX_SRCS-$(HAVE_MSA) += encoder/mips/msa/vp9_subtract_msa.c
VP9_CX_SRCS-$(HAVE_MSA) += encoder/mips/msa/vp9_temporal_filter_msa.c
VP9_CX_SRCS-$(HAVE_MSA) += encoder/mips/msa/vp9_variance_msa.c
VP9_CX_SRCS-yes := $(filter-out $(VP9_CX_SRCS_REMOVE-yes),$(VP9_CX_SRCS-yes))
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment