Index: llvm/include/llvm/CodeGen/ValueTypes.td
===================================================================
--- llvm/include/llvm/CodeGen/ValueTypes.td
+++ llvm/include/llvm/CodeGen/ValueTypes.td
@@ -54,177 +54,178 @@
 
 def v1i8    : ValueType<8,    30>;  //    1 x i8 vector value
 def v2i8    : ValueType<16,   31>;  //    2 x i8 vector value
-def v4i8    : ValueType<32,   32>;  //    4 x i8 vector value
-def v8i8    : ValueType<64,   33>;  //    8 x i8 vector value
-def v16i8   : ValueType<128,  34>;  //   16 x i8 vector value
-def v32i8   : ValueType<256,  35>;  //   32 x i8 vector value
-def v64i8   : ValueType<512,  36>;  //   64 x i8 vector value
-def v128i8  : ValueType<1024, 37>;  //  128 x i8 vector value
-def v256i8  : ValueType<2048, 38>;  //  256 x i8 vector value
-def v512i8  : ValueType<4096, 39>;  //  512 x i8 vector value
-def v1024i8 : ValueType<8192, 40>;  // 1024 x i8 vector value
-
-def v1i16   : ValueType<16,   41>;  //   1 x i16 vector value
-def v2i16   : ValueType<32,   42>;  //   2 x i16 vector value
-def v3i16   : ValueType<48,   43>;  //   3 x i16 vector value
-def v4i16   : ValueType<64,   44>;  //   4 x i16 vector value
-def v8i16   : ValueType<128,  45>;  //   8 x i16 vector value
-def v16i16  : ValueType<256,  46>;  //  16 x i16 vector value
-def v32i16  : ValueType<512,  47>;  //  32 x i16 vector value
-def v64i16  : ValueType<1024, 48>;  //  64 x i16 vector value
-def v128i16 : ValueType<2048, 49>;  // 128 x i16 vector value
-def v256i16 : ValueType<4096, 50>;  // 256 x i16 vector value
-def v512i16 : ValueType<8192, 51>;  // 512 x i16 vector value
-
-def v1i32    : ValueType<32,    52>;  //    1 x i32 vector value
-def v2i32    : ValueType<64,    53>;  //    2 x i32 vector value
-def v3i32    : ValueType<96,    54>;  //    3 x i32 vector value
-def v4i32    : ValueType<128,   55>;  //    4 x i32 vector value
-def v5i32    : ValueType<160,   56>;  //    5 x i32 vector value
-def v6i32    : ValueType<192,   57>;  //    6 x f32 vector value
-def v7i32    : ValueType<224,   58>;  //    7 x f32 vector value
-def v8i32    : ValueType<256,   59>;  //    8 x i32 vector value
-def v16i32   : ValueType<512,   60>;  //   16 x i32 vector value
-def v32i32   : ValueType<1024,  61>;  //   32 x i32 vector value
-def v64i32   : ValueType<2048,  62>;  //   64 x i32 vector value
-def v128i32  : ValueType<4096,  63>;  //  128 x i32 vector value
-def v256i32  : ValueType<8192,  64>;  //  256 x i32 vector value
-def v512i32  : ValueType<16384, 65>;  //  512 x i32 vector value
-def v1024i32 : ValueType<32768, 66>;  // 1024 x i32 vector value
-def v2048i32 : ValueType<65536, 67>;  // 2048 x i32 vector value
-
-def v1i64   : ValueType<64,    68>;  //   1 x i64 vector value
-def v2i64   : ValueType<128,   69>;  //   2 x i64 vector value
-def v3i64   : ValueType<192,   70>;  //   3 x i64 vector value
-def v4i64   : ValueType<256,   71>;  //   4 x i64 vector value
-def v8i64   : ValueType<512,   72>;  //   8 x i64 vector value
-def v16i64  : ValueType<1024,  73>;  //  16 x i64 vector value
-def v32i64  : ValueType<2048,  74>;  //  32 x i64 vector value
-def v64i64  : ValueType<4096,  75>;  //  64 x i64 vector value
-def v128i64 : ValueType<8192,  76>;  // 128 x i64 vector value
-def v256i64 : ValueType<16384, 77>;  // 256 x i64 vector value
-
-def v1i128 : ValueType<128, 78>;  //  1 x i128 vector value
-
-def v1f16    : ValueType<16,     79>;  //    1 x f16 vector value
-def v2f16    : ValueType<32,     80>;  //    2 x f16 vector value
-def v3f16    : ValueType<48,     81>;  //    3 x f16 vector value
-def v4f16    : ValueType<64,     82>;  //    4 x f16 vector value
-def v8f16    : ValueType<128,    83>;  //    8 x f16 vector value
-def v16f16   : ValueType<256,    84>;  //   16 x f16 vector value
-def v32f16   : ValueType<512,    85>;  //   32 x f16 vector value
-def v64f16   : ValueType<1024,   86>;  //   64 x f16 vector value
-def v128f16  : ValueType<2048,   87>;  //  128 x f16 vector value
-def v256f16  : ValueType<4096,   88>;  //  256 x f16 vector value
-def v512f16  : ValueType<8192,   89>;  //  512 x f16 vector value
-
-def v2bf16   : ValueType<32,     90>;  //    2 x bf16 vector value
-def v3bf16   : ValueType<48,     91>;  //    3 x bf16 vector value
-def v4bf16   : ValueType<64,     92>;  //    4 x bf16 vector value
-def v8bf16   : ValueType<128,    93>;  //    8 x bf16 vector value
-def v16bf16  : ValueType<256,    94>;  //   16 x bf16 vector value
-def v32bf16  : ValueType<512,    95>;  //   32 x bf16 vector value
-def v64bf16  : ValueType<1024,   96>;  //   64 x bf16 vector value
-def v128bf16 : ValueType<2048,   97>;  //  128 x bf16 vector value
-
-def v1f32    : ValueType<32,     98>;  //    1 x f32 vector value
-def v2f32    : ValueType<64,     99>;  //    2 x f32 vector value
-def v3f32    : ValueType<96,    100>;  //    3 x f32 vector value
-def v4f32    : ValueType<128,   101>;  //    4 x f32 vector value
-def v5f32    : ValueType<160,   102>;  //    5 x f32 vector value
-def v6f32    : ValueType<192,   103>;  //    6 x f32 vector value
-def v7f32    : ValueType<224,   104>;  //    7 x f32 vector value
-def v8f32    : ValueType<256,   105>;  //    8 x f32 vector value
