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/AMDGPU/promote-alloca-to-lds-selec...

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

136 lines
7.5 KiB
LLVM
Raw Normal View History

AMDGPU: Fix breaking IR on instructions with multiple pointer operands The promote alloca pass would attempt to promote an alloca with a select, icmp, or phi user, even though the other operand was from a non-promotable source, producing a select on two different pointer types. Only do this if we know that both operands derive from the same alloca. In the future we should be able to relax this to an alloca which will also be promoted. llvm-svn: 269265
2016-05-12 09:58:58 +08:00
; RUN: opt -S -mtriple=amdgcn-unknown-amdhsa -mcpu=kaveri -amdgpu-promote-alloca < %s | FileCheck %s
AMDGPU: Fix using old address spaces in some tests llvm-svn: 348385
2018-12-06 01:34:59 +08:00
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5"
AMDGPU: Fix breaking IR on instructions with multiple pointer operands The promote alloca pass would attempt to promote an alloca with a select, icmp, or phi user, even though the other operand was from a non-promotable source, producing a select on two different pointer types. Only do this if we know that both operands derive from the same alloca. In the future we should be able to relax this to an alloca which will also be promoted. llvm-svn: 269265
2016-05-12 09:58:58 +08:00
; CHECK-LABEL: @lds_promoted_alloca_select_invalid_pointer_operand(
; CHECK: %alloca = alloca i32
AMDGPU: Fix using old address spaces in some tests llvm-svn: 348385
2018-12-06 01:34:59 +08:00
; CHECK: select i1 undef, i32 addrspace(5)* undef, i32 addrspace(5)* %alloca
AMDGPU: Mark all unspecified CC functions in tests as amdgpu_kernel Currently the default C calling convention functions are treated the same as compute kernels. Make this explicit so the default calling convention can be changed to a non-kernel. Converted with perl -pi -e 's/define void/define amdgpu_kernel void/' on the relevant test directories (and undoing in one place that actually wanted a non-kernel). llvm-svn: 298444
2017-03-22 05:39:51 +08:00
define amdgpu_kernel void @lds_promoted_alloca_select_invalid_pointer_operand() #0 {
AMDGPU: Fix using old address spaces in some tests llvm-svn: 348385
2018-12-06 01:34:59 +08:00
%alloca = alloca i32, align 4, addrspace(5)
%select = select i1 undef, i32 addrspace(5)* undef, i32 addrspace(5)* %alloca
store i32 0, i32 addrspace(5)* %select, align 4
AMDGPU: Fix breaking IR on instructions with multiple pointer operands The promote alloca pass would attempt to promote an alloca with a select, icmp, or phi user, even though the other operand was from a non-promotable source, producing a select on two different pointer types. Only do this if we know that both operands derive from the same alloca. In the future we should be able to relax this to an alloca which will also be promoted. llvm-svn: 269265
2016-05-12 09:58:58 +08:00
ret void
}
; CHECK-LABEL: @lds_promote_alloca_select_two_derived_pointers(
; CHECK: [[ARRAYGEP:%[0-9]+]] = getelementptr inbounds [256 x [16 x i32]], [256 x [16 x i32]] addrspace(3)* @lds_promote_alloca_select_two_derived_pointers.alloca, i32 0, i32 %{{[0-9]+}}
; CHECK: %ptr0 = getelementptr inbounds [16 x i32], [16 x i32] addrspace(3)* [[ARRAYGEP]], i32 0, i32 %a
; CHECK: %ptr1 = getelementptr inbounds [16 x i32], [16 x i32] addrspace(3)* [[ARRAYGEP]], i32 0, i32 %b
; CHECK: %select = select i1 undef, i32 addrspace(3)* %ptr0, i32 addrspace(3)* %ptr1
; CHECK: store i32 0, i32 addrspace(3)* %select, align 4
AMDGPU: Mark all unspecified CC functions in tests as amdgpu_kernel Currently the default C calling convention functions are treated the same as compute kernels. Make this explicit so the default calling convention can be changed to a non-kernel. Converted with perl -pi -e 's/define void/define amdgpu_kernel void/' on the relevant test directories (and undoing in one place that actually wanted a non-kernel). llvm-svn: 298444
2017-03-22 05:39:51 +08:00
define amdgpu_kernel void @lds_promote_alloca_select_two_derived_pointers(i32 %a, i32 %b) #0 {
AMDGPU: Fix using old address spaces in some tests llvm-svn: 348385
2018-12-06 01:34:59 +08:00
%alloca = alloca [16 x i32], align 4, addrspace(5)
%ptr0 = getelementptr inbounds [16 x i32], [16 x i32] addrspace(5)* %alloca, i32 0, i32 %a
%ptr1 = getelementptr inbounds [16 x i32], [16 x i32] addrspace(5)* %alloca, i32 0, i32 %b
%select = select i1 undef, i32 addrspace(5)* %ptr0, i32 addrspace(5)* %ptr1
store i32 0, i32 addrspace(5)* %select, align 4
AMDGPU: Fix breaking IR on instructions with multiple pointer operands The promote alloca pass would attempt to promote an alloca with a select, icmp, or phi user, even though the other operand was from a non-promotable source, producing a select on two different pointer types. Only do this if we know that both operands derive from the same alloca. In the future we should be able to relax this to an alloca which will also be promoted. llvm-svn: 269265
2016-05-12 09:58:58 +08:00
ret void
}
; FIXME: This should be promotable but requires knowing that both will be promoted first.
