[ValueTypes][RISCV] Add MVT for v1f16.

[craig.topper]

Summary

RISCV makes all fixed vector MVTs with size less than or equal
to a command line option legal.

This didn't include v1f16 because it was missing but did include v1f32 and v1f64.

One test is affectd where we did test this type, but it is a horizontal
reduction so it is non-sensical. Perhaps we should canonicalize that
away somewhere.

I'm not sure if we should be making v1 types legal, but this will at
least make RISCV consistent across all types.

[frasercrmck]

I think consistency is worthwhile, though I too question the sensibility of maintaining a v1 reduction.

Nit, that there's a typo in your commit message: affectd.

LGTM otherwise.

llvm/include/llvm/CodeGen/ValueTypes.td

 def v16i64 : ValueType<1024,62>;   // 16 x i64 vector value
 def v32i64 : ValueType<2048,63>;   // 32 x i64 vector value
 def v64i64 : ValueType<4096,64>;   // 64 x i64 vector value
 def v128i64: ValueType<8192,65>;   // 128 x i64 vector value
 def v256i64: ValueType<16384,66>;  // 256 x i64 vector value
 
 def v1i128 : ValueType<128, 67>;   //  1 x i128 vector value
 
-def v2f16    : ValueType<32 , 68>;    //    2 x f16 vector value
-def v3f16    : ValueType<48 , 69>;    //    3 x f16 vector value
-def v4f16    : ValueType<64 , 70>;    //    4 x f16 vector value
-def v8f16    : ValueType<128, 71>;    //    8 x f16 vector value
-def v16f16   : ValueType<256, 72>;    //   16 x f16 vector value
-def v32f16   : ValueType<512, 73>;    //   32 x f16 vector value
-def v64f16   : ValueType<1024, 74>;   //   64 x f16 vector value
-def v128f16  : ValueType<2048, 75>;   //  128 x f16 vector value
-def v2bf16   : ValueType<32 , 76>;    //    2 x bf16 vector value
-def v3bf16   : ValueType<48 , 77>;    //    3 x bf16 vector value
-def v4bf16   : ValueType<64 , 78>;    //    4 x bf16 vector value
-def v8bf16   : ValueType<128, 79>;    //    8 x bf16 vector value
-def v16bf16  : ValueType<256, 80>;    //   16 x bf16 vector value
-def v32bf16  : ValueType<512, 81>;    //   32 x bf16 vector value
-def v64bf16  : ValueType<1024, 82>;   //   64 x bf16 vector value
-def v128bf16 : ValueType<2048, 83>;   //  128 x bf16 vector value
-def v1f32    : ValueType<32 , 84>;    //    1 x f32 vector value
-def v2f32    : ValueType<64 , 85>;    //    2 x f32 vector value
-def v3f32    : ValueType<96 , 86>;    //    3 x f32 vector value
-def v4f32    : ValueType<128, 87>;    //    4 x f32 vector value
-def v5f32    : ValueType<160, 88>;    //    5 x f32 vector value
-def v8f32    : ValueType<256, 89>;    //    8 x f32 vector value
-def v16f32   : ValueType<512,  90>;   //   16 x f32 vector value
-def v32f32   : ValueType<1024, 91>;   //   32 x f32 vector value
-def v64f32   : ValueType<2048, 92>;   //   64 x f32 vector value
-def v128f32  : ValueType<4096, 93>;   //  128 x f32 vector value
-def v256f32  : ValueType<8182, 94>;   //  256 x f32 vector value
-def v512f32  : ValueType<16384, 95>;  //  512 x f32 vector value
-def v1024f32 : ValueType<32768, 96>;  // 1024 x f32 vector value
-def v2048f32 : ValueType<65536, 97>;  // 2048 x f32 vector value
-def v1f64    : ValueType<64, 98>;     //    1 x f64 vector value
-def v2f64    : ValueType<128, 99>;    //    2 x f64 vector value
-def v4f64    : ValueType<256, 100>;    //    4 x f64 vector value
-def v8f64    : ValueType<512, 101>;    //    8 x f64 vector value
-def v16f64   : ValueType<1024, 102>;   //   16 x f64 vector value