-def v16f32   : ValueType<512,   106>;  //   16 x f32 vector value
-def v32f32   : ValueType<1024,  107>;  //   32 x f32 vector value
-def v64f32   : ValueType<2048,  108>;  //   64 x f32 vector value
-def v128f32  : ValueType<4096,  109>;  //  128 x f32 vector value
-def v256f32  : ValueType<8192,  110>;  //  256 x f32 vector value
-def v512f32  : ValueType<16384, 111>;  //  512 x f32 vector value
-def v1024f32 : ValueType<32768, 112>;  // 1024 x f32 vector value
-def v2048f32 : ValueType<65536, 113>;  // 2048 x f32 vector value
-
-def v1f64    : ValueType<64,    114>;  //    1 x f64 vector value
-def v2f64    : ValueType<128,   115>;  //    2 x f64 vector value
-def v3f64    : ValueType<192,   116>;  //    3 x f64 vector value
-def v4f64    : ValueType<256,   117>;  //    4 x f64 vector value
-def v8f64    : ValueType<512,   118>;  //    8 x f64 vector value
-def v16f64   : ValueType<1024,  119>;  //   16 x f64 vector value
-def v32f64   : ValueType<2048,  120>;  //   32 x f64 vector value
-def v64f64   : ValueType<4096,  121>;  //   64 x f64 vector value
-def v128f64  : ValueType<8192,  122>;  //  128 x f64 vector value
-def v256f64  : ValueType<16384, 123>;  //  256 x f64 vector value
-
-def nxv1i1  : ValueType<1,  124>;  // n x  1 x i1  vector value
-def nxv2i1  : ValueType<2,  125>;  // n x  2 x i1  vector value
-def nxv4i1  : ValueType<4,  126>;  // n x  4 x i1  vector value
-def nxv8i1  : ValueType<8,  127>;  // n x  8 x i1  vector value
-def nxv16i1 : ValueType<16, 128>;  // n x 16 x i1  vector value
-def nxv32i1 : ValueType<32, 129>;  // n x 32 x i1  vector value
-def nxv64i1 : ValueType<64, 130>;  // n x 64 x i1  vector value
-
-def nxv1i8  : ValueType<8,   131>;  // n x  1 x i8  vector value
-def nxv2i8  : ValueType<16,  132>;  // n x  2 x i8  vector value
-def nxv4i8  : ValueType<32,  133>;  // n x  4 x i8  vector value
-def nxv8i8  : ValueType<64,  134>;  // n x  8 x i8  vector value
-def nxv16i8 : ValueType<128, 135>;  // n x 16 x i8  vector value
-def nxv32i8 : ValueType<256, 136>;  // n x 32 x i8  vector value
-def nxv64i8 : ValueType<512, 137>;  // n x 64 x i8  vector value
-
-def nxv1i16  : ValueType<16,  138>;  // n x  1 x i16 vector value
-def nxv2i16  : ValueType<32,  139>;  // n x  2 x i16 vector value
-def nxv4i16  : ValueType<64,  140>;  // n x  4 x i16 vector value
-def nxv8i16  : ValueType<128, 141>;  // n x  8 x i16 vector value
-def nxv16i16 : ValueType<256, 142>;  // n x 16 x i16 vector value
-def nxv32i16 : ValueType<512, 143>;  // n x 32 x i16 vector value
-
-def nxv1i32  : ValueType<32,   144>;  // n x  1 x i32 vector value
-def nxv2i32  : ValueType<64,   145>;  // n x  2 x i32 vector value
-def nxv4i32  : ValueType<128,  146>;  // n x  4 x i32 vector value
-def nxv8i32  : ValueType<256,  147>;  // n x  8 x i32 vector value
-def nxv16i32 : ValueType<512,  148>;  // n x 16 x i32 vector value
-def nxv32i32 : ValueType<1024, 149>;  // n x 32 x i32 vector value
-
-def nxv1i64  : ValueType<64,   150>;  // n x  1 x i64 vector value
-def nxv2i64  : ValueType<128,  151>;  // n x  2 x i64 vector value
-def nxv4i64  : ValueType<256,  152>;  // n x  4 x i64 vector value
-def nxv8i64  : ValueType<512,  153>;  // n x  8 x i64 vector value
-def nxv16i64 : ValueType<1024, 154>;  // n x 16 x i64 vector value
-def nxv32i64 : ValueType<2048, 155>;  // n x 32 x i64 vector value
-
-def nxv1f16  : ValueType<16,  156>;  // n x  1 x  f16 vector value
-def nxv2f16  : ValueType<32,  157>;  // n x  2 x  f16 vector value
-def nxv4f16  : ValueType<64,  158>;  // n x  4 x  f16 vector value
-def nxv8f16  : ValueType<128, 159>;  // n x  8 x  f16 vector value
-def nxv16f16 : ValueType<256, 160>;  // n x 16 x  f16 vector value
-def nxv32f16 : ValueType<512, 161>;  // n x 32 x  f16 vector value
-
-def nxv1bf16 : ValueType<16,  162>;  // n x  1 x bf16 vector value
-def nxv2bf16 : ValueType<32,  163>;  // n x  2 x bf16 vector value
-def nxv4bf16 : ValueType<64,  164>;  // n x  4 x bf16 vector value
-def nxv8bf16 : ValueType<128, 165>;  // n x  8 x bf16 vector value
-def nxv16bf16 : ValueType<256, 166>;  // n x 16 x bf16 vector value
-def nxv32bf16 : ValueType<512, 167>;  // n x 32 x bf16 vector value
-
-def nxv1f32  : ValueType<32,  168>;  // n x  1 x  f32 vector value
-def nxv2f32  : ValueType<64,  169>;  // n x  2 x  f32 vector value
-def nxv4f32  : ValueType<128, 170>;  // n x  4 x  f32 vector value
-def nxv8f32  : ValueType<256, 171>;  // n x  8 x  f32 vector value
-def nxv16f32 : ValueType<512, 172>;  // n x 16 x  f32 vector value
-
-def nxv1f64  : ValueType<64,  173>;  // n x  1 x  f64 vector value
-def nxv2f64  : ValueType<128, 174>;  // n x  2 x  f64 vector value
-def nxv4f64  : ValueType<256, 175>;  // n x  4 x  f64 vector value
-def nxv8f64  : ValueType<512, 176>;  // n x  8 x  f64 vector value
-
-def x86mmx    : ValueType<64,   177>;  // X86 MMX value
-def FlagVT    : ValueType<0,    178>;  // Pre-RA sched glue
-def isVoid    : ValueType<0,    179>;  // Produces no value
-def untyped   : ValueType<8,    180>;  // Produces an untyped value
-def funcref   : ValueType<0,    181>;  // WebAssembly's funcref type
-def externref : ValueType<0,    182>;  // WebAssembly's externref type
