6502/llvm-6502
1
0
Fork 0
mirror of https://github.com/c64scene-ar/llvm-6502.git synced 2024-09-02 11:30:10 +00:00
Code Issues Projects Releases Wiki Activity
f25a5f3290
llvm-6502/test/CodeGen/R600/llvm.memcpy.ll

365 lines
9.9 KiB
LLVM
Raw Normal View History

R600/SI: Implement getOptimalMemOpType The default guess uses i32. This needs an address space argument to really do the right thing in all cases. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214104 91177308-0d34-0410-b5e6-96231b3b80d8
2014-07-28 17:49:26 +00:00
; RUN: llc -march=r600 -mcpu=SI -verify-machineinstrs < %s | FileCheck -check-prefix=SI -check-prefix=FUNC %s
declare void @llvm.memcpy.p3i8.p3i8.i32(i8 addrspace(3)* nocapture, i8 addrspace(3)* nocapture, i32, i32, i1) nounwind
declare void @llvm.memcpy.p1i8.p1i8.i64(i8 addrspace(1)* nocapture, i8 addrspace(1)* nocapture, i64, i32, i1) nounwind
R600: Call EmitFunctionHeader() in the AsmPrinter to populate the ELF symbol table git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218776 91177308-0d34-0410-b5e6-96231b3b80d8
2014-10-01 17:15:17 +00:00
; FUNC-LABEL: {{^}}test_small_memcpy_i64_lds_to_lds_align1:
R600/SI: Implement getOptimalMemOpType The default guess uses i32. This needs an address space argument to really do the right thing in all cases. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214104 91177308-0d34-0410-b5e6-96231b3b80d8
2014-07-28 17:49:26 +00:00
; SI: DS_READ_U8
; SI: DS_WRITE_B8
; SI: DS_READ_U8
; SI: DS_WRITE_B8
; SI: DS_READ_U8
; SI: DS_WRITE_B8
; SI: DS_READ_U8
; SI: DS_WRITE_B8
; SI: DS_READ_U8
; SI: DS_WRITE_B8
R600: Increase nearby load scheduling threshold. This partially fixes weird looking load scheduling in memcpy test. The load clustering doesn't seem particularly smart, but this method seems to be partially deprecated so it might not be worth trying to fix. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214943 91177308-0d34-0410-b5e6-96231b3b80d8
2014-08-06 00:29:49 +00:00
R600/SI: Implement getOptimalMemOpType The default guess uses i32. This needs an address space argument to really do the right thing in all cases. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214104 91177308-0d34-0410-b5e6-96231b3b80d8
2014-07-28 17:49:26 +00:00
; SI: DS_READ_U8
; SI: DS_WRITE_B8
; SI: DS_READ_U8
; SI: DS_WRITE_B8
; SI: DS_READ_U8
; SI: DS_WRITE_B8
; SI: DS_READ_U8
; SI: DS_WRITE_B8
; SI: DS_READ_U8
; SI: DS_WRITE_B8
R600: Increase nearby load scheduling threshold. This partially fixes weird looking load scheduling in memcpy test. The load clustering doesn't seem particularly smart, but this method seems to be partially deprecated so it might not be worth trying to fix. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214943 91177308-0d34-0410-b5e6-96231b3b80d8
2014-08-06 00:29:49 +00:00
R600/SI: Implement getOptimalMemOpType The default guess uses i32. This needs an address space argument to really do the right thing in all cases. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214104 91177308-0d34-0410-b5e6-96231b3b80d8
2014-07-28 17:49:26 +00:00
; SI: DS_READ_U8
; SI: DS_WRITE_B8
; SI: DS_READ_U8
; SI: DS_WRITE_B8
; SI: DS_READ_U8
; SI: DS_WRITE_B8
; SI: DS_READ_U8
; SI: DS_WRITE_B8
; SI: DS_READ_U8
; SI: DS_READ_U8
