-//===- ARMScheduleV6.td - ARM v6 Scheduling Definitions ----*- tablegen -*-===//
-//
+//===-- ARMScheduleV6.td - ARM v6 Scheduling Definitions ---*- 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 the itinerary class data for the ARM v6 processors.
InstrItinData<IIC_iMOVsr , [InstrStage<2, [V6_Pipe]>], [3, 2, 1]>,
InstrItinData<IIC_iMOVix2 , [InstrStage<1, [V6_Pipe]>,
InstrStage<1, [V6_Pipe]>], [2]>,
+ InstrItinData<IIC_iMOVix2addpc,[InstrStage<1, [V6_Pipe]>,
+ InstrStage<1, [V6_Pipe]>,
+ InstrStage<1, [V6_Pipe]>], [3]>,
+ InstrItinData<IIC_iMOVix2ld , [InstrStage<1, [V6_Pipe]>,
+ InstrStage<1, [V6_Pipe]>,
+ InstrStage<1, [V6_Pipe]>], [5]>,
//
// Move instructions, conditional
InstrItinData<IIC_iCMOVi , [InstrStage<1, [V6_Pipe]>], [3]>,
InstrItinData<IIC_iCMOVr , [InstrStage<1, [V6_Pipe]>], [3, 2]>,
InstrItinData<IIC_iCMOVsi , [InstrStage<1, [V6_Pipe]>], [3, 1]>,
InstrItinData<IIC_iCMOVsr , [InstrStage<1, [V6_Pipe]>], [4, 2, 1]>,
+ InstrItinData<IIC_iCMOVix2 , [InstrStage<1, [V6_Pipe]>,
+ InstrStage<1, [V6_Pipe]>], [4]>,
//
// MVN instructions
InstrItinData<IIC_iMVNi , [InstrStage<1, [V6_Pipe]>], [2]>,
InstrItinData<IIC_iMAC32 , [InstrStage<2, [V6_Pipe]>], [5, 1, 1, 2]>,
InstrItinData<IIC_iMUL64 , [InstrStage<3, [V6_Pipe]>], [6, 1, 1]>,
InstrItinData<IIC_iMAC64 , [InstrStage<3, [V6_Pipe]>], [6, 1, 1, 2]>,
-
+
// Integer load pipeline
//
// Immediate offset
InstrItinData<IIC_iLoad_bh_siu,[InstrStage<2, [V6_Pipe]>], [5, 2, 2, 1]>,
//
- // Load multiple
- InstrItinData<IIC_iLoadm , [InstrStage<3, [V6_Pipe]>]>,
-
+ // Load multiple, def is the 5th operand.
+ InstrItinData<IIC_iLoad_m , [InstrStage<3, [V6_Pipe]>], [1, 1, 1, 1, 4]>,
+ //
+ // Load multiple + update, defs are the 1st and 5th operands.
+ InstrItinData<IIC_iLoad_mu , [InstrStage<3, [V6_Pipe]>], [2, 1, 1, 1, 4]>,
//
// Load multiple plus branch
- InstrItinData<IIC_iLoadmBr , [InstrStage<3, [V6_Pipe]>,
- InstrStage<1, [V6_Pipe]>]>,
+ InstrItinData<IIC_iLoad_mBr, [InstrStage<3, [V6_Pipe]>,
+ InstrStage<1, [V6_Pipe]>], [1, 2, 1, 1, 4]>,
//
// iLoadi + iALUr for t2LDRpci_pic.
InstrItinData<IIC_iLoadiALU, [InstrStage<1, [V6_Pipe]>,
InstrStage<1, [V6_Pipe]>], [3, 1]>,
+ //
+ // Pop, def is the 3rd operand.
+ InstrItinData<IIC_iPop , [InstrStage<3, [V6_Pipe]>], [1, 1, 4]>,
+ //
+ // Pop + branch, def is the 3rd operand.