-def x86amx    : ValueType<8192, 183>;  // X86 AMX value
-def i64x8     : ValueType<512,  184>;  // 8 Consecutive GPRs (AArch64)
+def v3i8    : ValueType<24,   32>;  //    3 x i8 vector value
+def v4i8    : ValueType<32,   33>;  //    4 x i8 vector value
+def v8i8    : ValueType<64,   34>;  //    8 x i8 vector value
+def v16i8   : ValueType<128,  35>;  //   16 x i8 vector value
+def v32i8   : ValueType<256,  36>;  //   32 x i8 vector value
+def v64i8   : ValueType<512,  37>;  //   64 x i8 vector value
+def v128i8  : ValueType<1024, 38>;  //  128 x i8 vector value
+def v256i8  : ValueType<2048, 39>;  //  256 x i8 vector value
+def v512i8  : ValueType<4096, 40>;  //  512 x i8 vector value
+def v1024i8 : ValueType<8192, 41>;  // 1024 x i8 vector value
+
+def v1i16   : ValueType<16,   42>;  //   1 x i16 vector value
+def v2i16   : ValueType<32,   43>;  //   2 x i16 vector value
+def v3i16   : ValueType<48,   44>;  //   3 x i16 vector value
+def v4i16   : ValueType<64,   45>;  //   4 x i16 vector value
+def v8i16   : ValueType<128,  46>;  //   8 x i16 vector value
+def v16i16  : ValueType<256,  47>;  //  16 x i16 vector value
+def v32i16  : ValueType<512,  48>;  //  32 x i16 vector value
+def v64i16  : ValueType<1024, 49>;  //  64 x i16 vector value
+def v128i16 : ValueType<2048, 50>;  // 128 x i16 vector value
+def v256i16 : ValueType<4096, 51>;  // 256 x i16 vector value
+def v512i16 : ValueType<8192, 52>;  // 512 x i16 vector value
+
+def v1i32    : ValueType<32,    53>;  //    1 x i32 vector value
+def v2i32    : ValueType<64,    54>;  //    2 x i32 vector value
+def v3i32    : ValueType<96,    55>;  //    3 x i32 vector value
+def v4i32    : ValueType<128,   56>;  //    4 x i32 vector value
+def v5i32    : ValueType<160,   57>;  //    5 x i32 vector value
+def v6i32    : ValueType<192,   58>;  //    6 x f32 vector value
+def v7i32    : ValueType<224,   59>;  //    7 x f32 vector value
+def v8i32    : ValueType<256,   60>;  //    8 x i32 vector value
+def v16i32   : ValueType<512,   61>;  //   16 x i32 vector value
+def v32i32   : ValueType<1024,  62>;  //   32 x i32 vector value
+def v64i32   : ValueType<2048,  63>;  //   64 x i32 vector value
+def v128i32  : ValueType<4096,  64>;  //  128 x i32 vector value
+def v256i32  : ValueType<8192,  65>;  //  256 x i32 vector value
+def v512i32  : ValueType<16384, 66>;  //  512 x i32 vector value
+def v1024i32 : ValueType<32768, 67>;  // 1024 x i32 vector value
+def v2048i32 : ValueType<65536, 68>;  // 2048 x i32 vector value
+
+def v1i64   : ValueType<64,    69>;  //   1 x i64 vector value
+def v2i64   : ValueType<128,   70>;  //   2 x i64 vector value
+def v3i64   : ValueType<192,   71>;  //   3 x i64 vector value
+def v4i64   : ValueType<256,   72>;  //   4 x i64 vector value
+def v8i64   : ValueType<512,   73>;  //   8 x i64 vector value
+def v16i64  : ValueType<1024,  74>;  //  16 x i64 vector value
+def v32i64  : ValueType<2048,  75>;  //  32 x i64 vector value
+def v64i64  : ValueType<4096,  76>;  //  64 x i64 vector value
+def v128i64 : ValueType<8192,  77>;  // 128 x i64 vector value
+def v256i64 : ValueType<16384, 78>;  // 256 x i64 vector value
+
+def v1i128 : ValueType<128, 79>;  //  1 x i128 vector value
+
+def v1f16    : ValueType<16,     80>;  //    1 x f16 vector value
+def v2f16    : ValueType<32,     81>;  //    2 x f16 vector value
+def v3f16    : ValueType<48,     82>;  //    3 x f16 vector value
+def v4f16    : ValueType<64,     83>;  //    4 x f16 vector value
+def v8f16    : ValueType<128,    84>;  //    8 x f16 vector value
+def v16f16   : ValueType<256,    85>;  //   16 x f16 vector value
+def v32f16   : ValueType<512,    86>;  //   32 x f16 vector value
+def v64f16   : ValueType<1024,   87>;  //   64 x f16 vector value
+def v128f16  : ValueType<2048,   88>;  //  128 x f16 vector value
+def v256f16  : ValueType<4096,   89>;  //  256 x f16 vector value
+def v512f16  : ValueType<8192,   90>;  //  512 x f16 vector value
+
+def v2bf16   : ValueType<32,     91>;  //    2 x bf16 vector value
+def v3bf16   : ValueType<48,     92>;  //    3 x bf16 vector value
+def v4bf16   : ValueType<64,     93>;  //    4 x bf16 vector value
+def v8bf16   : ValueType<128,    94>;  //    8 x bf16 vector value
+def v16bf16  : ValueType<256,    95>;  //   16 x bf16 vector value
+def v32bf16  : ValueType<512,    96>;  //   32 x bf16 vector value
+def v64bf16  : ValueType<1024,   97>;  //   64 x bf16 vector value
+def v128bf16 : ValueType<2048,   98>;  //  128 x bf16 vector value
+
+def v1f32    : ValueType<32,     99>;  //    1 x f32 vector value
+def v2f32    : ValueType<64,    100>;  //    2 x f32 vector value
+def v3f32    : ValueType<96,    101>;  //    3 x f32 vector value
+def v4f32    : ValueType<128,   102>;  //    4 x f32 vector value
+def v5f32    : ValueType<160,   103>;  //    5 x f32 vector value
+def v6f32    : ValueType<192,   104>;  //    6 x f32 vector value
+def v7f32    : ValueType<224,   105>;  //    7 x f32 vector value
+def v8f32    : ValueType<256,   106>;  //    8 x f32 vector value
+def v16f32   : ValueType<512,   107>;  //   16 x f32 vector value
+def v32f32   : ValueType<1024,  108>;  //   32 x f32 vector value
