//===- IntrinsicsARM.td - Defines ARM intrinsics -----------*- tablegen -*-===//
-//
+//
// The LLVM Compiler Infrastructure
//
// This file is distributed under the University of Illinois Open Source
// License. See LICENSE.TXT for details.
-//
+//
//===----------------------------------------------------------------------===//
//
// This file defines all of the ARM-specific intrinsics.
Intrinsic<[llvm_ptr_ty], [], [IntrNoMem]>;
}
+//===----------------------------------------------------------------------===//
+// Saturating Arithmentic
+
+let TargetPrefix = "arm" in { // All intrinsics start with "llvm.arm.".
+ def int_arm_qadd : GCCBuiltin<"__builtin_arm_qadd">,
+ Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty],
+ [IntrNoMem, Commutative]>;
+ def int_arm_qsub : GCCBuiltin<"__builtin_arm_qsub">,
+ Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrNoMem]>;
+ def int_arm_ssat : GCCBuiltin<"__builtin_arm_ssat">,
+ Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrNoMem]>;
+ def int_arm_usat : GCCBuiltin<"__builtin_arm_usat">,
+ Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrNoMem]>;
+}
+
+//===----------------------------------------------------------------------===//
+// VFP
+
+let TargetPrefix = "arm" in { // All intrinsics start with "llvm.arm.".
+ def int_arm_get_fpscr : GCCBuiltin<"__builtin_arm_get_fpscr">,
+ Intrinsic<[llvm_i32_ty], [], [IntrNoMem]>;
+ def int_arm_set_fpscr : GCCBuiltin<"__builtin_arm_set_fpscr">,
+ Intrinsic<[], [llvm_i32_ty], []>;
+ def int_arm_vcvtr : Intrinsic<[llvm_float_ty], [llvm_anyfloat_ty],
+ [IntrNoMem]>;
+ def int_arm_vcvtru : Intrinsic<[llvm_float_ty], [llvm_anyfloat_ty],
+ [IntrNoMem]>;
+}
+
//===----------------------------------------------------------------------===//
// Advanced SIMD (NEON)
class Neon_1Arg_Narrow_Intrinsic
: Intrinsic<[llvm_anyvector_ty],
[LLVMExtendedElementVectorType<0>], [IntrNoMem]>;
- class Neon_1Arg_Long_Intrinsic
- : Intrinsic<[llvm_anyvector_ty],
- [LLVMTruncatedElementVectorType<0>], [IntrNoMem]>;
class Neon_2Arg_Intrinsic
: Intrinsic<[llvm_anyvector_ty], [LLVMMatchType<0>, LLVMMatchType<0>],
[IntrNoMem]>;
[LLVMTruncatedElementVectorType<0>,
LLVMTruncatedElementVectorType<0>],
[IntrNoMem]>;
- class Neon_2Arg_Wide_Intrinsic
- : Intrinsic<[llvm_anyvector_ty],
- [LLVMMatchType<0>, LLVMTruncatedElementVectorType<0>],
- [IntrNoMem]>;
class Neon_3Arg_Intrinsic
: Intrinsic<[llvm_anyvector_ty],
[LLVMMatchType<0>, LLVMMatchType<0>, LLVMMatchType<0>],
def int_arm_neon_vqaddu : Neon_2Arg_Intrinsic;
def int_arm_neon_vaddhn : Neon_2Arg_Narrow_Intrinsic;
def int_arm_neon_vraddhn : Neon_2Arg_Narrow_Intrinsic;
- def int_arm_neon_vaddls : Neon_2Arg_Long_Intrinsic;
- def int_arm_neon_vaddlu : Neon_2Arg_Long_Intrinsic;
- def int_arm_neon_vaddws : Neon_2Arg_Wide_Intrinsic;
- def int_arm_neon_vaddwu : Neon_2Arg_Wide_Intrinsic;
// Vector Multiply.
def int_arm_neon_vmulp : Neon_2Arg_Intrinsic;
def int_arm_neon_vqdmulh : Neon_2Arg_Intrinsic;
def int_arm_neon_vqrdmulh : Neon_2Arg_Intrinsic;
- def int_arm_neon_vmulls : Neon_2Arg_Long_Intrinsic;
- def int_arm_neon_vmullu : Neon_2Arg_Long_Intrinsic;
def int_arm_neon_vmullp : Neon_2Arg_Long_Intrinsic;
def int_arm_neon_vqdmull : Neon_2Arg_Long_Intrinsic;
-
- // Vector Multiply and Accumulate/Subtract.
