RPCS3/llvm-mirror
1
0
Fork 0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-10-19 11:02:59 +02:00
Code Issues Projects Releases Wiki Activity

[PowerPC] Better scheduling for isel on P7/P8

Browse Source 
isel is actually a cracked instruction on the P7/P8, and must start a dispatch
group. The scheduling model should reflect this so that we don't bunch too many
of them together when possible.

Thanks to Bill Schmidt and Pat Haugen for helping to sort this out.

llvm-svn: 227758
This commit is contained in:
Hal Finkel 2015-02-01 17:52:16 +00:00
parent 83140dfa69
commit 4a7dffd074
10 changed files with 67 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
lib/Target/PowerPC/PPCInstr64Bit.td
View File

@ -686,7 +686,7 @@ defm RLWIMI8 : MForm_2r<20, (outs g8rc:$rA),
let isSelect = 1 in
def ISEL8 : AForm_4<31, 15,
(outs g8rc:$rT), (ins g8rc_nox0:$rA, g8rc:$rB, crbitrc:$cond),
"isel $rT, $rA, $rB, $cond", IIC_IntGeneral,
"isel $rT, $rA, $rB, $cond", IIC_IntISEL,
[]>;
} // Interpretation64Bit
} // hasSideEffects = 0

2
lib/Target/PowerPC/PPCInstrInfo.td
View File

@ -2377,7 +2377,7 @@ let PPC970_Unit = 1 in { // FXU Operations.
let isSelect = 1 in
def ISEL : AForm_4<31, 15,
(outs gprc:$rT), (ins gprc_nor0:$rA, gprc:$rB, crbitrc:$cond),
"isel $rT, $rA, $rB, $cond", IIC_IntGeneral,
"isel $rT, $rA, $rB, $cond", IIC_IntISEL,
[]>;
}

2
lib/Target/PowerPC/PPCSchedule.td
View File

@ -13,6 +13,7 @@
def IIC_IntSimple : InstrItinClass;
def IIC_IntGeneral : InstrItinClass;
def IIC_IntCompare : InstrItinClass;
def IIC_IntISEL : InstrItinClass;
def IIC_IntDivD : InstrItinClass;
def IIC_IntDivW : InstrItinClass;
def IIC_IntMFFS : InstrItinClass;
@ -217,6 +218,7 @@ include "PPCScheduleE5500.td"
// fsub IIC_FPAddSub
// fsubs IIC_FPGeneral
// icbi IIC_LdStICBI
// isel IIC_IntISEL
// isync IIC_SprISYNC
// lbz IIC_LdStLoad
// lbzu IIC_LdStLoadUpd

8
lib/Target/PowerPC/PPCSchedule440.td
View File

@ -121,6 +121,14 @@ def PPC440Itineraries : ProcessorItineraries<
[2, 0, 0],
[P440_GPR_Bypass,
P440_GPR_Bypass, P440_GPR_Bypass]>,
InstrItinData<IIC_IntISEL, [InstrStage<1, [P440_DISS1, P440_DISS2]>,
InstrStage<1, [P440_IRACC, P440_LRACC]>,
InstrStage<1, [P440_IEXE1, P440_JEXE1]>,
InstrStage<1, [P440_IEXE2, P440_JEXE2]>,
InstrStage<1, [P440_IWB, P440_JWB]>],
[2, 0, 0, 0],
[P440_GPR_Bypass,
P440_GPR_Bypass, P440_GPR_Bypass, NoBypass]>,
InstrItinData<IIC_IntCompare, [InstrStage<1, [P440_DISS1, P440_DISS2]>,
InstrStage<1, [P440_IRACC, P440_LRACC]>,
InstrStage<1, [P440_IEXE1, P440_JEXE1]>,

2
lib/Target/PowerPC/PPCScheduleA2.td
View File

@ -29,6 +29,8 @@ def PPCA2Itineraries : ProcessorItineraries<
[1, 0, 0]>,
InstrItinData<IIC_IntGeneral, [InstrStage<1, [A2_XU]>],
[2, 0, 0]>,
InstrItinData<IIC_IntISEL, [InstrStage<1, [A2_XU]>],
[2, 0, 0, 0]>,
InstrItinData<IIC_IntCompare, [InstrStage<1, [A2_XU]>],
[2, 0, 0]>,
InstrItinData<IIC_IntDivW, [InstrStage<1, [A2_XU]>],