R600: Increase nearby load scheduling threshold. This partially fixes weird looking load scheduling in memcpy test. The load clustering doesn't seem particularly smart, but this method seems to be partially deprecated so it might not be worth trying to fix. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214943 91177308-0d34-0410-b5e6-96231b3b80d8
2014-08-06 00:29:49 +00:00
R600/SI: Implement getOptimalMemOpType The default guess uses i32. This needs an address space argument to really do the right thing in all cases. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214104 91177308-0d34-0410-b5e6-96231b3b80d8
2014-07-28 17:49:26 +00:00
; SI: DS_READ_U8
; SI: DS_READ_U8
; SI: DS_READ_U8
; SI: DS_READ_U8
; SI: DS_READ_U8
; SI: DS_READ_U8
; SI: DS_READ_U8
; SI: DS_READ_U8
R600: Increase nearby load scheduling threshold. This partially fixes weird looking load scheduling in memcpy test. The load clustering doesn't seem particularly smart, but this method seems to be partially deprecated so it might not be worth trying to fix. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214943 91177308-0d34-0410-b5e6-96231b3b80d8
2014-08-06 00:29:49 +00:00
R600/SI: Implement getOptimalMemOpType The default guess uses i32. This needs an address space argument to really do the right thing in all cases. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214104 91177308-0d34-0410-b5e6-96231b3b80d8
2014-07-28 17:49:26 +00:00
; SI: DS_READ_U8
; SI: DS_READ_U8
; SI: DS_READ_U8
; SI: DS_READ_U8
; SI: DS_READ_U8
; SI: DS_READ_U8
; SI: DS_READ_U8
; SI: DS_READ_U8
R600/SI: Implement areLoadsFromSameBasePtr This currently has a noticable effect on the kernel argument loads. LDS and global loads are more problematic, I think because of how copies are currently inserted to ensure that the address is a VGPR. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214942 91177308-0d34-0410-b5e6-96231b3b80d8
2014-08-06 00:29:43 +00:00
; SI: DS_WRITE_B8
; SI: DS_WRITE_B8
; SI: DS_WRITE_B8
; SI: DS_WRITE_B8
; SI: DS_WRITE_B8
; SI: DS_WRITE_B8
; SI: DS_WRITE_B8
R600: Increase nearby load scheduling threshold. This partially fixes weird looking load scheduling in memcpy test. The load clustering doesn't seem particularly smart, but this method seems to be partially deprecated so it might not be worth trying to fix. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214943 91177308-0d34-0410-b5e6-96231b3b80d8
2014-08-06 00:29:49 +00:00
; SI: DS_WRITE_B8
; SI: DS_WRITE_B8
R600/SI: Implement areLoadsFromSameBasePtr This currently has a noticable effect on the kernel argument loads. LDS and global loads are more problematic, I think because of how copies are currently inserted to ensure that the address is a VGPR. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214942 91177308-0d34-0410-b5e6-96231b3b80d8
2014-08-06 00:29:43 +00:00
; SI: DS_WRITE_B8
; SI: DS_WRITE_B8
; SI: DS_WRITE_B8
; SI: DS_WRITE_B8
; SI: DS_WRITE_B8
; SI: DS_WRITE_B8
; SI: DS_WRITE_B8
; SI: DS_WRITE_B8
R600/SI: Implement getOptimalMemOpType The default guess uses i32. This needs an address space argument to really do the right thing in all cases. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214104 91177308-0d34-0410-b5e6-96231b3b80d8
2014-07-28 17:49:26 +00:00
; SI: DS_WRITE_B8
; SI: S_ENDPGM
define void @test_small_memcpy_i64_lds_to_lds_align1(i64 addrspace(3)* noalias %out, i64 addrspace(3)* noalias %in) nounwind {
%bcin = bitcast i64 addrspace(3)* %in to i8 addrspace(3)*