- def int_arm_neon_vmlals : Neon_3Arg_Long_Intrinsic;
- def int_arm_neon_vmlalu : Neon_3Arg_Long_Intrinsic;
- def int_arm_neon_vmlsls : Neon_3Arg_Long_Intrinsic;
- def int_arm_neon_vmlslu : Neon_3Arg_Long_Intrinsic;
def int_arm_neon_vqdmlal : Neon_3Arg_Long_Intrinsic;
def int_arm_neon_vqdmlsl : Neon_3Arg_Long_Intrinsic;
def int_arm_neon_vqsubu : Neon_2Arg_Intrinsic;
def int_arm_neon_vsubhn : Neon_2Arg_Narrow_Intrinsic;
def int_arm_neon_vrsubhn : Neon_2Arg_Narrow_Intrinsic;
-def int_arm_neon_vsubls : Neon_2Arg_Long_Intrinsic;
-def int_arm_neon_vsublu : Neon_2Arg_Long_Intrinsic;
-def int_arm_neon_vsubws : Neon_2Arg_Wide_Intrinsic;
-def int_arm_neon_vsubwu : Neon_2Arg_Wide_Intrinsic;
// Vector Absolute Compare.
let TargetPrefix = "arm" in {
// Vector Absolute Differences.
def int_arm_neon_vabds : Neon_2Arg_Intrinsic;
def int_arm_neon_vabdu : Neon_2Arg_Intrinsic;
-def int_arm_neon_vabdls : Neon_2Arg_Long_Intrinsic;
-def int_arm_neon_vabdlu : Neon_2Arg_Long_Intrinsic;
-
-// Vector Absolute Difference and Accumulate.
-def int_arm_neon_vabas : Neon_3Arg_Intrinsic;
-def int_arm_neon_vabau : Neon_3Arg_Intrinsic;
-def int_arm_neon_vabals : Neon_3Arg_Long_Intrinsic;
-def int_arm_neon_vabalu : Neon_3Arg_Long_Intrinsic;
// Vector Pairwise Add.
def int_arm_neon_vpadd : Neon_2Arg_Intrinsic;
def int_arm_neon_vcvtfxs2fp : Neon_CvtFxToFP_Intrinsic;
def int_arm_neon_vcvtfxu2fp : Neon_CvtFxToFP_Intrinsic;
-// Narrowing and Lengthening Vector Moves.
-def int_arm_neon_vmovn : Neon_1Arg_Narrow_Intrinsic;
+// Vector Conversions Between Half-Precision and Single-Precision.
+def int_arm_neon_vcvtfp2hf
+ : Intrinsic<[llvm_v4i16_ty], [llvm_v4f32_ty], [IntrNoMem]>;
+def int_arm_neon_vcvthf2fp
+ : Intrinsic<[llvm_v4f32_ty], [llvm_v4i16_ty], [IntrNoMem]>;
+
+// Narrowing Saturating Vector Moves.
def int_arm_neon_vqmovns : Neon_1Arg_Narrow_Intrinsic;
def int_arm_neon_vqmovnu : Neon_1Arg_Narrow_Intrinsic;
def int_arm_neon_vqmovnsu : Neon_1Arg_Narrow_Intrinsic;
-def int_arm_neon_vmovls : Neon_1Arg_Long_Intrinsic;
-def int_arm_neon_vmovlu : Neon_1Arg_Long_Intrinsic;
// Vector Table Lookup.
// The first 1-4 arguments are the table.
let TargetPrefix = "arm" in {
// De-interleaving vector loads from N-element structures.
+ // Source operands are the address and alignment.
def int_arm_neon_vld1 : Intrinsic<[llvm_anyvector_ty],
- [llvm_ptr_ty], [IntrReadArgMem]>;
+ [llvm_ptr_ty, llvm_i32_ty],
+ [IntrReadArgMem]>;
def int_arm_neon_vld2 : Intrinsic<[llvm_anyvector_ty, LLVMMatchType<0>],
- [llvm_ptr_ty], [IntrReadArgMem]>;
+ [llvm_ptr_ty, llvm_i32_ty],
+ [IntrReadArgMem]>;
def int_arm_neon_vld3 : Intrinsic<[llvm_anyvector_ty, LLVMMatchType<0>,
LLVMMatchType<0>],
- [llvm_ptr_ty], [IntrReadArgMem]>;
+ [llvm_ptr_ty, llvm_i32_ty],
+ [IntrReadArgMem]>;
def int_arm_neon_vld4 : Intrinsic<[llvm_anyvector_ty, LLVMMatchType<0>,
LLVMMatchType<0>, LLVMMatchType<0>],
- [llvm_ptr_ty], [IntrReadArgMem]>;
+ [llvm_ptr_ty, llvm_i32_ty],
+ [IntrReadArgMem]>;
// Vector load N-element structure to one lane.
