OSchip
/
llvm-project
mirror of https://github.com/THU-DSP-LAB/llvm-project.git
13
0
Fork 6
Code Issues Packages Projects Releases Wiki Activity

Enable SandyBridgeModel for all modern Intel P6 descendants. 

All Intel CPUs since Yonah look a lot alike, at least at the granularity
of the scheduling models. We can add more accurate models for
processors that aren't Sandy Bridge if required. Haswell will probably
need its own.

The Atom processor and anything based on NetBurst is completely
different. So are the non-Intel chips.

llvm-svn: 178080
This commit is contained in:
Jakob Stoklund Olesen 2013-03-26 22:19:12 +00:00
parent 2ebf08ecbd
commit 1ac7e662d4
4 changed files with 47 additions and 37 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

77
llvm/lib/Target/X86/X86.td
View File

@ -145,9 +145,6 @@ def ProcIntelAtom : SubtargetFeature<"atom", "X86ProcFamily", "IntelAtom",
class Proc<string Name, list<SubtargetFeature> Features>
: ProcessorModel<Name, GenericModel, Features>;
class AtomProc<string Name, list<SubtargetFeature> Features>
: ProcessorModel<Name, AtomModel, Features>;
def : Proc<"generic", []>;
def : Proc<"i386", []>;
def : Proc<"i486", []>;
@ -164,46 +161,58 @@ def : Proc<"pentium4", [FeatureSSE2]>;
def : Proc<"pentium4m", [FeatureSSE2, FeatureSlowBTMem]>;
def : Proc<"x86-64", [FeatureSSE2, Feature64Bit, FeatureSlowBTMem,
FeatureFastUAMem]>;
def : Proc<"yonah", [FeatureSSE3, FeatureSlowBTMem]>;
def : Proc<"prescott", [FeatureSSE3, FeatureSlowBTMem]>;
def : Proc<"nocona", [FeatureSSE3, FeatureCMPXCHG16B,
FeatureSlowBTMem]>;
def : Proc<"core2", [FeatureSSSE3, FeatureCMPXCHG16B,
FeatureSlowBTMem]>;
def : Proc<"penryn", [FeatureSSE41, FeatureCMPXCHG16B,
FeatureSlowBTMem]>;
def : AtomProc<"atom", [ProcIntelAtom, FeatureSSSE3, FeatureCMPXCHG16B,
FeatureMOVBE, FeatureSlowBTMem, FeatureLeaForSP,
FeatureSlowDivide, FeaturePadShortFunctions]>;
// Intel Core Duo.
def : ProcessorModel<"yonah", SandyBridgeModel,
[FeatureSSE3, FeatureSlowBTMem]>;
// NetBurst.
def : Proc<"prescott", [FeatureSSE3, FeatureSlowBTMem]>;
def : Proc<"nocona", [FeatureSSE3, FeatureCMPXCHG16B, FeatureSlowBTMem]>;
// Intel Core 2 Solo/Duo.
def : ProcessorModel<"core2", SandyBridgeModel,
[FeatureSSSE3, FeatureCMPXCHG16B, FeatureSlowBTMem]>;
def : ProcessorModel<"penryn", SandyBridgeModel,
[FeatureSSE41, FeatureCMPXCHG16B, FeatureSlowBTMem]>;
// Atom.
def : ProcessorModel<"atom", AtomModel,
[ProcIntelAtom, FeatureSSSE3, FeatureCMPXCHG16B,
FeatureMOVBE, FeatureSlowBTMem, FeatureLeaForSP,
FeatureSlowDivide, FeaturePadShortFunctions]>;
// "Arrandale" along with corei3 and corei5
def : Proc<"corei7", [FeatureSSE42, FeatureCMPXCHG16B,
FeatureSlowBTMem, FeatureFastUAMem,
FeaturePOPCNT, FeatureAES]>;
def : Proc<"nehalem", [FeatureSSE42, FeatureCMPXCHG16B,
FeatureSlowBTMem, FeatureFastUAMem,
FeaturePOPCNT]>;
def : ProcessorModel<"corei7", SandyBridgeModel,
[FeatureSSE42, FeatureCMPXCHG16B, FeatureSlowBTMem,
FeatureFastUAMem, FeaturePOPCNT, FeatureAES]>;
def : ProcessorModel<"nehalem", SandyBridgeModel,
[FeatureSSE42, FeatureCMPXCHG16B, FeatureSlowBTMem,
FeatureFastUAMem, FeaturePOPCNT]>;
// Westmere is a similar machine to nehalem with some additional features.
// Westmere is the corei3/i5/i7 path from nehalem to sandybridge
def : Proc<"westmere", [FeatureSSE42, FeatureCMPXCHG16B,
FeatureSlowBTMem, FeatureFastUAMem,
FeaturePOPCNT, FeatureAES, FeaturePCLMUL]>;
def : ProcessorModel<"westmere", SandyBridgeModel,
[FeatureSSE42, FeatureCMPXCHG16B, FeatureSlowBTMem,
FeatureFastUAMem, FeaturePOPCNT, FeatureAES,
FeaturePCLMUL]>;
// Sandy Bridge
// SSE is not listed here since llvm treats AVX as a reimplementation of SSE,
// rather than a superset.
def : Proc<"corei7-avx", [FeatureAVX, FeatureCMPXCHG16B, FeatureFastUAMem,
FeaturePOPCNT, FeatureAES, FeaturePCLMUL]>;
def : ProcessorModel<"corei7-avx", SandyBridgeModel,
[FeatureAVX, FeatureCMPXCHG16B, FeatureFastUAMem,
FeaturePOPCNT, FeatureAES, FeaturePCLMUL]>;
// Ivy Bridge
def : Proc<"core-avx-i", [FeatureAVX, FeatureCMPXCHG16B, FeatureFastUAMem,
FeaturePOPCNT, FeatureAES, FeaturePCLMUL,
FeatureRDRAND, FeatureF16C, FeatureFSGSBase]>;
def : ProcessorModel<"core-avx-i", SandyBridgeModel,
[FeatureAVX, FeatureCMPXCHG16B, FeatureFastUAMem,
FeaturePOPCNT, FeatureAES, FeaturePCLMUL, FeatureRDRAND,
FeatureF16C, FeatureFSGSBase]>;
// Haswell
def : Proc<"core-avx2", [FeatureAVX2, FeatureCMPXCHG16B, FeatureFastUAMem,
FeaturePOPCNT, FeatureAES, FeaturePCLMUL,
FeatureRDRAND, FeatureF16C, FeatureFSGSBase,
FeatureMOVBE, FeatureLZCNT, FeatureBMI,
FeatureBMI2, FeatureFMA,
FeatureRTM]>;
def : ProcessorModel<"core-avx2", SandyBridgeModel,
[FeatureAVX2, FeatureCMPXCHG16B, FeatureFastUAMem,
FeaturePOPCNT, FeatureAES, FeaturePCLMUL, FeatureRDRAND,
FeatureF16C, FeatureFSGSBase, FeatureMOVBE, FeatureLZCNT,
FeatureBMI, FeatureBMI2, FeatureFMA, FeatureRTM]>;
def : Proc<"k6", [FeatureMMX]>;
def : Proc<"k6-2", [Feature3DNow]>;