+def v64f32   : ValueType<2048,  109>;  //   64 x f32 vector value
+def v128f32  : ValueType<4096,  110>;  //  128 x f32 vector value
+def v256f32  : ValueType<8192,  111>;  //  256 x f32 vector value
+def v512f32  : ValueType<16384, 112>;  //  512 x f32 vector value
+def v1024f32 : ValueType<32768, 113>;  // 1024 x f32 vector value
+def v2048f32 : ValueType<65536, 114>;  // 2048 x f32 vector value
+
+def v1f64    : ValueType<64,    115>;  //    1 x f64 vector value
+def v2f64    : ValueType<128,   116>;  //    2 x f64 vector value
+def v3f64    : ValueType<192,   117>;  //    3 x f64 vector value
+def v4f64    : ValueType<256,   118>;  //    4 x f64 vector value
+def v8f64    : ValueType<512,   119>;  //    8 x f64 vector value
+def v16f64   : ValueType<1024,  120>;  //   16 x f64 vector value
+def v32f64   : ValueType<2048,  121>;  //   32 x f64 vector value
+def v64f64   : ValueType<4096,  122>;  //   64 x f64 vector value
+def v128f64  : ValueType<8192,  123>;  //  128 x f64 vector value
+def v256f64  : ValueType<16384, 124>;  //  256 x f64 vector value
+
+def nxv1i1  : ValueType<1,  125>;  // n x  1 x i1  vector value
+def nxv2i1  : ValueType<2,  126>;  // n x  2 x i1  vector value
+def nxv4i1  : ValueType<4,  127>;  // n x  4 x i1  vector value
+def nxv8i1  : ValueType<8,  128>;  // n x  8 x i1  vector value
+def nxv16i1 : ValueType<16, 129>;  // n x 16 x i1  vector value
+def nxv32i1 : ValueType<32, 130>;  // n x 32 x i1  vector value
+def nxv64i1 : ValueType<64, 131>;  // n x 64 x i1  vector value
+
+def nxv1i8  : ValueType<8,   132>;  // n x  1 x i8  vector value
+def nxv2i8  : ValueType<16,  133>;  // n x  2 x i8  vector value
+def nxv4i8  : ValueType<32,  134>;  // n x  4 x i8  vector value
+def nxv8i8  : ValueType<64,  135>;  // n x  8 x i8  vector value
+def nxv16i8 : ValueType<128, 136>;  // n x 16 x i8  vector value
+def nxv32i8 : ValueType<256, 137>;  // n x 32 x i8  vector value
+def nxv64i8 : ValueType<512, 138>;  // n x 64 x i8  vector value
+
+def nxv1i16  : ValueType<16,  139>;  // n x  1 x i16 vector value
+def nxv2i16  : ValueType<32,  140>;  // n x  2 x i16 vector value
+def nxv4i16  : ValueType<64,  141>;  // n x  4 x i16 vector value
+def nxv8i16  : ValueType<128, 142>;  // n x  8 x i16 vector value
+def nxv16i16 : ValueType<256, 143>;  // n x 16 x i16 vector value
+def nxv32i16 : ValueType<512, 144>;  // n x 32 x i16 vector value
+
+def nxv1i32  : ValueType<32,   145>;  // n x  1 x i32 vector value
+def nxv2i32  : ValueType<64,   146>;  // n x  2 x i32 vector value
+def nxv4i32  : ValueType<128,  147>;  // n x  4 x i32 vector value
+def nxv8i32  : ValueType<256,  148>;  // n x  8 x i32 vector value
+def nxv16i32 : ValueType<512,  149>;  // n x 16 x i32 vector value
+def nxv32i32 : ValueType<1024, 150>;  // n x 32 x i32 vector value
+
+def nxv1i64  : ValueType<64,   151>;  // n x  1 x i64 vector value
+def nxv2i64  : ValueType<128,  152>;  // n x  2 x i64 vector value
+def nxv4i64  : ValueType<256,  153>;  // n x  4 x i64 vector value
+def nxv8i64  : ValueType<512,  154>;  // n x  8 x i64 vector value
+def nxv16i64 : ValueType<1024, 155>;  // n x 16 x i64 vector value
+def nxv32i64 : ValueType<2048, 156>;  // n x 32 x i64 vector value
+
+def nxv1f16  : ValueType<16,  157>;  // n x  1 x  f16 vector value
+def nxv2f16  : ValueType<32,  158>;  // n x  2 x  f16 vector value
+def nxv4f16  : ValueType<64,  159>;  // n x  4 x  f16 vector value
+def nxv8f16  : ValueType<128, 160>;  // n x  8 x  f16 vector value
+def nxv16f16 : ValueType<256, 161>;  // n x 16 x  f16 vector value
+def nxv32f16 : ValueType<512, 162>;  // n x 32 x  f16 vector value
+
+def nxv1bf16 : ValueType<16, 163>;  // n x  1 x bf16 vector value
+def nxv2bf16 : ValueType<32, 164>;  // n x  2 x bf16 vector value
+def nxv4bf16 : ValueType<64, 165>;  // n x  4 x bf16 vector value
+def nxv8bf16 : ValueType<128, 166>;  // n x  8 x bf16 vector value
+def nxv16bf16 : ValueType<256, 167>;  // n x 16 x bf16 vector value
+def nxv32bf16 : ValueType<512, 168>;  // n x 32 x bf16 vector value
+
+def nxv1f32  : ValueType<32,  169>;  // n x  1 x  f32 vector value
+def nxv2f32  : ValueType<64,  170>;  // n x  2 x  f32 vector value
+def nxv4f32  : ValueType<128, 171>;  // n x  4 x  f32 vector value
+def nxv8f32  : ValueType<256, 172>;  // n x  8 x  f32 vector value
+def nxv16f32 : ValueType<512, 173>;  // n x 16 x  f32 vector value
+
+def nxv1f64  : ValueType<64,  174>;  // n x  1 x  f64 vector value
+def nxv2f64  : ValueType<128, 175>;  // n x  2 x  f64 vector value
+def nxv4f64  : ValueType<256, 176>;  // n x  4 x  f64 vector value
+def nxv8f64  : ValueType<512, 177>;  // n x  8 x  f64 vector value
+
+def x86mmx    : ValueType<64,   178>;  // X86 MMX value
+def FlagVT    : ValueType<0,    179>;  // Pre-RA sched glue
+def isVoid    : ValueType<0,    180>;  // Produces no value
+def untyped   : ValueType<8,    181>;  // Produces an untyped value
+def funcref   : ValueType<0,    182>;  // WebAssembly's funcref type
+def externref : ValueType<0,    183>;  // WebAssembly's externref type
+def x86amx    : ValueType<8192, 184>;  // X86 AMX value
+def i64x8     : ValueType<512,  185>;  // 8 Consecutive GPRs (AArch64)
 
