diff --git a/celt/x86/pitch_avx.c b/celt/x86/pitch_avx.c index 1a667dd746003192942488fe45c2e76c5ddaf93c..63dea97f1d3e633f8aa5a3c5844b2b20f5808708 100644 --- a/celt/x86/pitch_avx.c +++ b/celt/x86/pitch_avx.c @@ -56,7 +56,7 @@ static void xcorr_kernel_avx(const float *x, const float *y, float sum[8], int l if (i != len) { static const int mask[15] = {-1, -1, -1, -1, -1, -1, -1, 0, 0, 0, 0, 0, 0, 0, 0}; __m256i m; - m = _mm256_loadu_si256((__m256i*)(mask + 7+i-len)); + m = _mm256_loadu_si256((__m256i*)(void*)(mask + 7+i-len)); x0 = _mm256_maskload_ps(x+i, m); xsum0 = _mm256_fmadd_ps(x0, _mm256_maskload_ps(y+i , m), xsum0); xsum1 = _mm256_fmadd_ps(x0, _mm256_maskload_ps(y+i+1, m), xsum1); diff --git a/celt/x86/vq_sse2.c b/celt/x86/vq_sse2.c index 775042860d43c31cb76a9f4839d1008bcbe79943..4c4ebf8e2dde2991cf8ed776a1a6614acbfe07b0 100644 --- a/celt/x86/vq_sse2.c +++ b/celt/x86/vq_sse2.c @@ -75,7 +75,7 @@ opus_val16 op_pvq_search_sse2(celt_norm *_X, int *iy, int K, int N, int arch) sums = _mm_add_ps(sums, x4); /* Clear y and iy in case we don't do the projection. */ _mm_storeu_ps(&y[j], _mm_setzero_ps()); - _mm_storeu_si128((__m128i*)&iy[j], _mm_setzero_si128()); + _mm_storeu_si128((__m128i*)(void*)&iy[j], _mm_setzero_si128()); _mm_storeu_ps(&X[j], x4); _mm_storeu_ps(&signy[j], s4); } @@ -116,7 +116,7 @@ opus_val16 op_pvq_search_sse2(celt_norm *_X, int *iy, int K, int N, int arch) rx4 = _mm_mul_ps(x4, rcp4); iy4 = _mm_cvttps_epi32(rx4); pulses_sum = _mm_add_epi32(pulses_sum, iy4); - _mm_storeu_si128((__m128i*)&iy[j], iy4); + _mm_storeu_si128((__m128i*)(void*)&iy[j], iy4); y4 = _mm_cvtepi32_ps(iy4); xy4 = _mm_add_ps(xy4, _mm_mul_ps(x4, y4)); yy4 = _mm_add_ps(yy4, _mm_mul_ps(y4, y4)); @@ -205,10 +205,10 @@ opus_val16 op_pvq_search_sse2(celt_norm *_X, int *iy, int K, int N, int arch) { __m128i y4; __m128i s4; - y4 = _mm_loadu_si128((__m128i*)&iy[j]); + y4 = _mm_loadu_si128((__m128i*)(void*)&iy[j]); s4 = _mm_castps_si128(_mm_loadu_ps(&signy[j])); y4 = _mm_xor_si128(_mm_add_epi32(y4, s4), s4); - _mm_storeu_si128((__m128i*)&iy[j], y4); + _mm_storeu_si128((__m128i*)(void*)&iy[j], y4); } RESTORE_STACK; return yy; diff --git a/celt/x86/x86cpu.h b/celt/x86/x86cpu.h index 8bd69551b5a2fea77e0c6443387c68c565ca2a2c..fe46d1d948fbef1d1a94df0d152d3c69c4720ea1 100644 --- a/celt/x86/x86cpu.h +++ b/celt/x86/x86cpu.h @@ -68,6 +68,6 @@ int opus_select_arch(void); (_mm_cvtepi8_epi32(_mm_cvtsi32_si128(OP_LOADU_EPI32(x)))) #define