diff options
Diffstat (limited to 'common_audio/signal_processing/downsample_fast_neon.c')
| -rw-r--r-- | common_audio/signal_processing/downsample_fast_neon.c | 33 |
1 files changed, 20 insertions, 13 deletions
diff --git a/common_audio/signal_processing/downsample_fast_neon.c b/common_audio/signal_processing/downsample_fast_neon.c index 36fc0c8aee..f1b754b798 100644 --- a/common_audio/signal_processing/downsample_fast_neon.c +++ b/common_audio/signal_processing/downsample_fast_neon.c @@ -8,9 +8,11 @@ * be found in the AUTHORS file in the root of the source tree. */ +#include <arm_neon.h> + #include "common_audio/signal_processing/include/signal_processing_library.h" -#include <arm_neon.h> +#include "rtc_base/checks.h" // NEON intrinsics version of WebRtcSpl_DownsampleFast() // for ARM 32-bit/64-bit platforms. @@ -22,19 +24,24 @@ int WebRtcSpl_DownsampleFastNeon(const int16_t* data_in, size_t coefficients_length, int factor, size_t delay) { - size_t i = 0; - size_t j = 0; + // Using signed indexes to be able to compute negative i-j that + // is used to index data_in. + int i = 0; + int j = 0; int32_t out_s32 = 0; - size_t endpos = delay + factor * (data_out_length - 1) + 1; + int endpos = delay + factor * (data_out_length - 1) + 1; size_t res = data_out_length & 0x7; - size_t endpos1 = endpos - factor * res; + int endpos1 = endpos - factor * res; // Return error if any of the running conditions doesn't meet. if (data_out_length == 0 || coefficients_length == 0 - || data_in_length < endpos) { + || (int)data_in_length < endpos) { return -1; } + RTC_DCHECK_GE(endpos, 0); + RTC_DCHECK_GE(endpos1, 0); + // First part, unroll the loop 8 times, with 3 subcases // (factor == 2, 4, others). switch (factor) { @@ -46,7 +53,7 @@ int WebRtcSpl_DownsampleFastNeon(const int16_t* data_in, #if defined(WEBRTC_ARCH_ARM64) // Unroll the loop 2 times. - for (j = 0; j < coefficients_length - 1; j += 2) { + for (j = 0; j < (int)coefficients_length - 1; j += 2) { int32x2_t coeff32 = vld1_dup_s32((int32_t*)&coefficients[j]); int16x4_t coeff16x4 = vreinterpret_s16_s32(coeff32); int16x8x2_t in16x8x2 = vld2q_s16(&data_in[i - j - 1]); @@ -68,7 +75,7 @@ int WebRtcSpl_DownsampleFastNeon(const int16_t* data_in, out32x4_1 = vmlal_lane_s16(out32x4_1, in16x4_3, coeff16x4, 0); } - for (; j < coefficients_length; j++) { + for (; j < (int)coefficients_length; j++) { int16x4_t coeff16x4 = vld1_dup_s16(&coefficients[j]); int16x8x2_t in16x8x2 = vld2q_s16(&data_in[i - j]); @@ -87,7 +94,7 @@ int WebRtcSpl_DownsampleFastNeon(const int16_t* data_in, #else // On ARMv7, the loop unrolling 2 times results in performance // regression. - for (j = 0; j < coefficients_length; j++) { + for (j = 0; j < (int)coefficients_length; j++) { int16x4_t coeff16x4 = vld1_dup_s16(&coefficients[j]); int16x8x2_t in16x8x2 = vld2q_s16(&data_in[i - j]); @@ -114,7 +121,7 @@ int WebRtcSpl_DownsampleFastNeon(const int16_t* data_in, int32x4_t out32x4_1 = vdupq_n_s32(2048); // Unroll the loop 4 times. - for (j = 0; j < coefficients_length - 3; j += 4) { + for (j = 0; j < (int)coefficients_length - 3; j += 4) { int16x4_t coeff16x4 = vld1_s16(&coefficients[j]); int16x8x4_t in16x8x4 = vld4q_s16(&data_in[i - j - 3]); @@ -143,7 +150,7 @@ int WebRtcSpl_DownsampleFastNeon(const int16_t* data_in, out32x4_1 = vmlal_lane_s16(out32x4_1, in16x4_7, coeff16x4, 0); } - for (; j < coefficients_length; j++) { + for (; j < (int)coefficients_length; j++) { int16x4_t coeff16x4 = vld1_dup_s16(&coefficients[j]); int16x8x4_t in16x8x4 = vld4q_s16(&data_in[i - j]); @@ -174,7 +181,7 @@ int WebRtcSpl_DownsampleFastNeon(const int16_t* data_in, int32x4_t out32x4_0 = vdupq_n_s32(2048); int32x4_t out32x4_1 = vdupq_n_s32(2048); - for (j = 0; j < coefficients_length; j++) { + for (j = 0; j < (int)coefficients_length; j++) { int16x4_t coeff16x4 = vld1_dup_s16(&coefficients[j]); int16x4_t in16x4_0 = vld1_dup_s16(&data_in[i - j]); in16x4_0 = vld1_lane_s16(&data_in[i + factor - j], in16x4_0, 1); @@ -204,7 +211,7 @@ int WebRtcSpl_DownsampleFastNeon(const int16_t* data_in, for (; i < endpos; i += factor) { out_s32 = 2048; // Round value, 0.5 in Q12. - for (j = 0; j < coefficients_length; j++) { + for (j = 0; j < (int)coefficients_length; j++) { out_s32 = WebRtc_MulAccumW16(coefficients[j], data_in[i - j], out_s32); } |