Optimize aom_highbd_quantize_b_64x64 module
Added SSE2 variant for aom_highbd_quantize_b_64x64_c
When tested for multiple test cases observed 0.3%
average reduction in encoder time for speed = 4 preset.
Module gains improved by factor of ~1.5x w.r.t C code.
Change-Id: If17d4cd853f4db4d3caa2d1977939d441961558a
diff --git a/test/quantize_func_test.cc b/test/quantize_func_test.cc
index 2975c1d..e7afed9 100644
--- a/test/quantize_func_test.cc
+++ b/test/quantize_func_test.cc
@@ -398,6 +398,12 @@
TX_32X32, TYPE_B, AOM_BITS_10),
make_tuple(&aom_highbd_quantize_b_32x32_c, &aom_highbd_quantize_b_32x32_sse2,
TX_32X32, TYPE_B, AOM_BITS_12),
+ make_tuple(&aom_highbd_quantize_b_64x64_c, &aom_highbd_quantize_b_64x64_sse2,
+ TX_64X64, TYPE_B, AOM_BITS_8),
+ make_tuple(&aom_highbd_quantize_b_64x64_c, &aom_highbd_quantize_b_64x64_sse2,
+ TX_64X64, TYPE_B, AOM_BITS_10),
+ make_tuple(&aom_highbd_quantize_b_64x64_c, &aom_highbd_quantize_b_64x64_sse2,
+ TX_64X64, TYPE_B, AOM_BITS_12)
};
INSTANTIATE_TEST_CASE_P(SSE2, QuantizeTest,