; CHECK-LABEL: @lds_promote_alloca_select_two_allocas(
; CHECK: %alloca0 = alloca i32, i32 16, align 4
; CHECK: %alloca1 = alloca i32, i32 16, align 4
AMDGPU: Fix using old address spaces in some tests llvm-svn: 348385
2018-12-06 01:34:59 +08:00
; CHECK: %ptr0 = getelementptr inbounds i32, i32 addrspace(5)* %alloca0, i32 %a
; CHECK: %ptr1 = getelementptr inbounds i32, i32 addrspace(5)* %alloca1, i32 %b
; CHECK: %select = select i1 undef, i32 addrspace(5)* %ptr0, i32 addrspace(5)* %ptr1
AMDGPU: Mark all unspecified CC functions in tests as amdgpu_kernel Currently the default C calling convention functions are treated the same as compute kernels. Make this explicit so the default calling convention can be changed to a non-kernel. Converted with perl -pi -e 's/define void/define amdgpu_kernel void/' on the relevant test directories (and undoing in one place that actually wanted a non-kernel). llvm-svn: 298444
2017-03-22 05:39:51 +08:00
define amdgpu_kernel void @lds_promote_alloca_select_two_allocas(i32 %a, i32 %b) #0 {
AMDGPU: Fix using old address spaces in some tests llvm-svn: 348385
2018-12-06 01:34:59 +08:00
%alloca0 = alloca i32, i32 16, align 4, addrspace(5)
%alloca1 = alloca i32, i32 16, align 4, addrspace(5)
%ptr0 = getelementptr inbounds i32, i32 addrspace(5)* %alloca0, i32 %a
%ptr1 = getelementptr inbounds i32, i32 addrspace(5)* %alloca1, i32 %b
%select = select i1 undef, i32 addrspace(5)* %ptr0, i32 addrspace(5)* %ptr1
store i32 0, i32 addrspace(5)* %select, align 4
AMDGPU: Fix breaking IR on instructions with multiple pointer operands The promote alloca pass would attempt to promote an alloca with a select, icmp, or phi user, even though the other operand was from a non-promotable source, producing a select on two different pointer types. Only do this if we know that both operands derive from the same alloca. In the future we should be able to relax this to an alloca which will also be promoted. llvm-svn: 269265
2016-05-12 09:58:58 +08:00
ret void
}
; TODO: Maybe this should be canonicalized to select on the constant and GEP after.