-def v32f64   : ValueType<2048, 103>;  //   32 x f64 vector value
-def v64f64   : ValueType<4096, 104>;  //   64 x f64 vector value
-def v128f64  : ValueType<8192, 105>;  //  128 x f64 vector value
-def v256f64  : ValueType<16384, 106>; //  256 x f64 vector value
-
-def nxv1i1  : ValueType<1,  107>;  // n x  1 x i1  vector value
-def nxv2i1  : ValueType<2,  108>;  // n x  2 x i1  vector value
-def nxv4i1  : ValueType<4,  109>;  // n x  4 x i1  vector value
-def nxv8i1  : ValueType<8,  110>;  // n x  8 x i1  vector value
-def nxv16i1 : ValueType<16, 111>;  // n x 16 x i1  vector value
-def nxv32i1 : ValueType<32, 112>;  // n x 32 x i1  vector value
-def nxv64i1  : ValueType<64,113>;  // n x  64 x i1  vector value
-
-def nxv1i8  : ValueType<8,   114>;  // n x  1 x i8  vector value
-def nxv2i8  : ValueType<16,  115>;  // n x  2 x i8  vector value
-def nxv4i8  : ValueType<32,  116>;  // n x  4 x i8  vector value
-def nxv8i8  : ValueType<64,  117>;  // n x  8 x i8  vector value
-def nxv16i8 : ValueType<128, 118>;  // n x 16 x i8  vector value
-def nxv32i8 : ValueType<256, 119>;  // n x 32 x i8  vector value
-def nxv64i8  : ValueType<512,  120>;  // n x  64 x i8  vector value
-
-def nxv1i16 : ValueType<16,  121>; // n x  1 x i16 vector value
-def nxv2i16 : ValueType<32,  122>; // n x  2 x i16 vector value
-def nxv4i16 : ValueType<64,  123>; // n x  4 x i16 vector value
-def nxv8i16 : ValueType<128, 124>; // n x  8 x i16 vector value
-def nxv16i16: ValueType<256, 125>; // n x 16 x i16 vector value
-def nxv32i16: ValueType<512, 126>; // n x 32 x i16 vector value
-
-def nxv1i32 : ValueType<32,  127>; // n x  1 x i32 vector value
-def nxv2i32 : ValueType<64,  128>; // n x  2 x i32 vector value
-def nxv4i32 : ValueType<128, 129>; // n x  4 x i32 vector value
-def nxv8i32 : ValueType<256, 130>; // n x  8 x i32 vector value
-def nxv16i32: ValueType<512, 131>; // n x 16 x i32 vector value
-def nxv32i32: ValueType<1024,132>; // n x 32 x i32 vector value
-
-def nxv1i64 : ValueType<64,  133>; // n x  1 x i64 vector value
-def nxv2i64 : ValueType<128, 134>; // n x  2 x i64 vector value
-def nxv4i64 : ValueType<256, 135>; // n x  4 x i64 vector value
-def nxv8i64 : ValueType<512, 136>; // n x  8 x i64 vector value
-def nxv16i64: ValueType<1024,137>; // n x 16 x i64 vector value
-def nxv32i64: ValueType<2048,138>; // n x 32 x i64 vector value
-
-def nxv1f16  : ValueType<16, 139>; // n x  1 x  f16 vector value
-def nxv2f16  : ValueType<32, 140>; // n x  2 x  f16 vector value
-def nxv4f16  : ValueType<64, 141>; // n x  4 x  f16 vector value
-def nxv8f16  : ValueType<128,142>; // n x  8 x  f16 vector value
-def nxv16f16 : ValueType<256,143>; // n x 16 x  f16 vector value
-def nxv32f16 : ValueType<512,144>; // n x 32 x  f16 vector value
-def nxv1bf16 : ValueType<16, 145>; // n x  1 x bf16 vector value
-def nxv2bf16 : ValueType<32, 146>; // n x  2 x bf16 vector value
-def nxv4bf16 : ValueType<64, 147>; // n x  4 x bf16 vector value
-def nxv8bf16 : ValueType<128,148>; // n x  8 x bf16 vector value
-def nxv1f32  : ValueType<32, 149>; // n x  1 x  f32 vector value
-def nxv2f32  : ValueType<64, 150>; // n x  2 x  f32 vector value
-def nxv4f32  : ValueType<128,151>; // n x  4 x  f32 vector value
-def nxv8f32  : ValueType<256,152>; // n x  8 x  f32 vector value
-def nxv16f32 : ValueType<512,153>; // n x 16 x  f32 vector value
-def nxv1f64  : ValueType<64, 154>; // n x  1 x  f64 vector value
-def nxv2f64  : ValueType<128,155>; // n x  2 x  f64 vector value
