Add signaling for a maximum DRED quantizer.

Since any value of dQ > 0 will cause the initial quantizer to degrade to the format-implied maximum (15) with a sufficient number of DRED frames, allow signaling a maximum smaller than 15. This allows encoders to improve the minimum quality of long DRED sequences (at the expense of bitrate) without requiring a constant quantizer for all frames (dQ == 0).

dnn/vec_avx.h

@@ -34,7 +34,7 @@
 
 #include <immintrin.h>
 #include <math.h>
-
+#include "celt/x86/x86cpu.h"
 
 #define MAX_INPUTS (2048)
 
@@ -786,19 +786,24 @@ static inline void sparse_cgemv8x4(float *_out, const opus_int8 *w, const int *i
       {
          __m256i vxj;
          __m256i vw;
-         vxj = _mm256_set1_epi32(*(int*)(void*)&x[*idx++]);
+         int pos;
+         pos = (*idx++);
+         vxj = _mm256_set1_epi32(OP_LOADU_EPI32(&x[pos]));
          vw = _mm256_loadu_si256((const __m256i *)(void*)w);
          vy0 = opus_mm256_dpbusds_epi32(vy0, vxj, vw);
          w += 32;
-         vxj = _mm256_set1_epi32(*(int*)(void*)&x[*idx++]);
+         pos = (*idx++);
+         vxj = _mm256_set1_epi32(OP_LOADU_EPI32(&x[pos]));
          vw = _mm256_loadu_si256((const __m256i *)(void*)w);
          vy0 = opus_mm256_dpbusds_epi32(vy0, vxj, vw);
          w += 32;
-         vxj = _mm256_set1_epi32(*(int*)(void*)&x[*idx++]);
+         pos = (*idx++);
+         vxj = _mm256_set1_epi32(OP_LOADU_EPI32(&x[pos]));
          vw = _mm256_loadu_si256((const __m256i *)(void*)w);
          vy0 = opus_mm256_dpbusds_epi32(vy0, vxj, vw);
          w += 32;
-         vxj = _mm256_set1_epi32(*(int*)(void*)&x[*idx++]);
+         pos = (*idx++);
+         vxj = _mm256_set1_epi32(OP_LOADU_EPI32(&x[pos]));
          vw = _mm256_loadu_si256((const __m256i *)(void*)w);
          vy0 = opus_mm256_dpbusds_epi32(vy0, vxj, vw);
          w += 32;
@@ -810,7 +815,7 @@ 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*)(void*)&x[pos]);
+         vxj = _mm256_set1_epi32(OP_LOADU_EPI32(&x[pos]));
          vw = _mm256_loadu_si256((const __m256i *)(void*)w);
          vy0 = opus_mm256_dpbusds_epi32(vy0, vxj, vw);
          w += 32;