 def token      : ValueType<0, 248>;  // TokenTy
 def MetadataVT : ValueType<0, 249>;  // Metadata
Index: llvm/include/llvm/Support/MachineValueType.h
===================================================================
--- llvm/include/llvm/Support/MachineValueType.h
+++ llvm/include/llvm/Support/MachineValueType.h
@@ -81,109 +81,110 @@
 
       v1i8           =  30,   //    1 x i8
       v2i8           =  31,   //    2 x i8
-      v4i8           =  32,   //    4 x i8
-      v8i8           =  33,   //    8 x i8
-      v16i8          =  34,   //   16 x i8
-      v32i8          =  35,   //   32 x i8
-      v64i8          =  36,   //   64 x i8
-      v128i8         =  37,   //  128 x i8
-      v256i8         =  38,   //  256 x i8
-      v512i8         =  39,   //  512 x i8
-      v1024i8        =  40,   // 1024 x i8
-
-      v1i16          =  41,   //   1 x i16
-      v2i16          =  42,   //   2 x i16
-      v3i16          =  43,   //   3 x i16
-      v4i16          =  44,   //   4 x i16
-      v8i16          =  45,   //   8 x i16
-      v16i16         =  46,   //  16 x i16
-      v32i16         =  47,   //  32 x i16
-      v64i16         =  48,   //  64 x i16
-      v128i16        =  49,   // 128 x i16
-      v256i16        =  50,   // 256 x i16
-      v512i16        =  51,   // 512 x i16
-
-      v1i32          =  52,   //    1 x i32
-      v2i32          =  53,   //    2 x i32
-      v3i32          =  54,   //    3 x i32
-      v4i32          =  55,   //    4 x i32
-      v5i32          =  56,   //    5 x i32
-      v6i32          =  57,   //    6 x i32
-      v7i32          =  58,   //    7 x i32
-      v8i32          =  59,   //    8 x i32
-      v16i32         =  60,   //   16 x i32
-      v32i32         =  61,   //   32 x i32
-      v64i32         =  62,   //   64 x i32
-      v128i32        =  63,   //  128 x i32
-      v256i32        =  64,   //  256 x i32
-      v512i32        =  65,   //  512 x i32
-      v1024i32       =  66,   // 1024 x i32
-      v2048i32       =  67,   // 2048 x i32
-
-      v1i64          =  68,   //   1 x i64
-      v2i64          =  69,   //   2 x i64
-      v3i64          =  70,   //   3 x i64
-      v4i64          =  71,   //   4 x i64
-      v8i64          =  72,   //   8 x i64
-      v16i64         =  73,   //  16 x i64
-      v32i64         =  74,   //  32 x i64
-      v64i64         =  75,   //  64 x i64
-      v128i64        =  76,   // 128 x i64
-      v256i64        =  77,   // 256 x i64
-
-      v1i128         =  78,   //  1 x i128
+      v3i8           =  32,   //    3 x i8
+      v4i8           =  33,   //    4 x i8
+      v8i8           =  34,   //    8 x i8
+      v16i8          =  35,   //   16 x i8
+      v32i8          =  36,   //   32 x i8
+      v64i8          =  37,   //   64 x i8
+      v128i8         =  38,   //  128 x i8
+      v256i8         =  39,   //  256 x i8
+      v512i8         =  40,   //  512 x i8
+      v1024i8        =  41,   // 1024 x i8
+
+      v1i16          =  42,   //   1 x i16
+      v2i16          =  43,   //   2 x i16
+      v3i16          =  44,   //   3 x i16
+      v4i16          =  45,   //   4 x i16
+      v8i16          =  46,   //   8 x i16
+      v16i16         =  47,   //  16 x i16
+      v32i16         =  48,   //  32 x i16
+      v64i16         =  49,   //  64 x i16
+      v128i16        =  50,   // 128 x i16
+      v256i16        =  51,   // 256 x i16
+      v512i16        =  52,   // 512 x i16
+
+      v1i32          =  53,   //    1 x i32
+      v2i32          =  54,   //    2 x i32
+      v3i32          =  55,   //    3 x i32
+      v4i32          =  56,   //    4 x i32
+      v5i32          =  57,   //    5 x i32
+      v6i32          =  58,   //    6 x i32
+      v7i32          =  59,   //    7 x i32
+      v8i32          =  60,   //    8 x i32
+      v16i32         =  61,   //   16 x i32
+      v32i32         =  62,   //   32 x i32
+      v64i32         =  63,   //   64 x i32
+      v128i32        =  64,   //  128 x i32
+      v256i32        =  65,   //  256 x i32
+      v512i32        =  66,   //  512 x i32
+      v1024i32       =  67,   // 1024 x i32
+      v2048i32       =  68,   // 2048 x i32
+
+      v1i64          =  69,   //   1 x i64
+      v2i64          =  70,   //   2 x i64
+      v3i64          =  71,   //   3 x i64
+      v4i64          =  72,   //   4 x i64
+      v8i64          =  73,   //   8 x i64
+      v16i64         =  74,   //  16 x i64
+      v32i64         =  75,   //  32 x i64
+      v64i64         =  76,   //  64 x i64
+      v128i64        =  77,   // 128 x i64
+      v256i64        =  78,   // 256 x i64
+
+      v1i128         =  79,   //  1 x i128
 