%bcout = bitcast i64 addrspace(3)* %out to i8 addrspace(3)*
call void @llvm.memcpy.p3i8.p3i8.i32(i8 addrspace(3)* %bcout, i8 addrspace(3)* %bcin, i32 32, i32 1, i1 false) nounwind
ret void
}
R600: Call EmitFunctionHeader() in the AsmPrinter to populate the ELF symbol table git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218776 91177308-0d34-0410-b5e6-96231b3b80d8
2014-10-01 17:15:17 +00:00
; FUNC-LABEL: {{^}}test_small_memcpy_i64_lds_to_lds_align2:
R600/SI: Implement getOptimalMemOpType The default guess uses i32. This needs an address space argument to really do the right thing in all cases. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214104 91177308-0d34-0410-b5e6-96231b3b80d8
2014-07-28 17:49:26 +00:00
; SI: DS_READ_U16
; SI: DS_READ_U16
; SI: DS_READ_U16
; SI: DS_READ_U16
; SI: DS_READ_U16
; SI: DS_READ_U16
; SI: DS_READ_U16
; SI: DS_READ_U16
; SI: DS_READ_U16
; SI: DS_READ_U16
; SI: DS_READ_U16
; SI: DS_READ_U16
; SI: DS_READ_U16
; SI: DS_READ_U16
; SI: DS_READ_U16
; SI: DS_READ_U16
R600/SI: Implement areLoadsFromSameBasePtr This currently has a noticable effect on the kernel argument loads. LDS and global loads are more problematic, I think because of how copies are currently inserted to ensure that the address is a VGPR. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214942 91177308-0d34-0410-b5e6-96231b3b80d8
2014-08-06 00:29:43 +00:00
; SI: DS_WRITE_B16
; SI: DS_WRITE_B16
; SI: DS_WRITE_B16
; SI: DS_WRITE_B16
; SI: DS_WRITE_B16
; SI: DS_WRITE_B16
; SI: DS_WRITE_B16
R600/SI: Implement getOptimalMemOpType The default guess uses i32. This needs an address space argument to really do the right thing in all cases. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214104 91177308-0d34-0410-b5e6-96231b3b80d8
2014-07-28 17:49:26 +00:00
; SI: DS_WRITE_B16
R600: Increase nearby load scheduling threshold. This partially fixes weird looking load scheduling in memcpy test. The load clustering doesn't seem particularly smart, but this method seems to be partially deprecated so it might not be worth trying to fix. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214943 91177308-0d34-0410-b5e6-96231b3b80d8
2014-08-06 00:29:49 +00:00
; SI: DS_WRITE_B16
; SI: DS_WRITE_B16
; SI: DS_WRITE_B16
; SI: DS_WRITE_B16
; SI: DS_WRITE_B16
; SI: DS_WRITE_B16
; SI: DS_WRITE_B16
; SI: DS_WRITE_B16
R600/SI: Implement getOptimalMemOpType The default guess uses i32. This needs an address space argument to really do the right thing in all cases. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214104 91177308-0d34-0410-b5e6-96231b3b80d8
2014-07-28 17:49:26 +00:00
; SI: S_ENDPGM
define void @test_small_memcpy_i64_lds_to_lds_align2(i64 addrspace(3)* noalias %out, i64 addrspace(3)* noalias %in) nounwind {
%bcin = bitcast i64 addrspace(3)* %in to i8 addrspace(3)*
%bcout = bitcast i64 addrspace(3)* %out to i8 addrspace(3)*
call void @llvm.memcpy.p3i8.p3i8.i32(i8 addrspace(3)* %bcout, i8 addrspace(3)* %bcin, i32 32, i32 2, i1 false) nounwind
ret void
}
R600: Call EmitFunctionHeader() in the AsmPrinter to populate the ELF symbol table git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218776 91177308-0d34-0410-b5e6-96231b3b80d8
2014-10-01 17:15:17 +00:00
; FUNC-LABEL: {{^}}test_small_memcpy_i64_lds_to_lds_align4:
