RPCS3/llvm-mirror
1
0
Fork 0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-10-29 23:12:55 +01:00
Code Issues Projects Releases Wiki Activity
aa04918ad4
llvm-mirror/test/CodeGen/Mips/fpneeded.ll

150 lines
2.6 KiB
LLVM
Raw Normal View History

This is for an experimental option -mips-os16. The idea is to compile all Mips32 code as Mips16 unless it can't be compiled as Mips 16. For now this would happen as long as floating point instructions are not needed. Probably it would also make sense to compile as mips32 if atomic operations are needed too. There may be other cases too. A module pass prescans the IR and adds the mips16 or nomips16 attribute to functions depending on the functions needs. Mips 16 mode can result in a 40% code compression by utililizing 16 bit encoding of many instructions. The hope is for this to replace the traditional gcc way of dealing with Mips16 code using floating point which involves essentially using soft float but with a library implemented using mips32 floating point. This gcc method also requires creating stubs so that Mips32 code can interact with these Mips 16 functions that have floating point needs. My conjecture is that in reality this traditional gcc method would never win over this new method. I will be implementing the traditional gcc method also. Some of it is already done but I needed to do the stubs to finish the work and those required this mips16/32 mixed mode capability. I have more ideas for to make this new method much better and I think the old method will just live in llvm for anyone that needs the backward compatibility but I don't for what reason that would be needed. llvm-svn: 179185
2013-04-10 18:58:04 +02:00
; RUN: llc -march=mipsel -mcpu=mips32 -relocation-model=static -O3 < %s -mips-os16 | FileCheck %s -check-prefix=32
@x = global float 1.000000e+00, align 4
@y = global float 2.000000e+00, align 4
@zz = common global float 0.000000e+00, align 4
@z = common global float 0.000000e+00, align 4
define float @fv() #0 {
entry:
ret float 1.000000e+00
}
Fix a low hanging use of hasRawTextSupport. This also fixes the placement of the function label comment. It was being placed next to the mips16 directive instead of next to the label. llvm-svn: 199245
2014-01-14 19:57:12 +01:00
; 32: .set nomips16
This is for an experimental option -mips-os16. The idea is to compile all Mips32 code as Mips16 unless it can't be compiled as Mips 16. For now this would happen as long as floating point instructions are not needed. Probably it would also make sense to compile as mips32 if atomic operations are needed too. There may be other cases too. A module pass prescans the IR and adds the mips16 or nomips16 attribute to functions depending on the functions needs. Mips 16 mode can result in a 40% code compression by utililizing 16 bit encoding of many instructions. The hope is for this to replace the traditional gcc way of dealing with Mips16 code using floating point which involves essentially using soft float but with a library implemented using mips32 floating point. This gcc method also requires creating stubs so that Mips32 code can interact with these Mips 16 functions that have floating point needs. My conjecture is that in reality this traditional gcc method would never win over this new method. I will be implementing the traditional gcc method also. Some of it is already done but I needed to do the stubs to finish the work and those required this mips16/32 mixed mode capability. I have more ideas for to make this new method much better and I think the old method will just live in llvm for anyone that needs the backward compatibility but I don't for what reason that would be needed. llvm-svn: 179185
2013-04-10 18:58:04 +02:00
; 32: .ent fv
; 32: .set noreorder
; 32: .set nomacro
; 32: .set noat
; 32: jr $ra
; 32: .set at
; 32: .set macro
; 32: .set reorder
; 32: .end fv
define double @dv() #0 {
entry:
ret double 2.000000e+00
}
Fix a low hanging use of hasRawTextSupport. This also fixes the placement of the function label comment. It was being placed next to the mips16 directive instead of next to the label. llvm-svn: 199245
2014-01-14 19:57:12 +01:00
; 32: .set nomips16