; CHECK-LABEL: @lds_promote_alloca_select_two_derived_constant_pointers(
; CHECK: [[ARRAYGEP:%[0-9]+]] = getelementptr inbounds [256 x [16 x i32]], [256 x [16 x i32]] addrspace(3)* @lds_promote_alloca_select_two_derived_constant_pointers.alloca, i32 0, i32 %{{[0-9]+}}
; CHECK: %ptr0 = getelementptr inbounds [16 x i32], [16 x i32] addrspace(3)* [[ARRAYGEP]], i32 0, i32 1
; CHECK: %ptr1 = getelementptr inbounds [16 x i32], [16 x i32] addrspace(3)* [[ARRAYGEP]], i32 0, i32 3
; CHECK: %select = select i1 undef, i32 addrspace(3)* %ptr0, i32 addrspace(3)* %ptr1
; CHECK: store i32 0, i32 addrspace(3)* %select, align 4
AMDGPU: Mark all unspecified CC functions in tests as amdgpu_kernel Currently the default C calling convention functions are treated the same as compute kernels. Make this explicit so the default calling convention can be changed to a non-kernel. Converted with perl -pi -e 's/define void/define amdgpu_kernel void/' on the relevant test directories (and undoing in one place that actually wanted a non-kernel). llvm-svn: 298444
2017-03-22 05:39:51 +08:00
define amdgpu_kernel void @lds_promote_alloca_select_two_derived_constant_pointers() #0 {
AMDGPU: Fix using old address spaces in some tests llvm-svn: 348385
2018-12-06 01:34:59 +08:00
%alloca = alloca [16 x i32], align 4, addrspace(5)
%ptr0 = getelementptr inbounds [16 x i32], [16 x i32] addrspace(5)* %alloca, i32 0, i32 1
%ptr1 = getelementptr inbounds [16 x i32], [16 x i32] addrspace(5)* %alloca, i32 0, i32 3
%select = select i1 undef, i32 addrspace(5)* %ptr0, i32 addrspace(5)* %ptr1
store i32 0, i32 addrspace(5)* %select, align 4
AMDGPU: Fix breaking IR on instructions with multiple pointer operands The promote alloca pass would attempt to promote an alloca with a select, icmp, or phi user, even though the other operand was from a non-promotable source, producing a select on two different pointer types. Only do this if we know that both operands derive from the same alloca. In the future we should be able to relax this to an alloca which will also be promoted. llvm-svn: 269265
2016-05-12 09:58:58 +08:00
ret void
}
; CHECK-LABEL: @lds_promoted_alloca_select_input_select(
; CHECK: getelementptr inbounds [256 x [16 x i32]], [256 x [16 x i32]] addrspace(3)* @lds_promoted_alloca_select_input_select.alloca, i32 0, i32 %{{[0-9]+}}
; CHECK: %ptr0 = getelementptr inbounds [16 x i32], [16 x i32] addrspace(3)* %{{[0-9]+}}, i32 0, i32 %a
; CHECK: %ptr1 = getelementptr inbounds [16 x i32], [16 x i32] addrspace(3)* %{{[0-9]+}}, i32 0, i32 %b
; CHECK: %ptr2 = getelementptr inbounds [16 x i32], [16 x i32] addrspace(3)* %{{[0-9]+}}, i32 0, i32 %c
; CHECK: %select0 = select i1 undef, i32 addrspace(3)* %ptr0, i32 addrspace(3)* %ptr1
; CHECK: %select1 = select i1 undef, i32 addrspace(3)* %select0, i32 addrspace(3)* %ptr2
; CHECK: store i32 0, i32 addrspace(3)* %select1, align 4
AMDGPU: Mark all unspecified CC functions in tests as amdgpu_kernel Currently the default C calling convention functions are treated the same as compute kernels. Make this explicit so the default calling convention can be changed to a non-kernel. Converted with perl -pi -e 's/define void/define amdgpu_kernel void/' on the relevant test directories (and undoing in one place that actually wanted a non-kernel). llvm-svn: 298444
2017-03-22 05:39:51 +08:00
define amdgpu_kernel void @lds_promoted_alloca_select_input_select(i32 %a, i32 %b, i32 %c) #0 {
AMDGPU: Fix using old address spaces in some tests llvm-svn: 348385
2018-12-06 01:34:59 +08:00
%alloca = alloca [16 x i32], align 4, addrspace(5)
%ptr0 = getelementptr inbounds [16 x i32], [16 x i32] addrspace(5)* %alloca, i32 0, i32 %a
%ptr1 = getelementptr inbounds [16 x i32], [16 x i32] addrspace(5)* %alloca, i32 0, i32 %b
%ptr2 = getelementptr inbounds [16 x i32], [16 x i32] addrspace(5)* %alloca, i32 0, i32 %c
%select0 = select i1 undef, i32 addrspace(5)* %ptr0, i32 addrspace(5)* %ptr1
%select1 = select i1 undef, i32 addrspace(5)* %select0, i32 addrspace(5)* %ptr2
store i32 0, i32 addrspace(5)* %select1, align 4
AMDGPU: Fix breaking IR on instructions with multiple pointer operands The promote alloca pass would attempt to promote an alloca with a select, icmp, or phi user, even though the other operand was from a non-promotable source, producing a select on two different pointer types. Only do this if we know that both operands derive from the same alloca. In the future we should be able to relax this to an alloca which will also be promoted. llvm-svn: 269265