+ // Source operands are: the address, the N input vectors (since only one
+ // lane is assigned), the lane number, and the alignment.
def int_arm_neon_vld2lane : Intrinsic<[llvm_anyvector_ty, LLVMMatchType<0>],
[llvm_ptr_ty, LLVMMatchType<0>,
- LLVMMatchType<0>, llvm_i32_ty],
- [IntrReadArgMem]>;
+ LLVMMatchType<0>, llvm_i32_ty,
+ llvm_i32_ty], [IntrReadArgMem]>;
def int_arm_neon_vld3lane : Intrinsic<[llvm_anyvector_ty, LLVMMatchType<0>,
LLVMMatchType<0>],
[llvm_ptr_ty, LLVMMatchType<0>,
LLVMMatchType<0>, LLVMMatchType<0>,
- llvm_i32_ty], [IntrReadArgMem]>;
+ llvm_i32_ty, llvm_i32_ty],
+ [IntrReadArgMem]>;
def int_arm_neon_vld4lane : Intrinsic<[llvm_anyvector_ty, LLVMMatchType<0>,
LLVMMatchType<0>, LLVMMatchType<0>],
[llvm_ptr_ty, LLVMMatchType<0>,
LLVMMatchType<0>, LLVMMatchType<0>,
- LLVMMatchType<0>, llvm_i32_ty],
- [IntrReadArgMem]>;
+ LLVMMatchType<0>, llvm_i32_ty,
+ llvm_i32_ty], [IntrReadArgMem]>;
// Interleaving vector stores from N-element structures.
+ // Source operands are: the address, the N vectors, and the alignment.
def int_arm_neon_vst1 : Intrinsic<[],
- [llvm_ptr_ty, llvm_anyvector_ty],
- [IntrWriteArgMem]>;
+ [llvm_ptr_ty, llvm_anyvector_ty,
+ llvm_i32_ty], [IntrReadWriteArgMem]>;
def int_arm_neon_vst2 : Intrinsic<[],
[llvm_ptr_ty, llvm_anyvector_ty,
- LLVMMatchType<0>], [IntrWriteArgMem]>;
+ LLVMMatchType<0>, llvm_i32_ty],
+ [IntrReadWriteArgMem]>;
def int_arm_neon_vst3 : Intrinsic<[],
[llvm_ptr_ty, llvm_anyvector_ty,
- LLVMMatchType<0>, LLVMMatchType<0>],
- [IntrWriteArgMem]>;
+ LLVMMatchType<0>, LLVMMatchType<0>,
+ llvm_i32_ty], [IntrReadWriteArgMem]>;
def int_arm_neon_vst4 : Intrinsic<[],
[llvm_ptr_ty, llvm_anyvector_ty,
LLVMMatchType<0>, LLVMMatchType<0>,
- LLVMMatchType<0>], [IntrWriteArgMem]>;
+ LLVMMatchType<0>, llvm_i32_ty],
+ [IntrReadWriteArgMem]>;
// Vector store N-element structure from one lane.
+ // Source operands are: the address, the N vectors, the lane number, and
+ // the alignment.
def int_arm_neon_vst2lane : Intrinsic<[],
[llvm_ptr_ty, llvm_anyvector_ty,
- LLVMMatchType<0>, llvm_i32_ty],
- [IntrWriteArgMem]>;
+ LLVMMatchType<0>, llvm_i32_ty,
+ llvm_i32_ty], [IntrReadWriteArgMem]>;
def int_arm_neon_vst3lane : Intrinsic<[],
[llvm_ptr_ty, llvm_anyvector_ty,
LLVMMatchType<0>, LLVMMatchType<0>,
- llvm_i32_ty], [IntrWriteArgMem]>;
+ llvm_i32_ty, llvm_i32_ty],
+ [IntrReadWriteArgMem]>;
def int_arm_neon_vst4lane : Intrinsic<[],
[llvm_ptr_ty, llvm_anyvector_ty,
LLVMMatchType<0>, LLVMMatchType<0>,
- LLVMMatchType<0>, llvm_i32_ty],
- [IntrWriteArgMem]>;
+ LLVMMatchType<0>, llvm_i32_ty,
+ llvm_i32_ty], [IntrReadWriteArgMem]>;
}
projects / oota-llvm.git / blobdiff