-def nxv4f64  : ValueType<256,156>; // n x  4 x  f64 vector value
-def nxv8f64  : ValueType<512,157>; // n x  8 x  f64 vector value
-
-def x86mmx : ValueType<64,   158>; // X86 MMX value
-def FlagVT : ValueType<0,    159>; // Pre-RA sched glue
-def isVoid : ValueType<0,    160>; // Produces no value
-def untyped: ValueType<8,    161>; // Produces an untyped value
-def funcref : ValueType<0,   162>; // WebAssembly's funcref type
-def externref : ValueType<0, 163>; // WebAssembly's externref type
-def x86amx : ValueType<8192, 164>; // X86 AMX value
+def v1f16    : ValueType<16 , 68>;    //    1 x f16 vector value
+def v2f16    : ValueType<32 , 69>;    //    2 x f16 vector value
+def v3f16    : ValueType<48 , 70>;    //    3 x f16 vector value
+def v4f16    : ValueType<64 , 71>;    //    4 x f16 vector value
+def v8f16    : ValueType<128, 72>;    //    8 x f16 vector value
+def v16f16   : ValueType<256, 73>;    //   16 x f16 vector value
+def v32f16   : ValueType<512, 74>;    //   32 x f16 vector value
+def v64f16   : ValueType<1024, 75>;   //   64 x f16 vector value
+def v128f16  : ValueType<2048, 76>;   //  128 x f16 vector value
+def v2bf16   : ValueType<32 , 77>;    //    2 x bf16 vector value
+def v3bf16   : ValueType<48 , 78>;    //    3 x bf16 vector value
+def v4bf16   : ValueType<64 , 79>;    //    4 x bf16 vector value
+def v8bf16   : ValueType<128, 80>;    //    8 x bf16 vector value
+def v16bf16  : ValueType<256, 81>;    //   16 x bf16 vector value
+def v32bf16  : ValueType<512, 82>;    //   32 x bf16 vector value
+def v64bf16  : ValueType<1024, 83>;   //   64 x bf16 vector value
+def v128bf16 : ValueType<2048, 84>;   //  128 x bf16 vector value
+def v1f32    : ValueType<32 , 85>;    //    1 x f32 vector value
+def v2f32    : ValueType<64 , 86>;    //    2 x f32 vector value
+def v3f32    : ValueType<96 , 87>;    //    3 x f32 vector value
+def v4f32    : ValueType<128, 88>;    //    4 x f32 vector value
+def v5f32    : ValueType<160, 89>;    //    5 x f32 vector value
+def v8f32    : ValueType<256, 90>;    //    8 x f32 vector value
+def v16f32   : ValueType<512,  91>;   //   16 x f32 vector value
+def v32f32   : ValueType<1024, 92>;   //   32 x f32 vector value
+def v64f32   : ValueType<2048, 93>;   //   64 x f32 vector value
+def v128f32  : ValueType<4096, 94>;   //  128 x f32 vector value
+def v256f32  : ValueType<8182, 95>;   //  256 x f32 vector value
+def v512f32  : ValueType<16384, 96>;  //  512 x f32 vector value
+def v1024f32 : ValueType<32768, 97>;  // 1024 x f32 vector value
+def v2048f32 : ValueType<65536, 98>;  // 2048 x f32 vector value
+def v1f64    : ValueType<64, 99>;     //    1 x f64 vector value
+def v2f64    : ValueType<128, 100>;   //    2 x f64 vector value
+def v4f64    : ValueType<256, 101>;   //    4 x f64 vector value
+def v8f64    : ValueType<512, 102>;   //    8 x f64 vector value
+def v16f64   : ValueType<1024, 103>;  //   16 x f64 vector value
+def v32f64   : ValueType<2048, 104>;  //   32 x f64 vector value
+def v64f64   : ValueType<4096, 105>;  //   64 x f64 vector value
+def v128f64  : ValueType<8192, 106>;  //  128 x f64 vector value
+def v256f64  : ValueType<16384, 107>; //  256 x f64 vector value
+
+def nxv1i1  : ValueType<1,  108>;  // n x  1 x i1  vector value
+def nxv2i1  : ValueType<2,  109>;  // n x  2 x i1  vector value
+def nxv4i1  : ValueType<4,  110>;  // n x  4 x i1  vector value
+def nxv8i1  : ValueType<8,  111>;  // n x  8 x i1  vector value
+def nxv16i1 : ValueType<16, 112>;  // n x 16 x i1  vector value