This is for an experimental option -mips-os16. The idea is to compile all Mips32 code as Mips16 unless it can't be compiled as Mips 16. For now this would happen as long as floating point instructions are not needed. Probably it would also make sense to compile as mips32 if atomic operations are needed too. There may be other cases too. A module pass prescans the IR and adds the mips16 or nomips16 attribute to functions depending on the functions needs. Mips 16 mode can result in a 40% code compression by utililizing 16 bit encoding of many instructions. The hope is for this to replace the traditional gcc way of dealing with Mips16 code using floating point which involves essentially using soft float but with a library implemented using mips32 floating point. This gcc method also requires creating stubs so that Mips32 code can interact with these Mips 16 functions that have floating point needs. My conjecture is that in reality this traditional gcc method would never win over this new method. I will be implementing the traditional gcc method also. Some of it is already done but I needed to do the stubs to finish the work and those required this mips16/32 mixed mode capability. I have more ideas for to make this new method much better and I think the old method will just live in llvm for anyone that needs the backward compatibility but I don't for what reason that would be needed. llvm-svn: 179185
2013-04-10 18:58:04 +02:00
; 32: .ent dv
; 32: .set noreorder
; 32: .set nomacro
; 32: .set noat
; 32: jr $ra
; 32: .set at
; 32: .set macro
; 32: .set reorder
; 32: .end dv
define void @vf(float %x) #0 {
entry:
%x.addr = alloca float, align 4
store float %x, float* %x.addr, align 4
ret void
}
Fix a low hanging use of hasRawTextSupport. This also fixes the placement of the function label comment. It was being placed next to the mips16 directive instead of next to the label. llvm-svn: 199245
2014-01-14 19:57:12 +01:00
; 32: .set nomips16
This is for an experimental option -mips-os16. The idea is to compile all Mips32 code as Mips16 unless it can't be compiled as Mips 16. For now this would happen as long as floating point instructions are not needed. Probably it would also make sense to compile as mips32 if atomic operations are needed too. There may be other cases too. A module pass prescans the IR and adds the mips16 or nomips16 attribute to functions depending on the functions needs. Mips 16 mode can result in a 40% code compression by utililizing 16 bit encoding of many instructions. The hope is for this to replace the traditional gcc way of dealing with Mips16 code using floating point which involves essentially using soft float but with a library implemented using mips32 floating point. This gcc method also requires creating stubs so that Mips32 code can interact with these Mips 16 functions that have floating point needs. My conjecture is that in reality this traditional gcc method would never win over this new method. I will be implementing the traditional gcc method also. Some of it is already done but I needed to do the stubs to finish the work and those required this mips16/32 mixed mode capability. I have more ideas for to make this new method much better and I think the old method will just live in llvm for anyone that needs the backward compatibility but I don't for what reason that would be needed. llvm-svn: 179185
2013-04-10 18:58:04 +02:00
; 32: .ent vf
; 32: .set noreorder
; 32: .set nomacro
; 32: .set noat
; 32: jr $ra
; 32: .set at
; 32: .set macro
; 32: .set reorder
; 32: .end vf
define void @vd(double %x) #0 {
entry:
%x.addr = alloca double, align 8
store double %x, double* %x.addr, align 8
ret void
}
Fix a low hanging use of hasRawTextSupport. This also fixes the placement of the function label comment. It was being placed next to the mips16 directive instead of next to the label. llvm-svn: 199245
2014-01-14 19:57:12 +01:00
; 32: .set nomips16
This is for an experimental option -mips-os16. The idea is to compile all Mips32 code as Mips16 unless it can't be compiled as Mips 16. For now this would happen as long as floating point instructions are not needed. Probably it would also make sense to compile as mips32 if atomic operations are needed too. There may be other cases too. A module pass prescans the IR and adds the mips16 or nomips16 attribute to functions depending on the functions needs. Mips 16 mode can result in a 40% code compression by utililizing 16 bit encoding of many instructions. The hope is for this to replace the traditional gcc way of dealing with Mips16 code using floating point which involves essentially using soft float but with a library implemented using mips32 floating point. This gcc method also requires creating stubs so that Mips32 code can interact with these Mips 16 functions that have floating point needs. My conjecture is that in reality this traditional gcc method would never win over this new method. I will be implementing the traditional gcc method also. Some of it is already done but I needed to do the stubs to finish the work and those required this mips16/32 mixed mode capability. I have more ideas for to make this new method much better and I think the old method will just live in llvm for anyone that needs the backward compatibility but I don't for what reason that would be needed. llvm-svn: 179185
2013-04-10 18:58:04 +02:00
; 32: .ent vd
; 32: .set noreorder
; 32: .set nomacro
; 32: .set noat
; 32: jr $ra
