1 ; RUN: opt -S -mtriple=amdgcn-unknown-unknown -amdgpu-promote-alloca < %s | FileCheck -check-prefix=IR %s 2 ; RUN: llc -march=amdgcn -mcpu=tonga < %s | FileCheck -check-prefix=ASM %s 3 4 5 @global_array0 = internal unnamed_addr addrspace(3) global [750 x [10 x i32]] undef, align 4 6 @global_array1 = internal unnamed_addr addrspace(3) global [750 x [10 x i32]] undef, align 4 7 8 ; IR-LABEL: define void @promote_alloca_size_256(i32 addrspace(1)* nocapture %out, i32 addrspace(1)* nocapture %in) { 9 ; IR: alloca [10 x i32] 10 ; ASM-LABEL: {{^}}promote_alloca_size_256: 11 ; ASM: ; LDSByteSize: 60000 bytes/workgroup (compile time only) 12 13 define void @promote_alloca_size_256(i32 addrspace(1)* nocapture %out, i32 addrspace(1)* nocapture %in) { 14 entry: 15 %stack = alloca [10 x i32], align 4 16 %tmp = load i32, i32 addrspace(1)* %in, align 4 17 %arrayidx1 = getelementptr inbounds [10 x i32], [10 x i32]* %stack, i32 0, i32 %tmp 18 store i32 4, i32* %arrayidx1, align 4 19 %arrayidx2 = getelementptr inbounds i32, i32 addrspace(1)* %in, i32 1 20 %tmp1 = load i32, i32 addrspace(1)* %arrayidx2, align 4 21 %arrayidx3 = getelementptr inbounds [10 x i32], [10 x i32]* %stack, i32 0, i32 %tmp1 22 store i32 5, i32* %arrayidx3, align 4 23 %arrayidx10 = getelementptr inbounds [10 x i32], [10 x i32]* %stack, i32 0, i32 0 24 %tmp2 = load i32, i32* %arrayidx10, align 4 25 store i32 %tmp2, i32 addrspace(1)* %out, align 4 26 %arrayidx12 = getelementptr inbounds [10 x i32], [10 x i32]* %stack, i32 0, i32 1 27 %tmp3 = load i32, i32* %arrayidx12 28 %arrayidx13 = getelementptr inbounds i32, i32 addrspace(1)* %out, i32 1 29 store i32 %tmp3, i32 addrspace(1)* %arrayidx13 30 %v0 = getelementptr inbounds [750 x [10 x i32]], [750 x [10 x i32]] addrspace(3)* @global_array0, i32 0, i32 0, i32 0 31 store i32 %tmp3, i32 addrspace(3)* %v0 32 %v1 = getelementptr inbounds [750 x [10 x i32]], [750 x [10 x i32]] addrspace(3)* @global_array1, i32 0, i32 0, i32 0 33 store i32 %tmp3, i32 addrspace(3)* %v1 34 ret void 35 } 36