+def nxv32i1 : ValueType<32, 113>;  // n x 32 x i1  vector value
+def nxv64i1  : ValueType<64,114>;  // n x  64 x i1  vector value
+
+def nxv1i8  : ValueType<8,   115>;  // n x  1 x i8  vector value
+def nxv2i8  : ValueType<16,  116>;  // n x  2 x i8  vector value
+def nxv4i8  : ValueType<32,  117>;  // n x  4 x i8  vector value
+def nxv8i8  : ValueType<64,  118>;  // n x  8 x i8  vector value
+def nxv16i8 : ValueType<128, 119>;  // n x 16 x i8  vector value
+def nxv32i8 : ValueType<256, 120>;  // n x 32 x i8  vector value
+def nxv64i8  : ValueType<512, 121>; // n x  64 x i8  vector value
+
+def nxv1i16 : ValueType<16,  122>; // n x  1 x i16 vector value
+def nxv2i16 : ValueType<32,  123>; // n x  2 x i16 vector value
+def nxv4i16 : ValueType<64,  124>; // n x  4 x i16 vector value
+def nxv8i16 : ValueType<128, 125>; // n x  8 x i16 vector value
+def nxv16i16: ValueType<256, 126>; // n x 16 x i16 vector value
+def nxv32i16: ValueType<512, 127>; // n x 32 x i16 vector value
+
+def nxv1i32 : ValueType<32,  128>; // n x  1 x i32 vector value
+def nxv2i32 : ValueType<64,  129>; // n x  2 x i32 vector value
+def nxv4i32 : ValueType<128, 130>; // n x  4 x i32 vector value
+def nxv8i32 : ValueType<256, 131>; // n x  8 x i32 vector value
+def nxv16i32: ValueType<512, 132>; // n x 16 x i32 vector value
+def nxv32i32: ValueType<1024,133>; // n x 32 x i32 vector value
+
+def nxv1i64 : ValueType<64,  134>; // n x  1 x i64 vector value
+def nxv2i64 : ValueType<128, 135>; // n x  2 x i64 vector value
+def nxv4i64 : ValueType<256, 136>; // n x  4 x i64 vector value
+def nxv8i64 : ValueType<512, 137>; // n x  8 x i64 vector value
+def nxv16i64: ValueType<1024,138>; // n x 16 x i64 vector value
+def nxv32i64: ValueType<2048,139>; // n x 32 x i64 vector value
+
+def nxv1f16  : ValueType<16, 140>; // n x  1 x  f16 vector value
+def nxv2f16  : ValueType<32, 141>; // n x  2 x  f16 vector value
+def nxv4f16  : ValueType<64, 142>; // n x  4 x  f16 vector value
+def nxv8f16  : ValueType<128,143>; // n x  8 x  f16 vector value
+def nxv16f16 : ValueType<256,144>; // n x 16 x  f16 vector value
+def nxv32f16 : ValueType<512,145>; // n x 32 x  f16 vector value
+def nxv1bf16 : ValueType<16, 146>; // n x  1 x bf16 vector value
+def nxv2bf16 : ValueType<32, 147>; // n x  2 x bf16 vector value
+def nxv4bf16 : ValueType<64, 148>; // n x  4 x bf16 vector value
+def nxv8bf16 : ValueType<128,149>; // n x  8 x bf16 vector value
+def nxv1f32  : ValueType<32, 150>; // n x  1 x  f32 vector value
+def nxv2f32  : ValueType<64, 151>; // n x  2 x  f32 vector value
+def nxv4f32  : ValueType<128,152>; // n x  4 x  f32 vector value
+def nxv8f32  : ValueType<256,153>; // n x  8 x  f32 vector value
+def nxv16f32 : ValueType<512,154>; // n x 16 x  f32 vector value
+def nxv1f64  : ValueType<64, 155>; // n x  1 x  f64 vector value
+def nxv2f64  : ValueType<128,156>; // n x  2 x  f64 vector value
+def nxv4f64  : ValueType<256,157>; // n x  4 x  f64 vector value
+def nxv8f64  : ValueType<512,158>; // n x  8 x  f64 vector value
+
+def x86mmx : ValueType<64,   159>; // X86 MMX value
+def FlagVT : ValueType<0,    160>; // Pre-RA sched glue
+def isVoid : ValueType<0,    161>; // Produces no value
+def untyped: ValueType<8,    162>; // Produces an untyped value
+def funcref : ValueType<0,   163>; // WebAssembly's funcref type
+def externref : ValueType<0, 164>; // WebAssembly's externref type
+def x86amx : ValueType<8192, 165>; // X86 AMX value
 