; 32: .set at
; 32: .set macro
; 32: .set reorder
; 32: .end vd
define void @foo1() #0 {
entry:
store float 1.000000e+00, float* @zz, align 4
%0 = load float* @y, align 4
%1 = load float* @x, align 4
%add = fadd float %0, %1
store float %add, float* @z, align 4
ret void
}
Fix a low hanging use of hasRawTextSupport. This also fixes the placement of the function label comment. It was being placed next to the mips16 directive instead of next to the label. llvm-svn: 199245
2014-01-14 19:57:12 +01:00
; 32: .set nomips16
This is for an experimental option -mips-os16. The idea is to compile all Mips32 code as Mips16 unless it can't be compiled as Mips 16. For now this would happen as long as floating point instructions are not needed. Probably it would also make sense to compile as mips32 if atomic operations are needed too. There may be other cases too. A module pass prescans the IR and adds the mips16 or nomips16 attribute to functions depending on the functions needs. Mips 16 mode can result in a 40% code compression by utililizing 16 bit encoding of many instructions. The hope is for this to replace the traditional gcc way of dealing with Mips16 code using floating point which involves essentially using soft float but with a library implemented using mips32 floating point. This gcc method also requires creating stubs so that Mips32 code can interact with these Mips 16 functions that have floating point needs. My conjecture is that in reality this traditional gcc method would never win over this new method. I will be implementing the traditional gcc method also. Some of it is already done but I needed to do the stubs to finish the work and those required this mips16/32 mixed mode capability. I have more ideas for to make this new method much better and I think the old method will just live in llvm for anyone that needs the backward compatibility but I don't for what reason that would be needed. llvm-svn: 179185
2013-04-10 18:58:04 +02:00
; 32: .ent foo1
; 32: .set noreorder
; 32: .set nomacro
; 32: .set noat
; 32: jr $ra
; 32: .set at
; 32: .set macro
; 32: .set reorder
; 32: .end foo1
define void @foo2() #0 {
entry:
%0 = load float* @x, align 4
call void @vf(float %0)
ret void
}
Fix a low hanging use of hasRawTextSupport. This also fixes the placement of the function label comment. It was being placed next to the mips16 directive instead of next to the label. llvm-svn: 199245
2014-01-14 19:57:12 +01:00
; 32: .set nomips16
This is for an experimental option -mips-os16. The idea is to compile all Mips32 code as Mips16 unless it can't be compiled as Mips 16. For now this would happen as long as floating point instructions are not needed. Probably it would also make sense to compile as mips32 if atomic operations are needed too. There may be other cases too. A module pass prescans the IR and adds the mips16 or nomips16 attribute to functions depending on the functions needs. Mips 16 mode can result in a 40% code compression by utililizing 16 bit encoding of many instructions. The hope is for this to replace the traditional gcc way of dealing with Mips16 code using floating point which involves essentially using soft float but with a library implemented using mips32 floating point. This gcc method also requires creating stubs so that Mips32 code can interact with these Mips 16 functions that have floating point needs. My conjecture is that in reality this traditional gcc method would never win over this new method. I will be implementing the traditional gcc method also. Some of it is already done but I needed to do the stubs to finish the work and those required this mips16/32 mixed mode capability. I have more ideas for to make this new method much better and I think the old method will just live in llvm for anyone that needs the backward compatibility but I don't for what reason that would be needed. llvm-svn: 179185
2013-04-10 18:58:04 +02:00
; 32: .ent foo2
; 32: .set noreorder
; 32: .set nomacro
; 32: .set noat
; 32: jr $ra
; 32: .set at
; 32: .set macro
; 32: .set reorder
; 32: .end foo2
define void @foo3() #0 {
entry:
%call = call float @fv()
store float %call, float* @x, align 4
ret void
}
Fix a low hanging use of hasRawTextSupport. This also fixes the placement of the function label comment. It was being placed next to the mips16 directive instead of next to the label. llvm-svn: 199245
2014-01-14 19:57:12 +01:00
; 32: .set nomips16
