maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/llvm/test/CodeGen/PowerPC/float-to-int.ll

172 lines
3.6 KiB
LLVM
Raw Normal View History

[Power9] Exploit D-Form VSX Scalar memory ops that target full VSX register set This patch corresponds to review: The newly added VSX D-Form (register + offset) memory ops target the upper half of the VSX register set. The existing ones target the lower half. In order to unify these and have the ability to target all the VSX registers using D-Form operations, this patch defines Pseudo-ops for the loads/stores which are expanded post-RA. The expansion then choses the correct opcode based on the register that was allocated for the operation. llvm-svn: 283212
2016-10-04 19:25:52 +08:00
; RUN: llc -verify-machineinstrs < %s -mtriple=powerpc64-unknown-linux-gnu \
; RUN: -mcpu=a2 | FileCheck %s
; RUN: llc -verify-machineinstrs < %s -mtriple=powerpc64-unknown-linux-gnu \
; RUN: -mcpu=pwr7 -mattr=+vsx | FileCheck -check-prefix=CHECK-VSX %s
; RUN: llc -verify-machineinstrs < %s -mtriple=powerpc64-unknown-linux-gnu \
; RUN: -mcpu=g5
; RUN: llc -verify-machineinstrs < %s -mtriple=powerpc64-unknown-linux-gnu \
; RUN: -mcpu=pwr9 -mattr=-direct-move | FileCheck -check-prefix=CHECK-P9 %s
Add more PPC floating-point conversion instructions The P7 and A2 have additional floating-point conversion instructions which allow a direct two-instruction sequence (plus load/store) to convert from all combinations (signed/unsigned i32/i64) <--> (float/double) (on previous cores, only some combinations were directly available). llvm-svn: 178480
2013-04-02 01:52:07 +08:00
target datalayout = "E-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-f128:128:128-v128:128:128-n32:64"
target triple = "powerpc64-unknown-linux-gnu"
define i64 @foo(float %a) nounwind {
%x = fptosi float %a to i64
ret i64 %x
; CHECK: @foo
; CHECK: fctidz [[REG:[0-9]+]], 1
; CHECK: stfd [[REG]],
; CHECK: ld 3,
; CHECK: blr
[PowerPC] Make use of VSX f64 <-> i64 conversion instructions When VSX is available, these instructions should be used in preference to the older variants that only have access to the scalar floating-point registers. llvm-svn: 204559
2014-03-23 13:35:00 +08:00
; CHECK-VSX: @foo
; CHECK-VSX: xscvdpsxds [[REG:[0-9]+]], 1
; CHECK-VSX: stxsdx [[REG]],
; CHECK-VSX: ld 3,
; CHECK-VSX: blr
[Power9] Exploit D-Form VSX Scalar memory ops that target full VSX register set This patch corresponds to review: The newly added VSX D-Form (register + offset) memory ops target the upper half of the VSX register set. The existing ones target the lower half. In order to unify these and have the ability to target all the VSX registers using D-Form operations, this patch defines Pseudo-ops for the loads/stores which are expanded post-RA. The expansion then choses the correct opcode based on the register that was allocated for the operation. llvm-svn: 283212
2016-10-04 19:25:52 +08:00
; CHECK-LABEL-P9: @foo
; CHECK-P9: xscvdpsxds [[REG:[0-9]+]], 1
; CHECK-P9: stfd [[REG]],
; CHECK-P9: ld 3,
; CHECK-P9: blr
Add more PPC floating-point conversion instructions The P7 and A2 have additional floating-point conversion instructions which allow a direct two-instruction sequence (plus load/store) to convert from all combinations (signed/unsigned i32/i64) <--> (float/double) (on previous cores, only some combinations were directly available). llvm-svn: 178480
2013-04-02 01:52:07 +08:00
}
define i64 @foo2(double %a) nounwind {
%x = fptosi double %a to i64
ret i64 %x
; CHECK: @foo2
; CHECK: fctidz [[REG:[0-9]+]], 1
; CHECK: stfd [[REG]],
; CHECK: ld 3,
; CHECK: blr
[PowerPC] Make use of VSX f64 <-> i64 conversion instructions When VSX is available, these instructions should be used in preference to the older variants that only have access to the scalar floating-point registers. llvm-svn: 204559
2014-03-23 13:35:00 +08:00
; CHECK-VSX: @foo2
; CHECK-VSX: xscvdpsxds [[REG:[0-9]+]], 1
; CHECK-VSX: stxsdx [[REG]],
; CHECK-VSX: ld 3,
; CHECK-VSX: blr
[Power9] Exploit D-Form VSX Scalar memory ops that target full VSX register set This patch corresponds to review: The newly added VSX D-Form (register + offset) memory ops target the upper half of the VSX register set. The existing ones target the lower half. In order to unify these and have the ability to target all the VSX registers using D-Form operations, this patch defines Pseudo-ops for the loads/stores which are expanded post-RA. The expansion then choses the correct opcode based on the register that was allocated for the operation. llvm-svn: 283212
2016-10-04 19:25:52 +08:00
; CHECK-LABEL-P9: @foo2
; CHECK-P9: xscvdpsxds [[REG:[0-9]+]], 1
; CHECK-P9: stfd [[REG]],
; CHECK-P9: ld 3,
; CHECK-P9: blr
Add more PPC floating-point conversion instructions The P7 and A2 have additional floating-point conversion instructions which allow a direct two-instruction sequence (plus load/store) to convert from all combinations (signed/unsigned i32/i64) <--> (float/double) (on previous cores, only some combinations were directly available). llvm-svn: 178480
2013-04-02 01:52:07 +08:00
}
define i64 @foo3(float %a) nounwind {
%x = fptoui float %a to i64
ret i64 %x
; CHECK: @foo3
; CHECK: fctiduz [[REG:[0-9]+]], 1
; CHECK: stfd [[REG]],
; CHECK: ld 3,
; CHECK: blr
[PowerPC] Make use of VSX f64 <-> i64 conversion instructions When VSX is available, these instructions should be used in preference to the older variants that only have access to the scalar floating-point registers. llvm-svn: 204559
2014-03-23 13:35:00 +08:00
; CHECK-VSX: @foo3
; CHECK-VSX: xscvdpuxds [[REG:[0-9]+]], 1
; CHECK-VSX: stxsdx [[REG]],
; CHECK-VSX: ld 3,
; CHECK-VSX: blr
[Power9] Exploit D-Form VSX Scalar memory ops that target full VSX register set This patch corresponds to review: The newly added VSX D-Form (register + offset) memory ops target the upper half of the VSX register set. The existing ones target the lower half. In order to unify these and have the ability to target all the VSX registers using D-Form operations, this patch defines Pseudo-ops for the loads/stores which are expanded post-RA. The expansion then choses the correct opcode based on the register that was allocated for the operation. llvm-svn: 283212
2016-10-04 19:25:52 +08:00
; CHECK-LABEL-P9: @foo3
; CHECK-P9: xscvdpuxds [[REG:[0-9]+]], 1
; CHECK-P9: stfd [[REG]],
; CHECK-P9: ld 3,
; CHECK-P9: blr
Add more PPC floating-point conversion instructions The P7 and A2 have additional floating-point conversion instructions which allow a direct two-instruction sequence (plus load/store) to convert from all combinations (signed/unsigned i32/i64) <--> (float/double) (on previous cores, only some combinations were directly available). llvm-svn: 178480
2013-04-02 01:52:07 +08:00
}
define i64 @foo4(double %a) nounwind {
%x = fptoui double %a to i64
ret i64 %x
; CHECK: @foo4
; CHECK: fctiduz [[REG:[0-9]+]], 1
; CHECK: stfd [[REG]],
; CHECK: ld 3,
; CHECK: blr
[PowerPC] Make use of VSX f64 <-> i64 conversion instructions When VSX is available, these instructions should be used in preference to the older variants that only have access to the scalar floating-point registers. llvm-svn: 204559
2014-03-23 13:35:00 +08:00
; CHECK-VSX: @foo4
; CHECK-VSX: xscvdpuxds [[REG:[0-9]+]], 1
; CHECK-VSX: stxsdx [[REG]],
; CHECK-VSX: ld 3,
; CHECK-VSX: blr
[Power9] Exploit D-Form VSX Scalar memory ops that target full VSX register set This patch corresponds to review: The newly added VSX D-Form (register + offset) memory ops target the upper half of the VSX register set. The existing ones target the lower half. In order to unify these and have the ability to target all the VSX registers using D-Form operations, this patch defines Pseudo-ops for the loads/stores which are expanded post-RA. The expansion then choses the correct opcode based on the register that was allocated for the operation. llvm-svn: 283212
2016-10-04 19:25:52 +08:00
; CHECK-LABEL-P9: @foo4
; CHECK-P9: xscvdpuxds [[REG:[0-9]+]], 1
; CHECK-P9: stfd [[REG]],
; CHECK-P9: ld 3,
; CHECK-P9: blr