 
 def token  : ValueType<0  , 248>;   // TokenTy
 def MetadataVT: ValueType<0,249>;   // Metadata
 
 // Pseudo valuetype mapped to the current pointer size to any address space.
 // Should only be used in TableGen.
 def iPTRAny   : ValueType<0, 250>;

llvm/include/llvm/Support/MachineValueType.h

enum SimpleValueType : uint8_t {
       v128i64        =  65,   // 128 x i64
       v256i64        =  66,   // 256 x i64
 
       v1i128         =  67,   //  1 x i128
 
       FIRST_INTEGER_FIXEDLEN_VECTOR_VALUETYPE = v1i1,
       LAST_INTEGER_FIXEDLEN_VECTOR_VALUETYPE = v1i128,
 
-      v2f16          =  68,   //    2 x f16
-      v3f16          =  69,   //    3 x f16
-      v4f16          =  70,   //    4 x f16
-      v8f16          =  71,   //    8 x f16
-      v16f16         =  72,   //   16 x f16
-      v32f16         =  73,   //   32 x f16
-      v64f16         =  74,   //   64 x f16
-      v128f16        =  75,   //  128 x f16
-      v2bf16         =  76,   //    2 x bf16
-      v3bf16         =  77,   //    3 x bf16
-      v4bf16         =  78,   //    4 x bf16
-      v8bf16         =  79,   //    8 x bf16
-      v16bf16        =  80,   //   16 x bf16
-      v32bf16        =  81,   //   32 x bf16
-      v64bf16        =  82,   //   64 x bf16
-      v128bf16       =  83,   //  128 x bf16
-      v1f32          =  84,   //    1 x f32
-      v2f32          =  85,   //    2 x f32
-      v3f32          =  86,   //    3 x f32
-      v4f32          =  87,   //    4 x f32
-      v5f32          =  88,   //    5 x f32
-      v8f32          =  89,   //    8 x f32
-      v16f32         =  90,   //   16 x f32
-      v32f32         =  91,   //   32 x f32
-      v64f32         =  92,   //   64 x f32
-      v128f32        =  93,   //  128 x f32
-      v256f32        =  94,   //  256 x f32
-      v512f32        =  95,   //  512 x f32
-      v1024f32       =  96,   // 1024 x f32
-      v2048f32       =  97,   // 2048 x f32
-      v1f64          =  98,   //    1 x f64
-      v2f64          =  99,   //    2 x f64
-      v4f64          = 100,   //    4 x f64
-      v8f64          = 101,   //    8 x f64
-      v16f64         = 102,   //   16 x f64
-      v32f64         = 103,   //   32 x f64
-      v64f64         = 104,   //   64 x f64
-      v128f64        = 105,   //  128 x f64
-      v256f64        = 106,   //  256 x f64
+      v1f16          =  68,   //    1 x f16
+      v2f16          =  69,   //    2 x f16
+      v3f16          =  70,   //    3 x f16
+      v4f16          =  71,   //    4 x f16
+      v8f16          =  72,   //    8 x f16
+      v16f16         =  73,   //   16 x f16
+      v32f16         =  74,   //   32 x f16
+      v64f16         =  75,   //   64 x f16
+      v128f16        =  76,   //  128 x f16
+      v2bf16         =  77,   //    2 x bf16
+      v3bf16         =  78,   //    3 x bf16
+      v4bf16         =  79,   //    4 x bf16
+      v8bf16         =  80,   //    8 x bf16
+      v16bf16        =  81,   //   16 x bf16
+      v32bf16        =  82,   //   32 x bf16
+      v64bf16        =  83,   //   64 x bf16
+      v128bf16       =  84,   //  128 x bf16
+      v1f32          =  85,   //    1 x f32
+      v2f32          =  86,   //    2 x f32
+      v3f32          =  87,   //    3 x f32
+      v4f32          =  88,   //    4 x f32
+      v5f32          =  89,   //    5 x f32
+      v8f32          =  90,   //    8 x f32
+      v16f32         =  91,   //   16 x f32
+      v32f32         =  92,   //   32 x f32
+      v64f32         =  93,   //   64 x f32
+      v128f32        =  94,   //  128 x f32
+      v256f32        =  95,   //  256 x f32
+      v512f32        =  96,   //  512 x f32
+      v1024f32       =  97,   // 1024 x f32
+      v2048f32       =  98,   // 2048 x f32
+      v1f64          =  99,   //    1 x f64
+      v2f64          = 100,   //    2 x f64
+      v4f64          = 101,   //    4 x f64
+      v8f64          = 102,   //    8 x f64
+      v16f64         = 103,   //   16 x f64
+      v32f64         = 104,   //   32 x f64
+      v64f64         = 105,   //   64 x f64
+      v128f64        = 106,   //  128 x f64
+      v256f64        = 107,   //  256 x f64
 
-      FIRST_FP_FIXEDLEN_VECTOR_VALUETYPE = v2f16,
+      FIRST_FP_FIXEDLEN_VECTOR_VALUETYPE = v1f16,
       LAST_FP_FIXEDLEN_VECTOR_VALUETYPE = v256f64,
 