This is for an experimental option -mips-os16. The idea is to compile all Mips32 code as Mips16 unless it can't be compiled as Mips 16. For now this would happen as long as floating point instructions are not needed. Probably it would also make sense to compile as mips32 if atomic operations are needed too. There may be other cases too. A module pass prescans the IR and adds the mips16 or nomips16 attribute to functions depending on the functions needs. Mips 16 mode can result in a 40% code compression by utililizing 16 bit encoding of many instructions. The hope is for this to replace the traditional gcc way of dealing with Mips16 code using floating point which involves essentially using soft float but with a library implemented using mips32 floating point. This gcc method also requires creating stubs so that Mips32 code can interact with these Mips 16 functions that have floating point needs. My conjecture is that in reality this traditional gcc method would never win over this new method. I will be implementing the traditional gcc method also. Some of it is already done but I needed to do the stubs to finish the work and those required this mips16/32 mixed mode capability. I have more ideas for to make this new method much better and I think the old method will just live in llvm for anyone that needs the backward compatibility but I don't for what reason that would be needed. llvm-svn: 179185
2013-04-10 18:58:04 +02:00
; 32: .ent foo3
; 32: .set noreorder
; 32: .set nomacro
; 32: .set noat
; 32: jr $ra
; 32: .set at
; 32: .set macro
; 32: .set reorder
; 32: .end foo3
Update to remove the no-frame-pointer-elim-non-leaf flag if it was set to 'false'. llvm-svn: 189068
2013-08-22 23:28:54 +02:00
attributes #0 = { nounwind "less-precise-fpmad"="false" "no-frame-pointer-elim"="true" "no-frame-pointer-elim-non-leaf" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "unsafe-fp-math"="false" "use-soft-float"="false" }
This is for an experimental option -mips-os16. The idea is to compile all Mips32 code as Mips16 unless it can't be compiled as Mips 16. For now this would happen as long as floating point instructions are not needed. Probably it would also make sense to compile as mips32 if atomic operations are needed too. There may be other cases too. A module pass prescans the IR and adds the mips16 or nomips16 attribute to functions depending on the functions needs. Mips 16 mode can result in a 40% code compression by utililizing 16 bit encoding of many instructions. The hope is for this to replace the traditional gcc way of dealing with Mips16 code using floating point which involves essentially using soft float but with a library implemented using mips32 floating point. This gcc method also requires creating stubs so that Mips32 code can interact with these Mips 16 functions that have floating point needs. My conjecture is that in reality this traditional gcc method would never win over this new method. I will be implementing the traditional gcc method also. Some of it is already done but I needed to do the stubs to finish the work and those required this mips16/32 mixed mode capability. I have more ideas for to make this new method much better and I think the old method will just live in llvm for anyone that needs the backward compatibility but I don't for what reason that would be needed. llvm-svn: 179185
2013-04-10 18:58:04 +02:00
define void @vv() #0 {
entry:
ret void
}
Fix a low hanging use of hasRawTextSupport. This also fixes the placement of the function label comment. It was being placed next to the mips16 directive instead of next to the label. llvm-svn: 199245
2014-01-14 19:57:12 +01:00
; 32: .set mips16
This is for an experimental option -mips-os16. The idea is to compile all Mips32 code as Mips16 unless it can't be compiled as Mips 16. For now this would happen as long as floating point instructions are not needed. Probably it would also make sense to compile as mips32 if atomic operations are needed too. There may be other cases too. A module pass prescans the IR and adds the mips16 or nomips16 attribute to functions depending on the functions needs. Mips 16 mode can result in a 40% code compression by utililizing 16 bit encoding of many instructions. The hope is for this to replace the traditional gcc way of dealing with Mips16 code using floating point which involves essentially using soft float but with a library implemented using mips32 floating point. This gcc method also requires creating stubs so that Mips32 code can interact with these Mips 16 functions that have floating point needs. My conjecture is that in reality this traditional gcc method would never win over this new method. I will be implementing the traditional gcc method also. Some of it is already done but I needed to do the stubs to finish the work and those required this mips16/32 mixed mode capability. I have more ideas for to make this new method much better and I think the old method will just live in llvm for anyone that needs the backward compatibility but I don't for what reason that would be needed. llvm-svn: 179185
2013-04-10 18:58:04 +02:00
; 32: .ent vv
; 32: save {{.+}}
; 32: restore {{.+}}
; 32: .end vv
Reference in New Issue Copy Permalink