2016-05-12 09:58:58 +08:00
ret void
}
AMDGPU: Mark all unspecified CC functions in tests as amdgpu_kernel Currently the default C calling convention functions are treated the same as compute kernels. Make this explicit so the default calling convention can be changed to a non-kernel. Converted with perl -pi -e 's/define void/define amdgpu_kernel void/' on the relevant test directories (and undoing in one place that actually wanted a non-kernel). llvm-svn: 298444
2017-03-22 05:39:51 +08:00
define amdgpu_kernel void @lds_promoted_alloca_select_input_phi(i32 %a, i32 %b, i32 %c) #0 {
AMDGPU: Fix breaking IR on instructions with multiple pointer operands The promote alloca pass would attempt to promote an alloca with a select, icmp, or phi user, even though the other operand was from a non-promotable source, producing a select on two different pointer types. Only do this if we know that both operands derive from the same alloca. In the future we should be able to relax this to an alloca which will also be promoted. llvm-svn: 269265
2016-05-12 09:58:58 +08:00
entry:
AMDGPU: Fix using old address spaces in some tests llvm-svn: 348385
2018-12-06 01:34:59 +08:00
%alloca = alloca [16 x i32], align 4, addrspace(5)
%ptr0 = getelementptr inbounds [16 x i32], [16 x i32] addrspace(5)* %alloca, i32 0, i32 %a
%ptr1 = getelementptr inbounds [16 x i32], [16 x i32] addrspace(5)* %alloca, i32 0, i32 %b
store i32 0, i32 addrspace(5)* %ptr0
AMDGPU: Fix breaking IR on instructions with multiple pointer operands The promote alloca pass would attempt to promote an alloca with a select, icmp, or phi user, even though the other operand was from a non-promotable source, producing a select on two different pointer types. Only do this if we know that both operands derive from the same alloca. In the future we should be able to relax this to an alloca which will also be promoted. llvm-svn: 269265
2016-05-12 09:58:58 +08:00
br i1 undef, label %bb1, label %bb2
bb1:
AMDGPU: Fix using old address spaces in some tests llvm-svn: 348385
2018-12-06 01:34:59 +08:00
%ptr2 = getelementptr inbounds [16 x i32], [16 x i32] addrspace(5)* %alloca, i32 0, i32 %c
%select0 = select i1 undef, i32 addrspace(5)* undef, i32 addrspace(5)* %ptr2
store i32 0, i32 addrspace(5)* %ptr1
AMDGPU: Fix breaking IR on instructions with multiple pointer operands The promote alloca pass would attempt to promote an alloca with a select, icmp, or phi user, even though the other operand was from a non-promotable source, producing a select on two different pointer types. Only do this if we know that both operands derive from the same alloca. In the future we should be able to relax this to an alloca which will also be promoted. llvm-svn: 269265
2016-05-12 09:58:58 +08:00
br label %bb2
bb2:
AMDGPU: Fix using old address spaces in some tests llvm-svn: 348385
2018-12-06 01:34:59 +08:00
%phi.ptr = phi i32 addrspace(5)* [ %ptr0, %entry ], [ %select0, %bb1 ]
%select1 = select i1 undef, i32 addrspace(5)* %phi.ptr, i32 addrspace(5)* %ptr1
store i32 0, i32 addrspace(5)* %select1, align 4
AMDGPU: Fix breaking IR on instructions with multiple pointer operands The promote alloca pass would attempt to promote an alloca with a select, icmp, or phi user, even though the other operand was from a non-promotable source, producing a select on two different pointer types. Only do this if we know that both operands derive from the same alloca. In the future we should be able to relax this to an alloca which will also be promoted. llvm-svn: 269265
2016-05-12 09:58:58 +08:00
ret void
}
AMDGPU: Handle alloca promoting with null operands If the second pointer in a multi-pointer instruction is a constant, we can replace the type. llvm-svn: 269945
2016-05-18 23:57:21 +08:00
; CHECK-LABEL: @select_null_rhs(
; CHECK-NOT: alloca
; CHECK: select i1 %tmp2, double addrspace(3)* %{{[0-9]+}}, double addrspace(3)* null
AMDGPU: Mark all unspecified CC functions in tests as amdgpu_kernel Currently the default C calling convention functions are treated the same as compute kernels. Make this explicit so the default calling convention can be changed to a non-kernel. Converted with perl -pi -e 's/define void/define amdgpu_kernel void/' on the relevant test directories (and undoing in one place that actually wanted a non-kernel). llvm-svn: 298444
