CuPBoP/examples/vecadd/vecadd-cuda-nvptx64-nvidia-...

74 lines
1.7 KiB
ArmAsm

//
// Generated by LLVM NVPTX Back-End
//
.version 7.5
.target sm_50
.address_size 64
// .globl _Z6vecAddPdS_S_i
.global .align 1 .b8 blockIdx[1];
.global .align 1 .b8 blockDim[1];
.global .align 1 .b8 threadIdx[1];
.visible .entry _Z6vecAddPdS_S_i(
.param .u64 _Z6vecAddPdS_S_i_param_0,
.param .u64 _Z6vecAddPdS_S_i_param_1,
.param .u64 _Z6vecAddPdS_S_i_param_2,
.param .u32 _Z6vecAddPdS_S_i_param_3
)
{
.local .align 8 .b8 __local_depot0[32];
.reg .b64 %SP;
.reg .b64 %SPL;
.reg .pred %p<2>;
.reg .b32 %r<9>;
.reg .b64 %rd<18>;
.reg .f64 %fd<4>;
mov.u64 %SPL, __local_depot0;
cvta.local.u64 %SP, %SPL;
ld.param.u32 %r1, [_Z6vecAddPdS_S_i_param_3];
ld.param.u64 %rd3, [_Z6vecAddPdS_S_i_param_2];
ld.param.u64 %rd2, [_Z6vecAddPdS_S_i_param_1];
ld.param.u64 %rd1, [_Z6vecAddPdS_S_i_param_0];
cvta.to.global.u64 %rd4, %rd3;
cvta.global.u64 %rd5, %rd4;
cvta.to.global.u64 %rd6, %rd2;
cvta.global.u64 %rd7, %rd6;
cvta.to.global.u64 %rd8, %rd1;
cvta.global.u64 %rd9, %rd8;
st.u64 [%SP+0], %rd9;
st.u64 [%SP+8], %rd7;
st.u64 [%SP+16], %rd5;
st.u32 [%SP+24], %r1;
mov.u32 %r2, %ctaid.x;
mov.u32 %r3, %ntid.x;
mul.lo.s32 %r4, %r2, %r3;
mov.u32 %r5, %tid.x;
add.s32 %r6, %r4, %r5;
st.u32 [%SP+28], %r6;
ld.u32 %r7, [%SP+28];
ld.u32 %r8, [%SP+24];
setp.ge.s32 %p1, %r7, %r8;
@%p1 bra LBB0_2;
bra.uni LBB0_1;
LBB0_1:
ld.u64 %rd10, [%SP+0];
ld.s32 %rd11, [%SP+28];
shl.b64 %rd12, %rd11, 3;
add.s64 %rd13, %rd10, %rd12;
ld.f64 %fd1, [%rd13];
ld.u64 %rd14, [%SP+8];
add.s64 %rd15, %rd14, %rd12;
ld.f64 %fd2, [%rd15];
add.rn.f64 %fd3, %fd1, %fd2;
ld.u64 %rd16, [%SP+16];
add.s64 %rd17, %rd16, %rd12;
st.f64 [%rd17], %fd3;
bra.uni LBB0_2;
LBB0_2:
ret;
}