       FIRST_FIXEDLEN_VECTOR_VALUETYPE = v1i1,
       LAST_FIXEDLEN_VECTOR_VALUETYPE = v256f64,
 
-      nxv1i1         = 107,   // n x  1 x i1
-      nxv2i1         = 108,   // n x  2 x i1
-      nxv4i1         = 109,   // n x  4 x i1
-      nxv8i1         = 110,   // n x  8 x i1
-      nxv16i1        = 111,   // n x 16 x i1
-      nxv32i1        = 112,   // n x 32 x i1
-      nxv64i1        = 113,   // n x  64 x i1
-
-      nxv1i8         = 114,   // n x  1 x i8
-      nxv2i8         = 115,   // n x  2 x i8
-      nxv4i8         = 116,   // n x  4 x i8
-      nxv8i8         = 117,   // n x  8 x i8
-      nxv16i8        = 118,   // n x 16 x i8
-      nxv32i8        = 119,   // n x 32 x i8
-      nxv64i8        = 120,   // n x  64 x i8
-
-      nxv1i16        = 121,  // n x  1 x i16
-      nxv2i16        = 122,  // n x  2 x i16
-      nxv4i16        = 123,  // n x  4 x i16
-      nxv8i16        = 124,  // n x  8 x i16
-      nxv16i16       = 125,  // n x 16 x i16
-      nxv32i16       = 126,  // n x 32 x i16
-
-      nxv1i32        = 127,  // n x  1 x i32
-      nxv2i32        = 128,  // n x  2 x i32
-      nxv4i32        = 129,  // n x  4 x i32
-      nxv8i32        = 130,  // n x  8 x i32
-      nxv16i32       = 131,  // n x 16 x i32
-      nxv32i32       = 132,  // n x 32 x i32
-
-      nxv1i64        = 133,  // n x  1 x i64
-      nxv2i64        = 134,  // n x  2 x i64
-      nxv4i64        = 135,  // n x  4 x i64
-      nxv8i64        = 136,  // n x  8 x i64
-      nxv16i64       = 137,  // n x 16 x i64
-      nxv32i64       = 138,  // n x 32 x i64
+      nxv1i1         = 108,   // n x  1 x i1
+      nxv2i1         = 109,   // n x  2 x i1
+      nxv4i1         = 110,   // n x  4 x i1
+      nxv8i1         = 111,   // n x  8 x i1
+      nxv16i1        = 112,   // n x 16 x i1
+      nxv32i1        = 113,   // n x 32 x i1
+      nxv64i1        = 114,   // n x  64 x i1
+
+      nxv1i8         = 115,   // n x  1 x i8
+      nxv2i8         = 116,   // n x  2 x i8
+      nxv4i8         = 117,   // n x  4 x i8
+      nxv8i8         = 118,   // n x  8 x i8
+      nxv16i8        = 119,   // n x 16 x i8
+      nxv32i8        = 120,   // n x 32 x i8
+      nxv64i8        = 121,   // n x  64 x i8
+
+      nxv1i16        = 122,  // n x  1 x i16
+      nxv2i16        = 123,  // n x  2 x i16
+      nxv4i16        = 124,  // n x  4 x i16
+      nxv8i16        = 125,  // n x  8 x i16
+      nxv16i16       = 126,  // n x 16 x i16
+      nxv32i16       = 127,  // n x 32 x i16
+
+      nxv1i32        = 128,  // n x  1 x i32
+      nxv2i32        = 129,  // n x  2 x i32
+      nxv4i32        = 130,  // n x  4 x i32
+      nxv8i32        = 131,  // n x  8 x i32
+      nxv16i32       = 132,  // n x 16 x i32
+      nxv32i32       = 133,  // n x 32 x i32
+
+      nxv1i64        = 134,  // n x  1 x i64
+      nxv2i64        = 135,  // n x  2 x i64
+      nxv4i64        = 136,  // n x  4 x i64
+      nxv8i64        = 137,  // n x  8 x i64
+      nxv16i64       = 138,  // n x 16 x i64
+      nxv32i64       = 139,  // n x 32 x i64
 
       FIRST_INTEGER_SCALABLE_VECTOR_VALUETYPE = nxv1i1,
       LAST_INTEGER_SCALABLE_VECTOR_VALUETYPE = nxv32i64,
 
-      nxv1f16        = 139,   // n x   1 x f16
-      nxv2f16        = 140,  // n x  2 x f16
-      nxv4f16        = 141,  // n x  4 x f16
-      nxv8f16        = 142,  // n x  8 x f16
-      nxv16f16       = 143,   // n x  16 x f16
-      nxv32f16       = 144,   // n x  32 x f16
-      nxv1bf16       = 145,  // n x  1 x bf16
-      nxv2bf16       = 146,  // n x  2 x bf16
-      nxv4bf16       = 147,  // n x  4 x bf16
-      nxv8bf16       = 148,  // n x  8 x bf16
-      nxv1f32        = 149,  // n x  1 x f32
-      nxv2f32        = 150,  // n x  2 x f32
-      nxv4f32        = 151,  // n x  4 x f32
-      nxv8f32        = 152,  // n x  8 x f32
-      nxv16f32       = 153,  // n x 16 x f32
-      nxv1f64        = 154,  // n x  1 x f64
-      nxv2f64        = 155,  // n x  2 x f64
-      nxv4f64        = 156,  // n x  4 x f64
-      nxv8f64        = 157,  // n x  8 x f64
+      nxv1f16        = 140,  // n x  1 x f16
+      nxv2f16        = 141,  // n x  2 x f16
+      nxv4f16        = 142,  // n x  4 x f16
+      nxv8f16        = 143,  // n x  8 x f16
+      nxv16f16       = 144,  // n x 16 x f16
+      nxv32f16       = 145,  // n x 32 x f16
+      nxv1bf16       = 146,  // n x  1 x bf16
+      nxv2bf16       = 147,  // n x  2 x bf16
+      nxv4bf16       = 148,  // n x  4 x bf16
+      nxv8bf16       = 149,  // n x  8 x bf16
+      nxv1f32        = 150,  // n x  1 x f32
+      nxv2f32        = 151,  // n x  2 x f32
+      nxv4f32        = 152,  // n x  4 x f32
+      nxv8f32        = 153,  // n x  8 x f32
+      nxv16f32       = 154,  // n x 16 x f32
+      nxv1f64        = 155,  // n x  1 x f64
+      nxv2f64        = 156,  // n x  2 x f64
+      nxv4f64        = 157,  // n x  4 x f64
+      nxv8f64        = 158,  // n x  8 x f64
 