       FIRST_INTEGER_FIXEDLEN_VECTOR_VALUETYPE = v1i1,
       LAST_INTEGER_FIXEDLEN_VECTOR_VALUETYPE = v1i128,
 
-      v1f16          =  79,   //    1 x f16
-      v2f16          =  80,   //    2 x f16
-      v3f16          =  81,   //    3 x f16
-      v4f16          =  82,   //    4 x f16
-      v8f16          =  83,   //    8 x f16
-      v16f16         =  84,   //   16 x f16
-      v32f16         =  85,   //   32 x f16
-      v64f16         =  86,   //   64 x f16
-      v128f16        =  87,   //  128 x f16
-      v256f16        =  88,   //  256 x f16
-      v512f16        =  89,   //  256 x f16
-
-      v2bf16         =  90,   //    2 x bf16
-      v3bf16         =  91,   //    3 x bf16
-      v4bf16         =  92,   //    4 x bf16
-      v8bf16         =  93,   //    8 x bf16
-      v16bf16        =  94,   //   16 x bf16
-      v32bf16        =  95,   //   32 x bf16
-      v64bf16        =  96,   //   64 x bf16
-      v128bf16       =  97,   //  128 x bf16
-
-      v1f32          =  98,   //    1 x f32
-      v2f32          =  99,   //    2 x f32
-      v3f32          = 100,   //    3 x f32
-      v4f32          = 101,   //    4 x f32
-      v5f32          = 102,   //    5 x f32
-      v6f32          = 103,   //    6 x f32
-      v7f32          = 104,   //    7 x f32
-      v8f32          = 105,   //    8 x f32
-      v16f32         = 106,   //   16 x f32
-      v32f32         = 107,   //   32 x f32
-      v64f32         = 108,   //   64 x f32
-      v128f32        = 109,   //  128 x f32
-      v256f32        = 110,   //  256 x f32
-      v512f32        = 111,   //  512 x f32
-      v1024f32       = 112,   // 1024 x f32
-      v2048f32       = 113,   // 2048 x f32
-
-      v1f64          = 114,   //    1 x f64
-      v2f64          = 115,   //    2 x f64
-      v3f64          = 116,   //    3 x f64
-      v4f64          = 117,   //    4 x f64
-      v8f64          = 118,   //    8 x f64
-      v16f64         = 119,   //   16 x f64
-      v32f64         = 120,   //   32 x f64
-      v64f64         = 121,   //   64 x f64
-      v128f64        = 122,   //  128 x f64
-      v256f64        = 123,   //  256 x f64
+      v1f16          =  80,   //    1 x f16
+      v2f16          =  81,   //    2 x f16
+      v3f16          =  82,   //    3 x f16
+      v4f16          =  83,   //    4 x f16
+      v8f16          =  84,   //    8 x f16
+      v16f16         =  85,   //   16 x f16
+      v32f16         =  86,   //   32 x f16
+      v64f16         =  87,   //   64 x f16
+      v128f16        =  88,   //  128 x f16
+      v256f16        =  89,   //  256 x f16
+      v512f16        =  90,   //  256 x f16
+
+      v2bf16         =  91,   //    2 x bf16
+      v3bf16         =  92,   //    3 x bf16
+      v4bf16         =  93,   //    4 x bf16
+      v8bf16         =  94,   //    8 x bf16
+      v16bf16        =  95,   //   16 x bf16
+      v32bf16        =  96,   //   32 x bf16
+      v64bf16        =  97,   //   64 x bf16
+      v128bf16       =  98,   //  128 x bf16
+
+      v1f32          =  99,   //    1 x f32
+      v2f32          = 100,   //    2 x f32
+      v3f32          = 101,   //    3 x f32
+      v4f32          = 102,   //    4 x f32
+      v5f32          = 103,   //    5 x f32
+      v6f32          = 104,   //    6 x f32
+      v7f32          = 105,   //    7 x f32
+      v8f32          = 106,   //    8 x f32
+      v16f32         = 107,   //   16 x f32
+      v32f32         = 108,   //   32 x f32
+      v64f32         = 109,   //   64 x f32
+      v128f32        = 110,   //  128 x f32
+      v256f32        = 111,   //  256 x f32
+      v512f32        = 112,   //  512 x f32
+      v1024f32       = 113,   // 1024 x f32
+      v2048f32       = 114,   // 2048 x f32
+
+      v1f64          = 115,   //    1 x f64
+      v2f64          = 116,   //    2 x f64
+      v3f64          = 117,   //    3 x f64
+      v4f64          = 118,   //    4 x f64
+      v8f64          = 119,   //    8 x f64
+      v16f64         = 120,   //   16 x f64
+      v32f64         = 121,   //   32 x f64
+      v64f64         = 122,   //   64 x f64
+      v128f64        = 123,   //  128 x f64
+      v256f64        = 124,   //  256 x f64
 