Add more PPC floating-point conversion instructions The P7 and A2 have additional floating-point conversion instructions which allow a direct two-instruction sequence (plus load/store) to convert from all combinations (signed/unsigned i32/i64) <--> (float/double) (on previous cores, only some combinations were directly available). llvm-svn: 178480
2013-04-02 01:52:07 +08:00
}
define i32 @goo(float %a) nounwind {
%x = fptosi float %a to i32
ret i32 %x
; CHECK: @goo
; CHECK: fctiwz [[REG:[0-9]+]], 1
; CHECK: stfiwx [[REG]],
; CHECK: lwz 3,
; CHECK: blr
[PowerPC] Make use of VSX f64 <-> i64 conversion instructions When VSX is available, these instructions should be used in preference to the older variants that only have access to the scalar floating-point registers. llvm-svn: 204559
2014-03-23 13:35:00 +08:00
; CHECK-VSX: @goo
; CHECK-VSX: xscvdpsxws [[REG:[0-9]+]], 1
; CHECK-VSX: stfiwx [[REG]],
; CHECK-VSX: lwz 3,
; CHECK-VSX: blr
Add more PPC floating-point conversion instructions The P7 and A2 have additional floating-point conversion instructions which allow a direct two-instruction sequence (plus load/store) to convert from all combinations (signed/unsigned i32/i64) <--> (float/double) (on previous cores, only some combinations were directly available). llvm-svn: 178480
2013-04-02 01:52:07 +08:00
}
define i32 @goo2(double %a) nounwind {
%x = fptosi double %a to i32
ret i32 %x
; CHECK: @goo2
; CHECK: fctiwz [[REG:[0-9]+]], 1
; CHECK: stfiwx [[REG]],
; CHECK: lwz 3,
; CHECK: blr
[PowerPC] Make use of VSX f64 <-> i64 conversion instructions When VSX is available, these instructions should be used in preference to the older variants that only have access to the scalar floating-point registers. llvm-svn: 204559
2014-03-23 13:35:00 +08:00
; CHECK-VSX: @goo2
; CHECK-VSX: xscvdpsxws [[REG:[0-9]+]], 1
; CHECK-VSX: stfiwx [[REG]],
; CHECK-VSX: lwz 3,
; CHECK-VSX: blr
Add more PPC floating-point conversion instructions The P7 and A2 have additional floating-point conversion instructions which allow a direct two-instruction sequence (plus load/store) to convert from all combinations (signed/unsigned i32/i64) <--> (float/double) (on previous cores, only some combinations were directly available). llvm-svn: 178480
2013-04-02 01:52:07 +08:00
}
define i32 @goo3(float %a) nounwind {
%x = fptoui float %a to i32
ret i32 %x
; CHECK: @goo3
; CHECK: fctiwuz [[REG:[0-9]+]], 1
; CHECK: stfiwx [[REG]],
; CHECK: lwz 3,
; CHECK: blr
[PowerPC] Make use of VSX f64 <-> i64 conversion instructions When VSX is available, these instructions should be used in preference to the older variants that only have access to the scalar floating-point registers. llvm-svn: 204559
2014-03-23 13:35:00 +08:00
; CHECK-VSX: @goo3
; CHECK-VSX: xscvdpuxws [[REG:[0-9]+]], 1
; CHECK-VSX: stfiwx [[REG]],
; CHECK-VSX: lwz 3,
; CHECK-VSX: blr
Add more PPC floating-point conversion instructions The P7 and A2 have additional floating-point conversion instructions which allow a direct two-instruction sequence (plus load/store) to convert from all combinations (signed/unsigned i32/i64) <--> (float/double) (on previous cores, only some combinations were directly available). llvm-svn: 178480
2013-04-02 01:52:07 +08:00
}
define i32 @goo4(double %a) nounwind {
%x = fptoui double %a to i32
ret i32 %x
; CHECK: @goo4
; CHECK: fctiwuz [[REG:[0-9]+]], 1
; CHECK: stfiwx [[REG]],
; CHECK: lwz 3,
; CHECK: blr
[PowerPC] Make use of VSX f64 <-> i64 conversion instructions When VSX is available, these instructions should be used in preference to the older variants that only have access to the scalar floating-point registers. llvm-svn: 204559
2014-03-23 13:35:00 +08:00
; CHECK-VSX: @goo4
; CHECK-VSX: xscvdpuxws [[REG:[0-9]+]], 1
; CHECK-VSX: stfiwx [[REG]],
; CHECK-VSX: lwz 3,
; CHECK-VSX: blr
Add more PPC floating-point conversion instructions The P7 and A2 have additional floating-point conversion instructions which allow a direct two-instruction sequence (plus load/store) to convert from all combinations (signed/unsigned i32/i64) <--> (float/double) (on previous cores, only some combinations were directly available). llvm-svn: 178480
2013-04-02 01:52:07 +08:00
}