+ InstrItinData<IIC_iPop_Br, [InstrStage<3, [V6_Pipe]>,
+ InstrStage<1, [V6_Pipe]>], [1, 2, 4]>,
+
// Integer store pipeline
//
// Immediate offset
InstrItinData<IIC_iStore_bh_siu,[InstrStage<2, [V6_Pipe]>], [2, 2, 2, 1]>,
//
// Store multiple
- InstrItinData<IIC_iStorem , [InstrStage<3, [V6_Pipe]>]>,
-
+ InstrItinData<IIC_iStore_m , [InstrStage<3, [V6_Pipe]>]>,
+ //
+ // Store multiple + update
+ InstrItinData<IIC_iStore_mu , [InstrStage<3, [V6_Pipe]>], [2]>,
+
// Branch
//
// no delay slots, so the latency of a branch is unimportant
// Double-precision FP MAC
InstrItinData<IIC_fpMAC64 , [InstrStage<2, [V6_Pipe]>], [9, 2, 2, 2]>,
//
+ // Single-precision Fused FP MAC
+ InstrItinData<IIC_fpFMAC32, [InstrStage<1, [V6_Pipe]>], [9, 2, 2, 2]>,
+ //
+ // Double-precision Fused FP MAC
+ InstrItinData<IIC_fpFMAC64, [InstrStage<2, [V6_Pipe]>], [9, 2, 2, 2]>,
+ //
// Single-precision FP DIV
InstrItinData<IIC_fpDIV32 , [InstrStage<15, [V6_Pipe]>], [20, 2, 2]>,
//
// Double-precision FP SQRT
InstrItinData<IIC_fpSQRT64 , [InstrStage<29, [V6_Pipe]>], [34, 2, 2]>,
//
+ // Integer to Single-precision Move
+ InstrItinData<IIC_fpMOVIS, [InstrStage<1, [V6_Pipe]>], [10, 1]>,
+ //
+ // Integer to Double-precision Move
+ InstrItinData<IIC_fpMOVID, [InstrStage<1, [V6_Pipe]>], [10, 1, 1]>,
+ //
+ // Single-precision to Integer Move
+ InstrItinData<IIC_fpMOVSI, [InstrStage<1, [V6_Pipe]>], [10, 1]>,
+ //
+ // Double-precision to Integer Move
+ InstrItinData<IIC_fpMOVDI, [InstrStage<1, [V6_Pipe]>], [10, 10, 1]>,
+ //
// Single-precision FP Load
InstrItinData<IIC_fpLoad32 , [InstrStage<1, [V6_Pipe]>], [5, 2, 2]>,
//
InstrItinData<IIC_fpLoad64 , [InstrStage<1, [V6_Pipe]>], [5, 2, 2]>,
//
// FP Load Multiple
- InstrItinData<IIC_fpLoadm , [InstrStage<3, [V6_Pipe]>]>,
+ InstrItinData<IIC_fpLoad_m , [InstrStage<3, [V6_Pipe]>], [2, 1, 1, 5]>,
+ //
+ // FP Load Multiple + update
+ InstrItinData<IIC_fpLoad_mu, [InstrStage<3, [V6_Pipe]>], [3, 2, 1, 1, 5]>,
//
// Single-precision FP Store
InstrItinData<IIC_fpStore32 , [InstrStage<1, [V6_Pipe]>], [2, 2, 2]>,
InstrItinData<IIC_fpStore64 , [InstrStage<1, [V6_Pipe]>], [2, 2, 2]>,
//
// FP Store Multiple
- InstrItinData<IIC_fpStorem , [InstrStage<3, [V6_Pipe]>]>
+ InstrItinData<IIC_fpStore_m, [InstrStage<3, [V6_Pipe]>], [2, 2, 2, 2]>,
+ //
+ // FP Store Multiple + update
+ InstrItinData<IIC_fpStore_mu,[InstrStage<3, [V6_Pipe]>], [3, 2, 2, 2, 2]>
]>;