OP_CVTEPI16_EPI32_M64(x) \ - (_mm_cvtepi16_epi32(_mm_loadl_epi64((__m128i *)(x)))) + (_mm_cvtepi16_epi32(_mm_loadl_epi64((__m128i *)(void*)(x)))) #endif diff --git a/dnn/vec_avx.h b/dnn/vec_avx.h index bc93313013b4ef57381e397aa4854cec3b5756d7..39002b110577ae3b442947db5a8e12d84ac516ee 100644 --- a/dnn/vec_avx.h +++ b/dnn/vec_avx.h @@ -335,7 +335,7 @@ static inline void vector_ps_to_epi8(unsigned char *x, const float *_x, int len) xi = _mm256_permute4x64_epi64(xi, 0xD8); xi = _mm256_packus_epi16(xi, _mm256_setzero_si256()); xi = _mm256_permutevar8x32_epi32(xi, _mm256_setr_epi32(0,1, 0,0, 0,0, 0,0)); - _mm256_storeu_si256 ((__m256i *)&x[i], xi); + _mm256_storeu_si256 ((__m256i *)(void*)&x[i], xi); } } @@ -794,20 +794,20 @@ static inline void sparse_cgemv8x4(float *_out, const opus_int8 *w, const int *i { __m256i vxj; __m256i vw; - vxj = _mm256_set1_epi32(*(int*)&x[*idx++]); - vw = _mm256_loadu_si256((const __m256i *)w); + vxj = _mm256_set1_epi32(*(int*)(void*)&x[*idx++]); + vw = _mm256_loadu_si256((const __m256i *)(void*)w); vy0 = opus_mm256_dpbusds_epi32(vy0, vxj, vw); w += 32; - vxj = _mm256_set1_epi32(*(int*)&x[*idx++]); - vw = _mm256_loadu_si256((const __m256i *)w); + vxj = _mm256_set1_epi32(*(int*)(void*)&x[*idx++]); + vw = _mm256_loadu_si256((const __m256i *)(void*)w); vy0 = opus_mm256_dpbusds_epi32(vy0, vxj, vw); w += 32; - vxj = _mm256_set1_epi32(*(int*)&x[*idx++]); - vw = _mm256_loadu_si256((const __m256i *)w); + vxj = _mm256_set1_epi32(*(int*)(void*)&x[*idx++]); + vw = _mm256_loadu_si256((const __m256i *)(void*)w); vy0 = opus_mm256_dpbusds_epi32(vy0, vxj, vw); w += 32; - vxj = _mm256_set1_epi32(*(int*)&x[*idx++]); - vw = _mm256_loadu_si256((const __m256i *)w); + vxj = _mm256_set1_epi32(*(int*)(void*)&x[*idx++]); + vw = _mm256_loadu_si256((const __m256i *)(void*)w); vy0 = opus_mm256_dpbusds_epi32(vy0, vxj, vw); w += 32; } @@ -818,8 +818,8 @@ static inline void sparse_cgemv8x4(float *_out, const opus_int8 *w, const int *i __m256i vw; int pos; pos = (*idx++); - vxj = _mm256_set1_epi32(*(int*)&x[pos]); - vw = _mm256_loadu_si256((const __m256i *)w); + vxj = _mm256_set1_epi32(*(int*)(void*)&x[pos]); + vw = _mm256_loadu_si256((const __m256i *)(void*)w); vy0 = opus_mm256_dpbusds_epi32(vy0, vxj, vw); w += 32; } @@ -845,20 +845,20 @@ static inline void cgemv8x4(float *_out, const opus_int8 *w, const float *scale, { __m256i vxj; __m256i vw; - vxj = _mm256_set1_epi32(*(int*)&x[j]); - vw = _mm256_loadu_si256((const __m256i *)w); + vxj = _mm256_set1_epi32(*(int*)(void*)&x[j]); + vw = _mm256_loadu_si256((const __m256i *)(void*)w); vy0 = opus_mm256_dpbusds_epi32(vy0, vxj, vw); w += 32; - vxj = _mm256_set1_epi32(*(int*)&x[j+4]); - vw = _mm256_loadu_si256((const __m256i *)w); + vxj = _mm256_set1_epi32(*(int*)(void*)&x[j+4]); + vw = _mm256_loadu_si256((const __m256i *)(void*)w); vy0 = opus_mm256_dpbusds_epi32(vy0, vxj, vw); w += 32; - vxj = _mm256_set1_epi32(*(int*)&x[j+8]); - vw = _mm256_loadu_si256((const __m256i *)w); + vxj = _mm256_set1_epi32(*(int*)(void*)&x[j+8]); + vw = _mm256_loadu_si256((const __m256i *)(void*)w); vy0 = opus_mm256_dpbusds_epi32(vy0, vxj, vw); w += 32; - vxj = _mm256_set1_epi32(*(int*)&x[j+12]); - vw = _mm256_loadu_si256((const __m256i *)w); + vxj = _mm256_set1_epi32(*(int*)(void*)&x[j+12]); + vw = _mm256_loadu_si256((const __m256i *)(void*)w); vy0 = opus_mm256_dpbusds_epi32(vy0, vxj, vw); w += 32; } @@ -867,8 +867,8 @@ static inline void cgemv8x4(float *_out, const opus_int8 *w, const float *scale, { __m256i vxj; __m256i vw; - vxj = _mm256_set1_epi32(*(int*)&x[j]); - vw = _mm256_loadu_si256((const __m256i *)w); + vxj = _mm256_set1_epi32(*(int*)(void*)&x[j]); + vw = _mm256_loadu_si256((const __m256i *)(void*)w); vy0 = opus_mm256_dpbusds_epi32(vy0, vxj, vw); w += 32; } diff --git a/silk/x86/NSQ_del_dec_sse4_1.c b/silk/x86/NSQ_del_dec_sse4_1.c index a58a76cd8d21c46b1371b13212c25dd1eed33a8d..c7d47e5ea8cd7613caec0eb0d3b0a1659c7386ad 100644 --- a/silk/x86/NSQ_del_dec_sse4_1.c +++ b/silk/x86/NSQ_del_dec_sse4_1.c @@ -428,7 +428,7 @@ static OPUS_INLINE void silk_noise_shape_quantizer_del_dec_sse4_1( LTP_pred_Q14 = 2; { __m128i tmpa, tmpb, pred_lag_ptr_tmp; - pred_lag_ptr_tmp = _mm_loadu_si128( (__m128i *)(&pred_lag_ptr[ -3 ] ) ); + pred_lag_ptr_tmp = _mm_loadu_si128( (__m128i *)(void*)(&pred_lag_ptr[ -3 ] ) ); pred_lag_ptr_tmp = _mm_shuffle_epi32( pred_lag_ptr_tmp, 0x1B ); tmpa = _mm_mul_epi32( pred_lag_ptr_tmp, b_Q12_0123 ); tmpa = _mm_srli_si128( tmpa, 2 ); @@ -483,7 +483,7 @@ static OPUS_INLINE void silk_noise_shape_quantizer_del_dec_sse4_1( tmpb = _mm_setzero_si128(); /* step 1 */ - psLPC_Q14_tmp = _mm_loadu_si128( (__m128i *)(&psLPC_Q14[ -3 ] ) ); /* -3, -2 , -1, 0 */ + psLPC_Q14_tmp = _mm_loadu_si128( (__m128i *)(void*)(&psLPC_Q14[ -3 ] ) ); /* -3, -2 , -1, 0 */ psLPC_Q14_tmp = _mm_shuffle_epi32( psLPC_Q14_tmp, 0x1B ); /* 0, -1, -2, -3 */ tmpa = _mm_mul_epi32( psLPC_Q14_tmp, a_Q12_0123 ); /* 0, -1, -2, -3 * 0123 -> 0*0, 2*-2 */ @@ -497,7 +497,7 @@ static OPUS_INLINE void silk_noise_shape_quantizer_del_dec_sse4_1( tmpb = _mm_add_epi32( tmpb, psLPC_Q14_tmp ); /* step 2 */ - psLPC_Q14_tmp = _mm_loadu_si128( (__m128i *)(&psLPC_Q14[ -7 ] ) ); + psLPC_Q14_tmp = _mm_loadu_si128( (__m128i *)(void*)(&psLPC_Q14[ -7 ] ) ); psLPC_Q14_tmp = _mm_shuffle_epi32( psLPC_Q14_tmp, 0x1B ); tmpa = _mm_mul_epi32( psLPC_Q14_tmp, a_Q12_4567 ); tmpa = _mm_srli_epi64( tmpa, 16 ); @@ -512,7 +512,7 @@ static OPUS_INLINE void silk_noise_shape_quantizer_del_dec_sse4_1( if ( opus_likely( predictLPCOrder == 16 ) ) { /* step 3 */ - psLPC_Q14_tmp = _mm_loadu_si128( (__m128i *)(&psLPC_Q14[ -11 ] ) ); + psLPC_Q14_tmp = _mm_loadu_si128( (__m128i *)(void*)(&psLPC_Q14[ -11 ] ) ); psLPC_Q14_tmp = _mm_shuffle_epi32( psLPC_Q14_tmp, 0x1B ); tmpa = _mm_mul_epi32( psLPC_Q14_tmp, a_Q12_89AB ); tmpa = _mm_srli_epi64( tmpa, 16 ); @@ -525,7 +525,7 @@ static OPUS_INLINE void silk_noise_shape_quantizer_del_dec_sse4_1( tmpb = _mm_add_epi32( tmpb, psLPC_Q14_tmp ); /* step 4 */ - psLPC_Q14_tmp = _mm_loadu_si128( (__m128i *)(&psLPC_Q14[ -15 ] ) ); + psLPC_Q14_tmp = _mm_loadu_si128( (__m128i *)(void*)(&psLPC_Q14[ -15 ] ) ); psLPC_Q14_tmp = _mm_shuffle_epi32( psLPC_Q14_tmp, 0x1B ); tmpa = _mm_mul_epi32( psLPC_Q14_tmp, a_Q12_CDEF ); tmpa = _mm_srli_epi64( tmpa, 16 ); @@ -830,7 +830,7 @@ static OPUS_INLINE void silk_nsq_del_dec_scale_states_sse4_1( xmm_x16_x2x0 = _mm_blend_epi16( xmm_x16_x2x0, xmm_x16_x3x1, 0xCC ); - _mm_storeu_si128( (__m128i *)(&(x_sc_Q10[ i ] ) ), xmm_x16_x2x0 ); + _mm_storeu_si128( (__m128i *)(void*)(&(x_sc_Q10[ i ] ) ), xmm_x16_x2x0 ); } for( ; i < psEncC->subfr_length; i++ ) { @@ -862,7 +862,7 @@ static OPUS_INLINE void silk_nsq_del_dec_scale_states_sse4_1( for( i = NSQ->sLTP_shp_buf_idx - psEncC->ltp_mem_length; i < NSQ->sLTP_shp_buf_idx - 3; i += 4 ) { - xmm_sLTP_shp_Q14_x2x0 = _mm_loadu_si128( (__m128i *)(&(NSQ->sLTP_shp_Q14[ i ] ) ) ); + xmm_sLTP_shp_Q14_x2x0 = _mm_loadu_si128( (__m128i *)(void*)(&(NSQ->sLTP_shp_Q14[ i ] ) ) ); /* equal shift right 4 bytes*/ xmm_sLTP_shp_Q14_x3x1 = _mm_shuffle_epi32( xmm_sLTP_shp_Q14_x2x0, _MM_SHUFFLE( 0, 3, 2, 1 ) ); @@ -874,7 +874,7 @@ static OPUS_INLINE void silk_nsq_del_dec_scale_states_sse4_1( xmm_sLTP_shp_Q14_x2x0 = _mm_blend_epi16( xmm_sLTP_shp_Q14_x2x0, xmm_sLTP_shp_Q14_x3x1, 0xCC ); - _mm_storeu_si128( (__m128i *)(&(NSQ->sLTP_shp_Q14[ i ] ) ), xmm_sLTP_shp_Q14_x2x0 ); + _mm_storeu_si128( (__m128i *)(void*)(&(NSQ->sLTP_shp_Q14[ i ] ) ), xmm_sLTP_shp_Q14_x2x0 ); } for( ; i < NSQ->sLTP_shp_buf_idx; i++ ) { diff --git a/silk/x86/NSQ_sse4_1.c b/silk/x86/NSQ_sse4_1.c index d5ae1d3b1cd56ba6629830f256fd77204a9b89f3..51f5957c02258124ef6c6aff37ebabd009d1379f 100644 --- a/silk/x86/NSQ_sse4_1.c +++ b/silk/x86/NSQ_sse4_1.c @@ -338,21 +338,21 @@ static OPUS_INLINE void silk_noise_shape_quantizer_10_16_sse4_1( xmm_one = _mm_set_epi8( 1, 0, 3, 2, 5, 4, 7, 6, 9, 8, 11, 10, 13, 12, 15, 14 ); /* load a_Q12[0] - a_Q12[7] */ - a_Q12_01234567 = _mm_loadu_si128( (__m128i *)(&a_Q12[ 0 ] ) ); + a_Q12_01234567 = _mm_loadu_si128( (__m128i *)(void*)(&a_Q12[ 0 ] ) ); /* load a_Q12[ 8 ] - a_Q12[ 15 ] */ - a_Q12_89ABCDEF = _mm_loadu_si128( (__m128i *)(&a_Q12[ 8 ] ) ); + a_Q12_89ABCDEF = _mm_loadu_si128( (__m128i *)(void*)(&a_Q12[ 8 ] ) ); a_Q12_01234567 = _mm_shuffle_epi8( a_Q12_01234567, xmm_one ); a_Q12_89ABCDEF = _mm_shuffle_epi8( a_Q12_89ABCDEF, xmm_one ); /* load AR_shp_Q13 */ - AR_shp_Q13_76543210 = _mm_loadu_si128( (__m128i *)(&AR_shp_Q13[0] ) ); + AR_shp_Q13_76543210 = _mm_loadu_si128( (__m128i *)(void*)(&AR_shp_Q13[0] ) ); /* load psLPC_Q14 */ xmm_one = _mm_set_epi8(15, 14, 11, 10, 7, 6, 3, 2, 13, 12, 9, 8, 5, 4, 1, 0 ); - xmm_tempa = _mm_loadu_si128( (__m128i *)(&psLPC_Q14[-16]) ); - xmm_tempb = _mm_loadu_si128( (__m128i *)(&psLPC_Q14[-12]) ); + xmm_tempa = _mm_loadu_si128( (__m128i *)(void*)(&psLPC_Q14[-16]) ); + xmm_tempb = _mm_loadu_si128( (__m128i *)(void*)(&psLPC_Q14[-12]) ); xmm_tempa = _mm_shuffle_epi8( xmm_tempa, xmm_one ); xmm_tempb = _mm_shuffle_epi8( xmm_tempb, xmm_one ); @@ -360,8 +360,8 @@ static OPUS_INLINE void silk_noise_shape_quantizer_10_16_sse4_1( psLPC_Q14_hi_89ABCDEF = _mm_unpackhi_epi64( xmm_tempa, xmm_tempb ); psLPC_Q14_lo_89ABCDEF = _mm_unpacklo_epi64( xmm_tempa, xmm_tempb ); - xmm_tempa = _mm_loadu_si128( (__m128i *)(&psLPC_Q14[ -8 ]) ); - xmm_tempb = _mm_loadu_si128( (__m128i *)(&psLPC_Q14[ -4 ]) ); + xmm_tempa = _mm_loadu_si128( (__m128i *)(void*)(&psLPC_Q14[ -8 ]) ); + xmm_tempb = _mm_loadu_si128( (__m128i *)(void*)(&psLPC_Q14[ -4 ]) ); xmm_tempa = _mm_shuffle_epi8( xmm_tempa, xmm_one ); xmm_tempb = _mm_shuffle_epi8( xmm_tempb, xmm_one ); @@ -370,8 +370,8 @@ static OPUS_INLINE void silk_noise_shape_quantizer_10_16_sse4_1( psLPC_Q14_lo_01234567 = _mm_unpacklo_epi64( xmm_tempa, xmm_tempb ); /* load sAR2_Q14 */ - xmm_tempa = _mm_loadu_si128( (__m128i *)(&(NSQ->sAR2_Q14[ 0 ]) ) ); - xmm_tempb = _mm_loadu_si128( (__m128i *)(&(NSQ->sAR2_Q14[ 4 ]) ) ); + xmm_tempa = _mm_loadu_si128( (__m128i *)(void*)(&(NSQ->sAR2_Q14[ 0 ]) ) ); + xmm_tempb = _mm_loadu_si128( (__m128i *)(void*)(&(NSQ->sAR2_Q14[ 4 ]) ) ); xmm_tempa = _mm_shuffle_epi8( xmm_tempa, xmm_one ); xmm_tempb = _mm_shuffle_epi8( xmm_tempb, xmm_one ); @@ -443,7 +443,7 @@ static OPUS_INLINE void silk_noise_shape_quantizer_10_16_sse4_1( b_Q14_0123 = _mm_shuffle_epi32( b_Q14_3210, 0x1B ); /* loaded: [0] [-1] [-2] [-3] */ - pred_lag_ptr_0123 = _mm_loadu_si128( (__m128i *)(&pred_lag_ptr[ -3 ] ) ); + pred_lag_ptr_0123 = _mm_loadu_si128( (__m128i *)(void*)(&pred_lag_ptr[ -3 ] ) ); /* shuffle to [-3] [-2] [-1] [0] and to new xmm */ xmm_tempa = _mm_shuffle_epi32( pred_lag_ptr_0123, 0x1B ); /*64-bit multiply, a[2] * b[-2], a[0] * b[0] */ @@ -595,8 +595,8 @@ static OPUS_INLINE void silk_noise_shape_quantizer_10_16_sse4_1( /* write back sAR2_Q14 */ xmm_tempa = _mm_unpackhi_epi16( sAR2_Q14_lo_76543210, sAR2_Q14_hi_76543210 ); xmm_tempb = _mm_unpacklo_epi16( sAR2_Q14_lo_76543210, sAR2_Q14_hi_76543210 ); - _mm_storeu_si128( (__m128i *)(&NSQ->sAR2_Q14[ 4 ]), xmm_tempa ); - _mm_storeu_si128( (__m128i *)(&NSQ->sAR2_Q14[ 0 ]), xmm_tempb ); + _mm_storeu_si128( (__m128i *)(void*)(&NSQ->sAR2_Q14[ 4 ]), xmm_tempa ); + _mm_storeu_si128( (__m128i *)(void*)(&NSQ->sAR2_Q14[ 0 ]), xmm_tempb ); /* xq[ i ] = (opus_int16)silk_SAT16( silk_RSHIFT_ROUND( silk_SMULWW( psLPC_Q14[ i ], Gain_Q10 ), 8 ) ); */ { @@ -612,8 +612,8 @@ static OPUS_INLINE void silk_noise_shape_quantizer_10_16_sse4_1( /* process xq */ for (i = 0; i < length - 7; i += 8) { - xmm_xq_Q14_3210 = _mm_loadu_si128( (__m128i *)(&(psLPC_Q14[ i + 0 ] ) ) ); - xmm_xq_Q14_7654 = _mm_loadu_si128( (__m128i *)(&(psLPC_Q14[ i + 4 ] ) ) ); + xmm_xq_Q14_3210 = _mm_loadu_si128( (__m128i *)(void*)(&(psLPC_Q14[ i + 0 ] ) ) ); + xmm_xq_Q14_7654 = _mm_loadu_si128( (__m128i *)(void*)(&(psLPC_Q14[ i + 4 ] ) ) ); /* equal shift right 4 bytes*/ xmm_xq_Q14_x3x1 = _mm_shuffle_epi32( xmm_xq_Q14_3210, _MM_SHUFFLE( 0, 3, 2, 1 ) ); @@ -644,7 +644,7 @@ static OPUS_INLINE void silk_noise_shape_quantizer_10_16_sse4_1( xmm_xq_Q14_3210 = _mm_packs_epi32( xmm_xq_Q14_3210, xmm_xq_Q14_7654 ); /* save to xq */ - _mm_storeu_si128( (__m128i *)(&xq[ i ] ), xmm_xq_Q14_3210 ); + _mm_storeu_si128( (__m128i *)(void*)(&xq[ i ] ), xmm_xq_Q14_3210 ); } } for ( ; i < length; i++) @@ -698,7 +698,7 @@ static OPUS_INLINE