R600/SI: Implement getOptimalMemOpType The default guess uses i32. This needs an address space argument to really do the right thing in all cases. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214104 91177308-0d34-0410-b5e6-96231b3b80d8
2014-07-28 17:49:26 +00:00
; SI-DAG: DS_READ_B32
; SI-DAG: DS_WRITE_B32
; SI-DAG: DS_READ_B32
; SI-DAG: DS_WRITE_B32
; SI-DAG: DS_READ_B32
; SI-DAG: DS_WRITE_B32
; SI-DAG: DS_READ_B32
; SI-DAG: DS_WRITE_B32
; SI-DAG: DS_READ_B32
; SI-DAG: DS_WRITE_B32
; SI-DAG: DS_READ_B32
; SI-DAG: DS_WRITE_B32
; SI-DAG: DS_READ_B32
; SI-DAG: DS_WRITE_B32
; SI-DAG: DS_READ_B32
; SI-DAG: DS_WRITE_B32
; SI-DAG: DS_READ_B32
; SI-DAG: DS_WRITE_B32
; SI: S_ENDPGM
define void @test_small_memcpy_i64_lds_to_lds_align4(i64 addrspace(3)* noalias %out, i64 addrspace(3)* noalias %in) nounwind {
%bcin = bitcast i64 addrspace(3)* %in to i8 addrspace(3)*
%bcout = bitcast i64 addrspace(3)* %out to i8 addrspace(3)*
call void @llvm.memcpy.p3i8.p3i8.i32(i8 addrspace(3)* %bcout, i8 addrspace(3)* %bcin, i32 32, i32 4, i1 false) nounwind
ret void
}
; FIXME: Use 64-bit ops
R600: Call EmitFunctionHeader() in the AsmPrinter to populate the ELF symbol table git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218776 91177308-0d34-0410-b5e6-96231b3b80d8
2014-10-01 17:15:17 +00:00
; FUNC-LABEL: {{^}}test_small_memcpy_i64_lds_to_lds_align8:
R600/SI: Implement getOptimalMemOpType The default guess uses i32. This needs an address space argument to really do the right thing in all cases. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214104 91177308-0d34-0410-b5e6-96231b3b80d8
2014-07-28 17:49:26 +00:00
; SI-DAG: DS_READ_B32
; SI-DAG: DS_WRITE_B32
; SI-DAG: DS_READ_B32
; SI-DAG: DS_WRITE_B32
; SI-DAG: DS_READ_B32
; SI-DAG: DS_WRITE_B32
; SI-DAG: DS_READ_B32
; SI-DAG: DS_WRITE_B32
; SI-DAG: DS_READ_B32
; SI-DAG: DS_WRITE_B32
; SI-DAG: DS_READ_B32
; SI-DAG: DS_WRITE_B32
; SI-DAG: DS_READ_B32
; SI-DAG: DS_WRITE_B32
; SI-DAG: DS_READ_B32
; SI-DAG: DS_WRITE_B32
; SI-DAG: DS_READ_B32
; SI-DAG: DS_WRITE_B32
; SI-DAG: S_ENDPGM
define void @test_small_memcpy_i64_lds_to_lds_align8(i64 addrspace(3)* noalias %out, i64 addrspace(3)* noalias %in) nounwind {
%bcin = bitcast i64 addrspace(3)* %in to i8 addrspace(3)*
%bcout = bitcast i64 addrspace(3)* %out to i8 addrspace(3)*
call void @llvm.memcpy.p3i8.p3i8.i32(i8 addrspace(3)* %bcout, i8 addrspace(3)* %bcin, i32 32, i32 8, i1 false) nounwind
ret void
}
R600: Call EmitFunctionHeader() in the AsmPrinter to populate the ELF symbol table git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218776 91177308-0d34-0410-b5e6-96231b3b80d8
2014-10-01 17:15:17 +00:00
; FUNC-LABEL: {{^}}test_small_memcpy_i64_global_to_global_align1:
R600/SI: Implement getOptimalMemOpType The default guess uses i32. This needs an address space argument to really do the right thing in all cases. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214104 91177308-0d34-0410-b5e6-96231b3b80d8
2014-07-28 17:49:26 +00:00
; SI-DAG: BUFFER_LOAD_UBYTE
; SI-DAG: BUFFER_STORE_BYTE
; SI-DAG: BUFFER_LOAD_UBYTE
; SI-DAG: BUFFER_STORE_BYTE
; SI-DAG: BUFFER_LOAD_UBYTE
; SI-DAG: BUFFER_STORE_BYTE
; SI-DAG: BUFFER_LOAD_UBYTE
; SI-DAG: BUFFER_STORE_BYTE
; SI-DAG: BUFFER_LOAD_UBYTE
; SI-DAG: BUFFER_STORE_BYTE
; SI-DAG: BUFFER_LOAD_UBYTE
; SI-DAG: BUFFER_STORE_BYTE
; SI-DAG: BUFFER_LOAD_UBYTE
; SI-DAG: BUFFER_STORE_BYTE
; SI-DAG: BUFFER_LOAD_UBYTE
; SI-DAG: BUFFER_STORE_BYTE
; SI-DAG: BUFFER_LOAD_UBYTE
; SI-DAG: BUFFER_STORE_BYTE
