[AArch64] Add missing ASIMD FP convert instructions to scheduling model (llvm#115146)

Some ASIMD FP convert instructions have incorrect scheduling
information. These instructions currently have latency 2, throughput 4
and utilise pipeline V. This patch corrects the scheduling models to
match the relevant Software Optimization Guide.

The V1 and V2 Software Optimization Guide show that ASIMD FP convert
instructions should all utilise pipelines V02. Their execution latency
and throughput should also differ depending on form. See section 3.17
"ASIMD floating-point instructions" in the Neoverse-V1 and Neoverse-V2
Software Optimization Guide for characteristics of instruction
performance.

Reference:
- V1 SOG: https://developer.arm.com/documentation/109897/latest/
- V2 SOG: https://developer.arm.com/documentation/109898/latest/

Author: Rin Dobrescu

llvm/lib/Target/AArch64/AArch64SchedNeoverseV1.td

@@ -1015,16 +1015,34 @@ def : InstRW<[V1Write_3c_1V02], (instregex "^FCVTNv[24]i32$",
                                            "^FCVTXN(v[24]f32|v1i64)$")>;
 
 // ASIMD FP convert, other, D-form F32 and Q-form F64
-def : InstRW<[V1Write_3c_1V02], (instregex "^[FSU]CVT[AMNPZ][SU]v2f(32|64)$",
-                                           "^[SU]CVTFv2f(32|64)$")>;
+def : InstRW<[V1Write_3c_1V02], (instregex "^FCVT[AMNPZ][SU]v2f(32|64)$",
+                                           "^FCVT[AMNPZ][SU]v2i(32|64)_shift$",
+                                           "^FCVT[AMNPZ][SU]v1i64$",
+                                           "^FCVTZ[SU]d$",
+                                           "^[SU]CVTFv2f(32|64)$",
+                                           "^[SU]CVTFv2i(32|64)_shift$",
+                                           "^[SU]CVTFv1i64$",
+                                           "^[SU]CVTFd$")>;
 
 // ASIMD FP convert, other, D-form F16 and Q-form F32
-def : InstRW<[V1Write_4c_2V02], (instregex "^[FSU]CVT[AMNPZ][SU]v4f(16|32)$",
-                                           "^[SU]CVTFv4f(16|32)$")>;
+def : InstRW<[V1Write_4c_2V02], (instregex "^FCVT[AMNPZ][SU]v4f(16|32)$",
+                                           "^FCVT[AMNPZ][SU]v4i(16|32)_shift$",
+                                           "^FCVT[AMNPZ][SU]v1i32$",
+                                           "^FCVTZ[SU]s$",
+                                           "^[SU]CVTFv4f(16|32)$",
+                                           "^[SU]CVTFv4i(16|32)_shift$",
+                                           "^[SU]CVTFv1i32$",
+                                           "^[SU]CVTFs$")>;
 
 // ASIMD FP convert, other, Q-form F16
-def : InstRW<[V1Write_6c_4V02], (instregex "^[FSU]CVT[AMNPZ][SU]v8f16$",
-                                           "^[SU]CVTFv8f16$")>;
+def : InstRW<[V1Write_6c_4V02], (instregex "^FCVT[AMNPZ][SU]v8f16$",
+                                           "^FCVT[AMNPZ][SU]v8i16_shift$",
+                                           "^FCVT[AMNPZ][SU]v1f16$",
+                                           "^FCVTZ[SU]h$",
+                                           "^[SU]CVTFv8f16$",
+                                           "^[SU]CVTFv8i16_shift$",
+                                           "^[SU]CVTFv1i16$",
+                                           "^[SU]CVTFh$")>;
 
 // ASIMD FP divide, D-form, F16
 // ASIMD FP square root, D-form, F16

llvm/lib/Target/AArch64/AArch64SchedNeoverseV2.td

@@ -1567,25 +1567,31 @@ def : InstRW<[V2Write_3c_1V02], (instregex "^FCVTN(v2|v4)i32",
 
 // ASIMD FP convert, other, D-form F32 and Q-form F64
 def : InstRW<[V2Write_3c_1V02], (instregex "^FCVT[AMNPZ][SU]v2f(32|64)$",
+                                           "^FCVT[AMNPZ][SU]v2i(32|64)_shift$",
                                            "^FCVT[AMNPZ][SU]v1i64$",
                                            "^FCVTZ[SU]d$",
                                            "^[SU]CVTFv2f(32|64)$",
+                                           "^[SU]CVTFv2i(32|64)_shift$",
                                            "^[SU]CVTFv1i64$",
                                            "^[SU]CVTFd$")>;
 
 // ASIMD FP convert, other, D-form F16 and Q-form F32
 def : InstRW<[V2Write_4c_2V02], (instregex "^FCVT[AMNPZ][SU]v4f(16|32)$",
+                                           "^FCVT[AMNPZ][SU]v4i(16|32)_shift$",
                                            "^FCVT[AMNPZ][SU]v1i32$",
                                            "^FCVTZ[SU]s$",
                                            "^[SU]CVTFv4f(16|32)$",
+                                           "^[SU]CVTFv4i(16|32)_shift$",
                                            "^[SU]CVTFv1i32$",
                                            "^[SU]CVTFs$")>;
 
 // ASIMD FP convert, other, Q-form F16
 def : InstRW<[V2Write_6c_4V02], (instregex "^FCVT[AMNPZ][SU]v8f16$",
+                                           "^FCVT[AMNPZ][SU]v8i16_shift$",
                                            "^FCVT[AMNPZ][SU]v1f16$",
                                            "^FCVTZ[SU]h$",
                                            "^[SU]CVTFv8f16$",
+                                           "^[SU]CVTFv8i16_shift$",
                                            "^[SU]CVTFv1i16$",
                                            "^[SU]CVTFh$")>;