1048   for (size_t cr_block_start = 0; cr_block_start < c; cr_block_start += cr) {  in xnn_pack_f32_dwconv_ghw_w()  local
1049     const size_t cr_block_size = min(c - cr_block_start, cr);  in xnn_pack_f32_dwconv_ghw_w()
1052         *packed_w++ = b[cr_block_start + cr_block_offset];  in xnn_pack_f32_dwconv_ghw_w()
1064           const float kv = k[((cr_block_start + cr_block_offset) * h + y) * w + x];  in xnn_pack_f32_dwconv_ghw_w()
1083   for (size_t cr_block_start = 0; cr_block_start < c; cr_block_start += cr) {  in xnn_pack_f16_dwconv_ghw_w()  local
1084     const size_t cr_block_size = min(c - cr_block_start, cr);  in xnn_pack_f16_dwconv_ghw_w()
1087         *packed_w++ = b[cr_block_start + cr_block_offset];  in xnn_pack_f16_dwconv_ghw_w()
1099           const uint16_t kv = k[((cr_block_start + cr_block_offset) * h + y) * w + x];  in xnn_pack_f16_dwconv_ghw_w()
1120   for (size_t cr_block_start = 0; cr_block_start < c; cr_block_start += cr) {  in xnn_pack_qu8_dwconv_ghw_w()  local
1121     const size_t cr_block_size = min(c - cr_block_start, cr);  in xnn_pack_qu8_dwconv_ghw_w()
1125         *((int32_t*) packed_w) = b[cr_block_start + cr_block_offset] + boff;  in xnn_pack_qu8_dwconv_ghw_w()
1139           const uint8_t kv = k[((cr_block_start + cr_block_offset) * h + y) * w + x];  in xnn_pack_qu8_dwconv_ghw_w()
1161   for (size_t cr_block_start = 0; cr_block_start < c; cr_block_start += cr) {  in xnn_pack_qs8_dwconv_ghw_w()  local
1162     const size_t cr_block_size = min(c - cr_block_start, cr);  in xnn_pack_qs8_dwconv_ghw_w()
1166         *((int32_t*) packed_w) = b[cr_block_start + cr_block_offset];  in xnn_pack_qs8_dwconv_ghw_w()
1180           const int8_t kv = k[((cr_block_start + cr_block_offset) * h + y) * w + x];  in xnn_pack_qs8_dwconv_ghw_w()
1201   for (size_t cr_block_start = 0; cr_block_start < c; cr_block_start += cr) {  in xnn_pack_f32_dwconv_hwg_w()  local
1202     const size_t cr_block_size = min(c - cr_block_start, cr);  in xnn_pack_f32_dwconv_hwg_w()
1205         *packed_w++ = b[cr_block_start + cr_block_offset];  in xnn_pack_f32_dwconv_hwg_w()
1217           const float kv = k[(y * w + x) * c + (cr_block_start + cr_block_offset)];  in xnn_pack_f32_dwconv_hwg_w()
1236   for (size_t cr_block_start = 0; cr_block_start < c; cr_block_start += cr) {  in xnn_pack_f16_dwconv_hwg_w()  local
1237     const size_t cr_block_size = min(c - cr_block_start, cr);  in xnn_pack_f16_dwconv_hwg_w()
1240         *packed_w++ = b[cr_block_start + cr_block_offset];  in xnn_pack_f16_dwconv_hwg_w()
1252           const uint16_t kv = k[(y * w + x) * c + (cr_block_start + cr_block_offset)];  in xnn_pack_f16_dwconv_hwg_w()
1273   for (size_t cr_block_start = 0; cr_block_start < c; cr_block_start += cr) {  in xnn_pack_qu8_dwconv_hwg_w()  local
1274     const size_t cr_block_size = min(c - cr_block_start, cr);  in xnn_pack_qu8_dwconv_hwg_w()
1278         *((int32_t*) packed_w) = b[cr_block_start + cr_block_offset] + boff;  in xnn_pack_qu8_dwconv_hwg_w()
1292           const uint8_t kv = k[(y * w + x) * c + (cr_block_start + cr_block_offset)];  in xnn_pack_qu8_dwconv_hwg_w()
1314   for (size_t cr_block_start = 0; cr_block_start < c; cr_block_start += cr) {  in xnn_pack_qs8_dwconv_hwg_w()  local
1315     const size_t cr_block_size = min(c - cr_block_start, cr);  in xnn_pack_qs8_dwconv_hwg_w()
1319         *((int32_t*) packed_w) = b[cr_block_start + cr_block_offset];  in xnn_pack_qs8_dwconv_hwg_w()
1333           const int8_t kv = k[(y * w + x) * c + (cr_block_start + cr_block_offset)];  in xnn_pack_qs8_dwconv_hwg_w()
1581   for (size_t cr_block_start = 0; cr_block_start < c; cr_block_start += cr) {  in xnn_pack_f32_vmulcaddc_w()  local
1582     const size_t cr_block_size = min(c - cr_block_start, cr);  in xnn_pack_f32_vmulcaddc_w()
1584       *packed_w++ = s[cr_block_start + cr_block_offset];  in xnn_pack_f32_vmulcaddc_w()
1589         *packed_w++ = b[cr_block_start + cr_block_offset];  in xnn_pack_f32_vmulcaddc_w()
1609   for (size_t cr_block_start = 0; cr_block_start < c; cr_block_start += cr) {  in xnn_pack_f16_vmulcaddc_w()  local
1610     const size_t cr_block_size = min(c - cr_block_start, cr);  in xnn_pack_f16_vmulcaddc_w()
1612       *packed_w++ = s[cr_block_start + cr_block_offset];  in xnn_pack_f16_vmulcaddc_w()
1617         *packed_w++ = b[cr_block_start + cr_block_offset];  in xnn_pack_f16_vmulcaddc_w()