1
llvm/lib/Target/X86/X86Schedule.td
View File

@ -565,3 +565,4 @@ def GenericModel : SchedMachineModel {
}
include "X86ScheduleAtom.td"
include "X86SchedSandyBridge.td"

4
llvm/test/CodeGen/X86/misched-ilp.ll
View File

@ -1,5 +1,5 @@
; RUN: llc < %s -mtriple=x86_64-apple-macosx -mcpu=core2 -enable-misched -misched=ilpmax | FileCheck -check-prefix=MAX %s
; RUN: llc < %s -mtriple=x86_64-apple-macosx -mcpu=core2 -enable-misched -misched=ilpmin | FileCheck -check-prefix=MIN %s
; RUN: llc < %s -mtriple=x86_64-apple-macosx -mcpu=nocona -enable-misched -misched=ilpmax | FileCheck -check-prefix=MAX %s
; RUN: llc < %s -mtriple=x86_64-apple-macosx -mcpu=nocona -enable-misched -misched=ilpmin | FileCheck -check-prefix=MIN %s
;
; Basic verification of the ScheduleDAGILP metric.
;

2
llvm/test/CodeGen/X86/sink-hoist.ll
View File

@ -1,4 +1,4 @@
; RUN: llc < %s -march=x86-64 -asm-verbose=false -mtriple=x86_64-unknown-linux-gnu -mcpu=nehalem -post-RA-scheduler=true | FileCheck %s
; RUN: llc < %s -march=x86-64 -asm-verbose=false -mtriple=x86_64-unknown-linux-gnu -mcpu=nehalem -post-RA-scheduler=true -schedmodel=false | FileCheck %s
; Currently, floating-point selects are lowered to CFG triangles.
; This means that one side of the select is always unconditionally