2017-03-22 05:39:51 +08:00
define amdgpu_kernel void @select_null_rhs(double addrspace(1)* nocapture %arg, i32 %arg1) #1 {
AMDGPU: Handle alloca promoting with null operands If the second pointer in a multi-pointer instruction is a constant, we can replace the type. llvm-svn: 269945
2016-05-18 23:57:21 +08:00
bb:
AMDGPU: Fix using old address spaces in some tests llvm-svn: 348385
2018-12-06 01:34:59 +08:00
%tmp = alloca double, align 8, addrspace(5)
store double 0.000000e+00, double addrspace(5)* %tmp, align 8
AMDGPU: Handle alloca promoting with null operands If the second pointer in a multi-pointer instruction is a constant, we can replace the type. llvm-svn: 269945
2016-05-18 23:57:21 +08:00
%tmp2 = icmp eq i32 %arg1, 0
AMDGPU: Fix using old address spaces in some tests llvm-svn: 348385
2018-12-06 01:34:59 +08:00
%tmp3 = select i1 %tmp2, double addrspace(5)* %tmp, double addrspace(5)* null
store double 1.000000e+00, double addrspace(5)* %tmp3, align 8
%tmp4 = load double, double addrspace(5)* %tmp, align 8
AMDGPU: Handle alloca promoting with null operands If the second pointer in a multi-pointer instruction is a constant, we can replace the type. llvm-svn: 269945
2016-05-18 23:57:21 +08:00
store double %tmp4, double addrspace(1)* %arg
ret void
}
; CHECK-LABEL: @select_null_lhs(
; CHECK-NOT: alloca
; CHECK: select i1 %tmp2, double addrspace(3)* null, double addrspace(3)* %{{[0-9]+}}
AMDGPU: Mark all unspecified CC functions in tests as amdgpu_kernel Currently the default C calling convention functions are treated the same as compute kernels. Make this explicit so the default calling convention can be changed to a non-kernel. Converted with perl -pi -e 's/define void/define amdgpu_kernel void/' on the relevant test directories (and undoing in one place that actually wanted a non-kernel). llvm-svn: 298444
2017-03-22 05:39:51 +08:00
define amdgpu_kernel void @select_null_lhs(double addrspace(1)* nocapture %arg, i32 %arg1) #1 {
AMDGPU: Handle alloca promoting with null operands If the second pointer in a multi-pointer instruction is a constant, we can replace the type. llvm-svn: 269945
2016-05-18 23:57:21 +08:00
bb:
AMDGPU: Fix using old address spaces in some tests llvm-svn: 348385
2018-12-06 01:34:59 +08:00
%tmp = alloca double, align 8, addrspace(5)
store double 0.000000e+00, double addrspace(5)* %tmp, align 8
AMDGPU: Handle alloca promoting with null operands If the second pointer in a multi-pointer instruction is a constant, we can replace the type. llvm-svn: 269945
2016-05-18 23:57:21 +08:00
%tmp2 = icmp eq i32 %arg1, 0
AMDGPU: Fix using old address spaces in some tests llvm-svn: 348385
2018-12-06 01:34:59 +08:00
%tmp3 = select i1 %tmp2, double addrspace(5)* null, double addrspace(5)* %tmp
store double 1.000000e+00, double addrspace(5)* %tmp3, align 8
%tmp4 = load double, double addrspace(5)* %tmp, align 8
AMDGPU: Handle alloca promoting with null operands If the second pointer in a multi-pointer instruction is a constant, we can replace the type. llvm-svn: 269945
2016-05-18 23:57:21 +08:00
store double %tmp4, double addrspace(1)* %arg
ret void
}
AMDGPU: Switch backend default max workgroup size to 1024 Previously this would default to 256, not the maximum supported size of 1024. Using a maximum lower than the hardware maximum requires language runtimes to enforce this limit for correctness, which no language has correctly done. Switch the default to the conservatively correct maximum, and force frontends to opt-in to the more optimal 256 default maximum. I don't really understand why the changes in occupancy-levels.ll increased the computed occupancy, which I expected to decrease. I'm not sure if these tests should be forcing the old maximum.
2019-08-28 00:34:40 +08:00
attributes #0 = { norecurse nounwind "amdgpu-waves-per-eu"="1,1" "amdgpu-flat-work-group-size"="1,256" }
[CodeGen] Always use `printReg` to print registers in both MIR and debug output As part of the unification of the debug format and the MIR format, always use `printReg` to print all kinds of registers. Updated the tests using '_' instead of '%noreg' until we decide which one we want to be the default one. Differential Revision: https://reviews.llvm.org/D40421 llvm-svn: 319445
2017-12-01 00:12:24 +08:00
attributes #1 = { norecurse nounwind }