       FIRST_FP_FIXEDLEN_VECTOR_VALUETYPE = v1f16,
       LAST_FP_FIXEDLEN_VECTOR_VALUETYPE = v256f64,
@@ -191,70 +192,70 @@
       FIRST_FIXEDLEN_VECTOR_VALUETYPE = v1i1,
       LAST_FIXEDLEN_VECTOR_VALUETYPE = v256f64,
 
-      nxv1i1         = 124,   // n x  1 x i1
-      nxv2i1         = 125,   // n x  2 x i1
-      nxv4i1         = 126,   // n x  4 x i1
-      nxv8i1         = 127,   // n x  8 x i1
-      nxv16i1        = 128,   // n x 16 x i1
-      nxv32i1        = 129,   // n x 32 x i1
-      nxv64i1        = 130,   // n x 64 x i1
-
-      nxv1i8         = 131,   // n x  1 x i8
-      nxv2i8         = 132,   // n x  2 x i8
-      nxv4i8         = 133,   // n x  4 x i8
-      nxv8i8         = 134,   // n x  8 x i8
-      nxv16i8        = 135,   // n x 16 x i8
-      nxv32i8        = 136,   // n x 32 x i8
-      nxv64i8        = 137,   // n x 64 x i8
-
-      nxv1i16        = 138,  // n x  1 x i16
-      nxv2i16        = 139,  // n x  2 x i16
-      nxv4i16        = 140,  // n x  4 x i16
-      nxv8i16        = 141,  // n x  8 x i16
-      nxv16i16       = 142,  // n x 16 x i16
-      nxv32i16       = 143,  // n x 32 x i16
-
-      nxv1i32        = 144,  // n x  1 x i32
-      nxv2i32        = 145,  // n x  2 x i32
-      nxv4i32        = 146,  // n x  4 x i32
-      nxv8i32        = 147,  // n x  8 x i32
-      nxv16i32       = 148,  // n x 16 x i32
-      nxv32i32       = 149,  // n x 32 x i32
-
-      nxv1i64        = 150,  // n x  1 x i64
-      nxv2i64        = 151,  // n x  2 x i64
-      nxv4i64        = 152,  // n x  4 x i64
-      nxv8i64        = 153,  // n x  8 x i64
-      nxv16i64       = 154,  // n x 16 x i64
-      nxv32i64       = 155,  // n x 32 x i64
+      nxv1i1         = 125,   // n x  1 x i1
+      nxv2i1         = 126,   // n x  2 x i1
+      nxv4i1         = 127,   // n x  4 x i1
+      nxv8i1         = 128,   // n x  8 x i1
+      nxv16i1        = 129,   // n x 16 x i1
+      nxv32i1        = 130,   // n x 32 x i1
+      nxv64i1        = 131,   // n x 64 x i1
+
+      nxv1i8         = 132,   // n x  1 x i8
+      nxv2i8         = 133,   // n x  2 x i8
+      nxv4i8         = 134,   // n x  4 x i8
+      nxv8i8         = 135,   // n x  8 x i8
+      nxv16i8        = 136,   // n x 16 x i8
+      nxv32i8        = 137,   // n x 32 x i8
+      nxv64i8        = 138,   // n x 64 x i8
+
+      nxv1i16        = 139,  // n x  1 x i16
+      nxv2i16        = 140,  // n x  2 x i16
+      nxv4i16        = 141,  // n x  4 x i16
+      nxv8i16        = 142,  // n x  8 x i16
+      nxv16i16       = 143,  // n x 16 x i16
+      nxv32i16       = 144,  // n x 32 x i16
+
+      nxv1i32        = 145,  // n x  1 x i32
+      nxv2i32        = 146,  // n x  2 x i32
+      nxv4i32        = 147,  // n x  4 x i32
+      nxv8i32        = 148,  // n x  8 x i32
+      nxv16i32       = 149,  // n x 16 x i32
+      nxv32i32       = 150,  // n x 32 x i32
+
+      nxv1i64        = 151,  // n x  1 x i64
+      nxv2i64        = 152,  // n x  2 x i64
+      nxv4i64        = 153,  // n x  4 x i64
+      nxv8i64        = 154,  // n x  8 x i64
+      nxv16i64       = 155,  // n x 16 x i64
+      nxv32i64       = 156,  // n x 32 x i64
 