       FIRST_FP_SCALABLE_VECTOR_VALUETYPE = nxv1f16,
       LAST_FP_SCALABLE_VECTOR_VALUETYPE = nxv8f64,
 
       FIRST_SCALABLE_VECTOR_VALUETYPE = nxv1i1,
       LAST_SCALABLE_VECTOR_VALUETYPE = nxv8f64,
 
       FIRST_VECTOR_VALUETYPE = v1i1,
       LAST_VECTOR_VALUETYPE  = nxv8f64,
 
-      x86mmx         = 158,   // This is an X86 MMX value
+      x86mmx         = 159,   // This is an X86 MMX value
 
-      Glue           = 159,   // This glues nodes together during pre-RA sched
+      Glue           = 160,   // This glues nodes together during pre-RA sched
 
-      isVoid         = 160,   // This has no value
+      isVoid         = 161,   // This has no value
 
-      Untyped        = 161,   // This value takes a register, but has
+      Untyped        = 162,   // This value takes a register, but has
                               // unspecified type.  The register class
                               // will be determined by the opcode.
 
-      funcref        = 162,   // WebAssembly's funcref type
-      externref      = 163,   // WebAssembly's externref type
-      x86amx         = 164,   // This is an X86 AMX value
+      funcref        = 163,   // WebAssembly's funcref type
+      externref      = 164,   // WebAssembly's externref type
+      x86amx         = 165,   // This is an X86 AMX value
 
       FIRST_VALUETYPE =  1,   // This is always the beginning of the list.
-      LAST_VALUETYPE = 165,   // This always remains at the end of the list.
+      LAST_VALUETYPE = 166,   // This always remains at the end of the list.
 
       // This is the current maximum for LAST_VALUETYPE.
       // MVT::MAX_ALLOWED_VALUETYPE is used for asserts and to size bit vectors
       // This value must be a multiple of 32.
       MAX_ALLOWED_VALUETYPE = 192,
 
       // A value of type llvm::TokenTy
       token          = 248,
public:
     bool isFixedLengthVector() const {
       return (SimpleTy >= MVT::FIRST_FIXEDLEN_VECTOR_VALUETYPE &&
               SimpleTy <= MVT::LAST_FIXEDLEN_VECTOR_VALUETYPE);
     }
 
     /// Return true if this is a 16-bit vector type.
     bool is16BitVector() const {
       return (SimpleTy == MVT::v2i8  || SimpleTy == MVT::v1i16 ||
-              SimpleTy == MVT::v16i1);
+              SimpleTy == MVT::v16i1 || SimpleTy == MVT::v1f16);
     }
 
     /// Return true if this is a 32-bit vector type.
     bool is32BitVector() const {
       return (SimpleTy == MVT::v32i1 || SimpleTy == MVT::v4i8   ||
               SimpleTy == MVT::v2i16 || SimpleTy == MVT::v1i32  ||
               SimpleTy == MVT::v2f16 || SimpleTy == MVT::v2bf16 ||
               SimpleTy == MVT::v1f32);
MVT getVectorElementType() const {
       case v256i64:
       case nxv1i64:
       case nxv2i64:
       case nxv4i64:
       case nxv8i64:
       case nxv16i64:
       case nxv32i64: return i64;
       case v1i128: return i128;
+      case v1f16:
       case v2f16:
       case v3f16:
       case v4f16:
       case v8f16:
       case v16f16:
       case v32f16:
       case v64f16:
       case v128f16:
unsigned getVectorNumElements() const {
       case nxv2f32:
       case nxv2f64: return 2;
       case v1i1:
       case v1i8:
       case v1i16:
       case v1i32:
       case v1i64:
       case v1i128:
+      case v1f16:
       case v1f32:
       case v1f64:
       case nxv1i1:
       case nxv1i8:
       case nxv1i16:
       case nxv1i32:
       case nxv1i64:
       case nxv1f16:
TypeSize getSizeInBits() const {
       case v8i1: return TypeSize::Fixed(8);
       case nxv1i8:
       case nxv8i1: return TypeSize::Scalable(8);
       case i16 :
       case f16:
       case bf16:
       case v16i1:
       case v2i8:
-      case v1i16: return TypeSize::Fixed(16);
+      case v1i16:
+      case v1f16: return TypeSize::Fixed(16);
       case nxv16i1:
       case nxv2i8:
       case nxv1i16:
       case nxv1bf16:
       case nxv1f16: return TypeSize::Scalable(16);
       case f32 :
       case i32 :
       case v32i1:
static MVT getVectorVT(MVT VT, unsigned NumElements) {
         if (NumElements == 64) return MVT::v64i64;
         if (NumElements == 128) return MVT::v128i64;
         if (NumElements == 256) return MVT::v256i64;
         break;
       case MVT::i128:
         if (NumElements == 1)  return MVT::v1i128;
         break;
       case MVT::f16:
+        if (NumElements == 1)   return MVT::v1f16;
         if (NumElements == 2)   return MVT::v2f16;
         if (NumElements == 3)   return MVT::v3f16;
         if (NumElements == 4)   return MVT::v4f16;
         if (NumElements == 8)   return MVT::v8f16;
         if (NumElements == 16)  return MVT::v16f16;
         if (NumElements == 32)  return MVT::v32f16;
         if (NumElements == 64)  return MVT::v64f16;
         if (NumElements == 128) return MVT::v128f16;