void silk_nsq_scale_states_sse4_1( xmm_x16_x2x0 = _mm_blend_epi16( xmm_x16_x2x0, xmm_x16_x3x1, 0xCC ); - _mm_storeu_si128( (__m128i *)(&(x_sc_Q10[ i ] ) ), xmm_x16_x2x0 ); + _mm_storeu_si128( (__m128i *)(void*)(&(x_sc_Q10[ i ] ) ), xmm_x16_x2x0 ); } for( ; i < psEncC->subfr_length; i++ ) { @@ -729,7 +729,7 @@ static OPUS_INLINE void silk_nsq_scale_states_sse4_1( for( i = NSQ->sLTP_shp_buf_idx - psEncC->ltp_mem_length; i < NSQ->sLTP_shp_buf_idx - 3; i += 4 ) { - xmm_sLTP_shp_Q14_x2x0 = _mm_loadu_si128( (__m128i *)(&(NSQ->sLTP_shp_Q14[ i ] ) ) ); + xmm_sLTP_shp_Q14_x2x0 = _mm_loadu_si128( (__m128i *)(void*)(&(NSQ->sLTP_shp_Q14[ i ] ) ) ); /* equal shift right 4 bytes*/ xmm_sLTP_shp_Q14_x3x1 = _mm_shuffle_epi32( xmm_sLTP_shp_Q14_x2x0, _MM_SHUFFLE( 0, 3, 2, 1 ) ); @@ -741,7 +741,7 @@ static OPUS_INLINE void silk_nsq_scale_states_sse4_1( xmm_sLTP_shp_Q14_x2x0 = _mm_blend_epi16( xmm_sLTP_shp_Q14_x2x0, xmm_sLTP_shp_Q14_x3x1, 0xCC ); - _mm_storeu_si128( (__m128i *)(&(NSQ->sLTP_shp_Q14[ i ] ) ), xmm_sLTP_shp_Q14_x2x0 ); + _mm_storeu_si128( (__m128i *)(void*)(&(NSQ->sLTP_shp_Q14[ i ] ) ), xmm_sLTP_shp_Q14_x2x0 ); } for( ; i < NSQ->sLTP_shp_buf_idx; i++ ) { diff --git a/silk/x86/VAD_sse4_1.c b/silk/x86/VAD_sse4_1.c index e7eaf9714a74204c660648351cccb5eb7460d0c6..9e06bc79d050c3e64c2a2c1bb4abc6956ed118f1 100644 --- a/silk/x86/VAD_sse4_1.c +++ b/silk/x86/VAD_sse4_1.c @@ -144,7 +144,7 @@ opus_int silk_VAD_GetSA_Q8_sse4_1( /* O Return value, 0 if s for( i = 0; i < dec_subframe_length - 7; i += 8 ) { - xmm_X = _mm_loadu_si128( (__m128i *)&(X[ X_offset[ b ] + i + dec_subframe_offset ] ) ); + xmm_X = _mm_loadu_si128( (__m128i *)(void*)&(X[ X_offset[ b ] + i + dec_subframe_offset ] ) ); xmm_X = _mm_srai_epi16( xmm_X, 3 ); xmm_X = _mm_madd_epi16( xmm_X, xmm_X ); xmm_acc = _mm_add_epi32( xmm_acc, xmm_X ); diff --git a/silk/x86/VQ_WMat_EC_sse4_1.c b/silk/x86/VQ_WMat_EC_sse4_1.c index 2c7d18d05eab4a003b90771d8f23bc1ddda0eb21..df4626b60a973f4530155511daa68aad808a6f2a 100644 --- a/silk/x86/VQ_WMat_EC_sse4_1.c +++ b/silk/x86/VQ_WMat_EC_sse4_1.c @@ -65,7 +65,7 @@ void silk_VQ_WMat_EC_sse4_1( neg_xX_Q24[ 3 ] = -silk_LSHIFT32( xX_Q17[ 3 ], 7 ); neg_xX_Q24[ 4 ] = -silk_LSHIFT32( xX_Q17[ 4 ], 7 ); - v_XX_31_Q17 = _mm_loadu_si128( (__m128i *)(&XX_Q17[ 1 ] ) ); + v_XX_31_Q17 = _mm_loadu_si128( (__m128i *)(void*)(&XX_Q17[ 1 ] ) ); v_XX_42_Q17 = _mm_shuffle_epi32( v_XX_31_Q17, _MM_SHUFFLE( 0, 3, 2, 1 ) ); /* Loop over codebook */