       FIRST_INTEGER_SCALABLE_VECTOR_VALUETYPE = nxv1i1,
       LAST_INTEGER_SCALABLE_VECTOR_VALUETYPE = nxv32i64,
 
-      nxv1f16        = 156,  // n x  1 x f16
-      nxv2f16        = 157,  // n x  2 x f16
-      nxv4f16        = 158,  // n x  4 x f16
-      nxv8f16        = 159,  // n x  8 x f16
-      nxv16f16       = 160,  // n x 16 x f16
-      nxv32f16       = 161,  // n x 32 x f16
-
-      nxv1bf16       = 162,  // n x  1 x bf16
-      nxv2bf16       = 163,  // n x  2 x bf16
-      nxv4bf16       = 164,  // n x  4 x bf16
-      nxv8bf16       = 165,  // n x  8 x bf16
-      nxv16bf16      = 166,  // n x 16 x bf16
-      nxv32bf16      = 167,  // n x 32 x bf16
-
-      nxv1f32        = 168,  // n x  1 x f32
-      nxv2f32        = 169,  // n x  2 x f32
-      nxv4f32        = 170,  // n x  4 x f32
-      nxv8f32        = 171,  // n x  8 x f32
-      nxv16f32       = 172,  // n x 16 x f32
-
-      nxv1f64        = 173,  // n x  1 x f64
-      nxv2f64        = 174,  // n x  2 x f64
-      nxv4f64        = 175,  // n x  4 x f64
-      nxv8f64        = 176,  // n x  8 x f64
+      nxv1f16        = 157,  // n x  1 x f16
+      nxv2f16        = 158,  // n x  2 x f16
+      nxv4f16        = 159,  // n x  4 x f16
+      nxv8f16        = 160,  // n x  8 x f16
+      nxv16f16       = 161,  // n x 16 x f16
+      nxv32f16       = 162,  // n x 32 x f16
+
+      nxv1bf16       = 163,  // n x  1 x bf16
+      nxv2bf16       = 164,  // n x  2 x bf16
+      nxv4bf16       = 165,  // n x  4 x bf16
+      nxv8bf16       = 166,  // n x  8 x bf16
+      nxv16bf16      = 167,  // n x 16 x bf16
+      nxv32bf16      = 168,  // n x 32 x bf16
+
+      nxv1f32        = 169,  // n x  1 x f32
+      nxv2f32        = 170,  // n x  2 x f32
+      nxv4f32        = 171,  // n x  4 x f32
+      nxv8f32        = 172,  // n x  8 x f32
+      nxv16f32       = 173,  // n x 16 x f32
+
+      nxv1f64        = 174,  // n x  1 x f64
+      nxv2f64        = 175,  // n x  2 x f64
+      nxv4f64        = 176,  // n x  4 x f64
+      nxv8f64        = 177,  // n x  8 x f64
 
       FIRST_FP_SCALABLE_VECTOR_VALUETYPE = nxv1f16,
       LAST_FP_SCALABLE_VECTOR_VALUETYPE = nxv8f64,
@@ -265,20 +266,20 @@
       FIRST_VECTOR_VALUETYPE = v1i1,
       LAST_VECTOR_VALUETYPE  = nxv8f64,
 
-      x86mmx         = 177,    // This is an X86 MMX value
+      x86mmx         = 178,    // This is an X86 MMX value
 
-      Glue           = 178,    // This glues nodes together during pre-RA sched
+      Glue           = 179,    // This glues nodes together during pre-RA sched
 
-      isVoid         = 179,    // This has no value
+      isVoid         = 180,    // This has no value
 
-      Untyped        = 180,    // This value takes a register, but has
+      Untyped        = 181,    // This value takes a register, but has
                                // unspecified type.  The register class
                                // will be determined by the opcode.
 
-      funcref        = 181,    // WebAssembly's funcref type
-      externref      = 182,    // WebAssembly's externref type
-      x86amx         = 183,    // This is an X86 AMX value
-      i64x8          = 184,    // 8 Consecutive GPRs (AArch64)
+      funcref        = 182,    // WebAssembly's funcref type
+      externref      = 183,    // WebAssembly's externref type
+      x86amx         = 184,    // This is an X86 AMX value
+      i64x8          = 185,    // 8 Consecutive GPRs (AArch64)
 
       FIRST_VALUETYPE =  1,    // This is always the beginning of the list.
       LAST_VALUETYPE = i64x8,  // This always remains at the end of the list.
@@ -552,6 +553,7 @@
       case v64i4: return i4;
       case v1i8:
       case v2i8:
+      case v3i8:
       case v4i8:
       case v8i8:
       case v16i8:
@@ -816,6 +818,7 @@
       case nxv4bf16:
       case nxv4f32:
       case nxv4f64: return 4;
+      case v3i8:
       case v3i16:
       case v3i32:
       case v3i64:
@@ -925,6 +928,7 @@
       case nxv1i16:
       case nxv1bf16:
       case nxv1f16: return TypeSize::Scalable(16);
+      case v3i8: return TypeSize::Scalable(24);
       case f32 :
       case i32 :
       case v32i1:
@@ -1240,6 +1244,7 @@
       case MVT::i8:
         if (NumElements == 1)   return MVT::v1i8;
         if (NumElements == 2)   return MVT::v2i8;
+        if (NumElements == 3)   return MVT::v3i8;
         if (NumElements == 4)   return MVT::v4i8;
         if (NumElements == 8)   return MVT::v8i8;
         if (NumElements == 16)  return MVT::v16i8;
Index: llvm/lib/CodeGen/ValueTypes.cpp
===================================================================
--- llvm/lib/CodeGen/ValueTypes.cpp
+++ llvm/lib/CodeGen/ValueTypes.cpp
@@ -240,6 +240,8 @@
     return FixedVectorType::get(Type::getInt8Ty(Context), 1);
   case MVT::v2i8:
     return FixedVectorType::get(Type::getInt8Ty(Context), 2);
+  case MVT::v3i8:
+    return FixedVectorType::get(Type::getInt8Ty(Context), 3);
   case MVT::v4i8:
     return FixedVectorType::get(Type::getInt8Ty(Context), 4);
   case MVT::v8i8:
Index: llvm/utils/TableGen/CodeGenTarget.cpp
===================================================================
--- llvm/utils/TableGen/CodeGenTarget.cpp
+++ llvm/utils/TableGen/CodeGenTarget.cpp
@@ -95,6 +95,7 @@
   case MVT::v64i4:    return "MVT::v64i4";
   case MVT::v1i8:     return "MVT::v1i8";
   case MVT::v2i8:     return "MVT::v2i8";
+  case MVT::v3i8:     return "MVT::v3i8";
   case MVT::v4i8:     return "MVT::v4i8";
   case MVT::v8i8:     return "MVT::v8i8";
   case MVT::v16i8:    return "MVT::v16i8";