AnotherFoxGuy/webrtc-audio-processing
1
0
Fork 0
Code Pull Requests Activity

Fix up XMM intrinsics usage on MSVC

Browse Source 
Repplying 0a0050746bc20ef970b9f260d485e4367c7ba854 after M131 bump.
This commit is contained in:
Arun Raghavan 2023-12-15 16:06:05 -05:00
parent b5c48b97f6
commit ad563b095c
1 changed files with 20 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

30
webrtc/modules/audio_processing/aec3/matched_filter_avx2.cc
View File

@ -13,6 +13,16 @@
#include "modules/audio_processing/aec3/matched_filter.h" #include "modules/audio_processing/aec3/matched_filter.h"
#include "rtc_base/checks.h" #include "rtc_base/checks.h"
#ifdef _MSC_VER
// Visual Studio
#define LOOKUP_M128(v, i) v.m128_f32[i]
#define LOOKUP_M256(v, i) v.m256_f32[i]
#else
// GCC/Clang
#define LOOKUP_M128(v, i) v[i]
#define LOOKUP_M256(v, i) v[i]
#endif
namespace webrtc { namespace webrtc {
namespace aec3 { namespace aec3 {
@ -81,14 +91,14 @@ void MatchedFilterCore_AccumulatedError_AVX2(
s_inst_256_8 = _mm256_mul_ps(h_k_8, x_k_8); s_inst_256_8 = _mm256_mul_ps(h_k_8, x_k_8);
s_inst_hadd_256 = _mm256_hadd_ps(s_inst_256, s_inst_256_8); s_inst_hadd_256 = _mm256_hadd_ps(s_inst_256, s_inst_256_8);
s_inst_hadd_256 = _mm256_hadd_ps(s_inst_hadd_256, s_inst_hadd_256); s_inst_hadd_256 = _mm256_hadd_ps(s_inst_hadd_256, s_inst_hadd_256);
s_acum += s_inst_hadd_256[0]; s_acum += LOOKUP_M256(s_inst_hadd_256, 0);
e_128[0] = s_acum - y[i]; LOOKUP_M128(e_128, 0) = s_acum - y[i];
s_acum += s_inst_hadd_256[4]; s_acum += LOOKUP_M256(s_inst_hadd_256,4);
e_128[1] = s_acum - y[i]; LOOKUP_M128(e_128, 1) = s_acum - y[i];
s_acum += s_inst_hadd_256[1]; s_acum += LOOKUP_M256(s_inst_hadd_256,1);
e_128[2] = s_acum - y[i]; LOOKUP_M128(e_128, 2) = s_acum - y[i];
s_acum += s_inst_hadd_256[5]; s_acum += LOOKUP_M256(s_inst_hadd_256,5);
e_128[3] = s_acum - y[i]; LOOKUP_M128(e_128, 3) = s_acum - y[i];
__m128 accumulated_error = _mm_load_ps(a_p); __m128 accumulated_error = _mm_load_ps(a_p);
accumulated_error = _mm_fmadd_ps(e_128, e_128, accumulated_error); accumulated_error = _mm_fmadd_ps(e_128, e_128, accumulated_error);
@ -209,8 +219,8 @@ void MatchedFilterCore_AVX2(size_t x_start_index,
x2_sum_256 = _mm256_add_ps(x2_sum_256, x2_sum_256_8); x2_sum_256 = _mm256_add_ps(x2_sum_256, x2_sum_256_8);
s_256 = _mm256_add_ps(s_256, s_256_8); s_256 = _mm256_add_ps(s_256, s_256_8);
__m128 sum = hsum_ab(x2_sum_256, s_256); __m128 sum = hsum_ab(x2_sum_256, s_256);
x2_sum += sum[0]; x2_sum += LOOKUP_M128(sum, 0);
s += sum[1]; s += LOOKUP_M128(sum, 1);
// Compute the matched filter error. // Compute the matched filter error.
float e = y[i] - s; float e = y[i] - s;