llvm/lib/CodeGen/ValueTypes.cpp

Type *EVT::getTypeForEVT(LLVMContext &Context) const {
   case MVT::v64i64:
     return FixedVectorType::get(Type::getInt64Ty(Context), 64);
   case MVT::v128i64:
     return FixedVectorType::get(Type::getInt64Ty(Context), 128);
   case MVT::v256i64:
     return FixedVectorType::get(Type::getInt64Ty(Context), 256);
   case MVT::v1i128:
     return FixedVectorType::get(Type::getInt128Ty(Context), 1);
+  case MVT::v1f16:
+    return FixedVectorType::get(Type::getHalfTy(Context), 1);
   case MVT::v2f16:
     return FixedVectorType::get(Type::getHalfTy(Context), 2);
   case MVT::v3f16:
     return FixedVectorType::get(Type::getHalfTy(Context), 3);
   case MVT::v4f16:
     return FixedVectorType::get(Type::getHalfTy(Context), 4);
   case MVT::v8f16:
     return FixedVectorType::get(Type::getHalfTy(Context), 8);

llvm/test/CodeGen/RISCV/rvv/fixed-vectors-reduction-fp.ll

 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc -mtriple=riscv32 -target-abi=ilp32d -mattr=+experimental-v,+experimental-zfh,+f,+d -verify-machineinstrs -riscv-v-vector-bits-min=128 -verify-machineinstrs < %s | FileCheck %s --check-prefixes=CHECK
 ; RUN: llc -mtriple=riscv64 -target-abi=lp64d -mattr=+experimental-v,+experimental-zfh,+f,+d -verify-machineinstrs -riscv-v-vector-bits-min=128  -verify-machineinstrs < %s | FileCheck %s --check-prefixes=CHECK
 
 declare half @llvm.vector.reduce.fadd.v1f16(half, <1 x half>)
 
 define half @vreduce_fadd_v1f16(<1 x half>* %x, half %s) {
 ; CHECK-LABEL: vreduce_fadd_v1f16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    flh ft0, 0(a0)
+; CHECK-NEXT:    vsetivli a1, 1, e16,m1,ta,mu
+; CHECK-NEXT:    vle16.v v25, (a0)
+; CHECK-NEXT:    vfmv.f.s ft0, v25
 ; CHECK-NEXT:    fadd.h fa0, fa0, ft0
 ; CHECK-NEXT:    ret
   %v = load <1 x half>, <1 x half>* %x
   %red = call reassoc half @llvm.vector.reduce.fadd.v1f16(half %s, <1 x half> %v)
   ret half %red
 }
 
 define half @vreduce_ord_fadd_v1f16(<1 x half>* %x, half %s) {
 ; CHECK-LABEL: vreduce_ord_fadd_v1f16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    flh ft0, 0(a0)
-; CHECK-NEXT:    fadd.h fa0, fa0, ft0
+; CHECK-NEXT:    vsetivli a1, 1, e16,m1,ta,mu
+; CHECK-NEXT:    vle16.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e16,m1,ta,mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetivli a0, 1, e16,m1,ta,mu
+; CHECK-NEXT:    vfredosum.vs v25, v25, v26
+; CHECK-NEXT:    vfmv.f.s fa0, v25
 ; CHECK-NEXT:    ret
   %v = load <1 x half>, <1 x half>* %x
   %red = call half @llvm.vector.reduce.fadd.v1f16(half %s, <1 x half> %v)
   ret half %red
 }
 
 declare half @llvm.vector.reduce.fadd.v2f16(half, <2 x half>)