6
lib/Target/PowerPC/PPCScheduleE500mc.td
View File

@ -54,6 +54,12 @@ def PPCE500mcItineraries : ProcessorItineraries<
[4, 1, 1], // Latency = 1
[E500_GPR_Bypass,
E500_GPR_Bypass, E500_GPR_Bypass]>,
InstrItinData<IIC_IntISEL, [InstrStage<1, [E500_DIS0, E500_DIS1], 0>,
InstrStage<1, [E500_SFX0, E500_SFX1]>],
[4, 1, 1, 1], // Latency = 1
[E500_GPR_Bypass,
E500_GPR_Bypass, E500_GPR_Bypass,
E500_CR_Bypass]>,
InstrItinData<IIC_IntCompare, [InstrStage<1, [E500_DIS0, E500_DIS1], 0>,
InstrStage<1, [E500_SFX0, E500_SFX1]>],
[5, 1, 1], // Latency = 1 or 2

6
lib/Target/PowerPC/PPCScheduleE5500.td
View File

@ -58,6 +58,12 @@ def PPCE5500Itineraries : ProcessorItineraries<
[5, 2, 2], // Latency = 1
[E5500_GPR_Bypass,
E5500_GPR_Bypass, E5500_GPR_Bypass]>,
InstrItinData<IIC_IntISEL, [InstrStage<1, [E5500_DIS0, E5500_DIS1], 0>,
InstrStage<1, [E5500_SFX0, E5500_SFX1]>],
[5, 2, 2, 2], // Latency = 1
[E5500_GPR_Bypass,
E5500_GPR_Bypass, E5500_GPR_Bypass,
E5500_CR_Bypass]>,
InstrItinData<IIC_IntCompare, [InstrStage<1, [E5500_DIS0, E5500_DIS1], 0>,
InstrStage<1, [E5500_SFX0, E5500_SFX1]>],
[6, 2, 2], // Latency = 1 or 2

4
lib/Target/PowerPC/PPCScheduleP7.td
View File

@ -89,6 +89,10 @@ def P7Itineraries : ProcessorItineraries<
P7_DU3, P7_DU4], 0>,
InstrStage<1, [P7_FX1, P7_FX2]>],
[1, 1, 1]>,
InstrItinData<IIC_IntISEL, [InstrStage<1, [P7_DU1], 0>,
InstrStage<1, [P7_FX1, P7_FX2], 0>,
InstrStage<1, [P7_BRU]>],
[1, 1, 1, 1]>,
InstrItinData<IIC_IntCompare , [InstrStage<1, [P7_DU1, P7_DU2,
P7_DU3, P7_DU4], 0>,
InstrStage<1, [P7_FX1, P7_FX2]>],

4
lib/Target/PowerPC/PPCScheduleP8.td
View File

@ -66,6 +66,10 @@ def P8Itineraries : ProcessorItineraries<
InstrStage<1, [P8_FXU1, P8_FXU2, P8_LU1,
P8_LU2, P8_LSU1, P8_LSU2]>],
[1, 1, 1]>,
InstrItinData<IIC_IntISEL, [InstrStage<1, [P8_DU1], 0>,
InstrStage<1, [P8_FXU1, P8_FXU2], 0>,
InstrStage<1, [P8_BRU]>],
[1, 1, 1, 1]>,
InstrItinData<IIC_IntCompare , [InstrStage<1, [P8_DU1, P8_DU2, P8_DU3,
P8_DU4, P8_DU5, P8_DU6], 0>,
InstrStage<1, [P8_FXU1, P8_FXU2]>],

33
test/CodeGen/PowerPC/p8-isel-sched.ll Normal file
View File

@ -0,0 +1,33 @@
; RUN: llc -mcpu=pwr8 < %s | FileCheck %s
target datalayout = "E-m:e-i64:64-n32:64"
target triple = "powerpc64-unknown-linux-gnu"
; Function Attrs: nounwind
define void @foo(i32* nocapture %r1, i32* nocapture %r2, i32* nocapture %r3, i32* nocapture %r4, i32 signext %a, i32 signext %b, i32 signext %c, i32 signext %d) #0 {
entry:
%tobool = icmp ne i32 %a, 0
%cond = select i1 %tobool, i32 %b, i32 %c
store i32 %cond, i32* %r1, align 4
%cond5 = select i1 %tobool, i32 %b, i32 %d
store i32 %cond5, i32* %r2, align 4
%add = add nsw i32 %b, 1
%sub = add nsw i32 %d, -2
%cond10 = select i1 %tobool, i32 %add, i32 %sub
store i32 %cond10, i32* %r3, align 4
%add13 = add nsw i32 %b, 3
%sub15 = add nsw i32 %d, -5
%cond17 = select i1 %tobool, i32 %add13, i32 %sub15
store i32 %cond17, i32* %r4, align 4
ret void
}
; Make sure that we don't schedule all of the isels together, they should be
; intermixed with the adds because each isel starts a new dispatch group.
; CHECK-LABEL: @foo
; CHECK: isel
; CHECK: addi
; CHECK: isel
; CHECK: blr
attributes #0 = { nounwind }