; SI-DAG: BUFFER_LOAD_UBYTE
; SI-DAG: BUFFER_STORE_BYTE
; SI-DAG: BUFFER_LOAD_UBYTE
; SI-DAG: BUFFER_STORE_BYTE
; SI-DAG: BUFFER_LOAD_UBYTE
; SI-DAG: BUFFER_STORE_BYTE
; SI-DAG: BUFFER_LOAD_UBYTE
; SI-DAG: BUFFER_STORE_BYTE
; SI-DAG: BUFFER_LOAD_UBYTE
; SI-DAG: BUFFER_STORE_BYTE
; SI-DAG: BUFFER_LOAD_UBYTE
; SI-DAG: BUFFER_STORE_BYTE
; SI-DAG: BUFFER_LOAD_UBYTE
; SI-DAG: BUFFER_STORE_BYTE
; SI-DAG: BUFFER_LOAD_UBYTE
; SI-DAG: BUFFER_STORE_BYTE
; SI-DAG: BUFFER_LOAD_UBYTE
; SI-DAG: BUFFER_STORE_BYTE
; SI-DAG: BUFFER_LOAD_UBYTE
; SI-DAG: BUFFER_STORE_BYTE
; SI-DAG: BUFFER_LOAD_UBYTE
; SI-DAG: BUFFER_STORE_BYTE
; SI-DAG: BUFFER_LOAD_UBYTE
; SI-DAG: BUFFER_STORE_BYTE
; SI-DAG: BUFFER_LOAD_UBYTE
; SI-DAG: BUFFER_STORE_BYTE
; SI-DAG: BUFFER_LOAD_UBYTE
; SI-DAG: BUFFER_STORE_BYTE
; SI-DAG: BUFFER_LOAD_UBYTE
; SI-DAG: BUFFER_STORE_BYTE
; SI-DAG: BUFFER_LOAD_UBYTE
; SI-DAG: BUFFER_STORE_BYTE
; SI-DAG: BUFFER_LOAD_UBYTE
; SI-DAG: BUFFER_STORE_BYTE
; SI-DAG: BUFFER_LOAD_UBYTE
; SI-DAG: BUFFER_STORE_BYTE
; SI-DAG: BUFFER_LOAD_UBYTE
; SI-DAG: BUFFER_STORE_BYTE
; SI-DAG: BUFFER_LOAD_UBYTE
; SI-DAG: BUFFER_STORE_BYTE
; SI-DAG: BUFFER_LOAD_UBYTE
; SI-DAG: BUFFER_STORE_BYTE
; SI-DAG: BUFFER_LOAD_UBYTE
; SI-DAG: BUFFER_STORE_BYTE
; SI-DAG: BUFFER_LOAD_UBYTE
; SI-DAG: BUFFER_STORE_BYTE
; SI: S_ENDPGM
define void @test_small_memcpy_i64_global_to_global_align1(i64 addrspace(1)* noalias %out, i64 addrspace(1)* noalias %in) nounwind {
%bcin = bitcast i64 addrspace(1)* %in to i8 addrspace(1)*
%bcout = bitcast i64 addrspace(1)* %out to i8 addrspace(1)*
call void @llvm.memcpy.p1i8.p1i8.i64(i8 addrspace(1)* %bcout, i8 addrspace(1)* %bcin, i64 32, i32 1, i1 false) nounwind
ret void
}
R600: Call EmitFunctionHeader() in the AsmPrinter to populate the ELF symbol table git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218776 91177308-0d34-0410-b5e6-96231b3b80d8
2014-10-01 17:15:17 +00:00
; FUNC-LABEL: {{^}}test_small_memcpy_i64_global_to_global_align2:
R600/SI: Implement getOptimalMemOpType The default guess uses i32. This needs an address space argument to really do the right thing in all cases. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214104 91177308-0d34-0410-b5e6-96231b3b80d8
2014-07-28 17:49:26 +00:00
; SI-DAG: BUFFER_LOAD_USHORT
; SI-DAG: BUFFER_LOAD_USHORT
; SI-DAG: BUFFER_LOAD_USHORT
; SI-DAG: BUFFER_LOAD_USHORT
; SI-DAG: BUFFER_LOAD_USHORT
; SI-DAG: BUFFER_LOAD_USHORT
; SI-DAG: BUFFER_LOAD_USHORT
; SI-DAG: BUFFER_LOAD_USHORT
; SI-DAG: BUFFER_LOAD_USHORT
; SI-DAG: BUFFER_LOAD_USHORT
; SI-DAG: BUFFER_LOAD_USHORT
; SI-DAG: BUFFER_LOAD_USHORT
; SI-DAG: BUFFER_LOAD_USHORT
; SI-DAG: BUFFER_LOAD_USHORT
; SI-DAG: BUFFER_LOAD_USHORT
; SI-DAG: BUFFER_LOAD_USHORT
R600: Increase nearby load scheduling threshold. This partially fixes weird looking load scheduling in memcpy test. The load clustering doesn't seem particularly smart, but this method seems to be partially deprecated so it might not be worth trying to fix. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214943 91177308-0d34-0410-b5e6-96231b3b80d8
2014-08-06 00:29:49 +00:00
; SI-DAG: BUFFER_STORE_SHORT
; SI-DAG: BUFFER_STORE_SHORT
; SI-DAG: BUFFER_STORE_SHORT
; SI-DAG: BUFFER_STORE_SHORT
; SI-DAG: BUFFER_STORE_SHORT
; SI-DAG: BUFFER_STORE_SHORT
; SI-DAG: BUFFER_STORE_SHORT
; SI-DAG: BUFFER_STORE_SHORT
; SI-DAG: BUFFER_STORE_SHORT
; SI-DAG: BUFFER_STORE_SHORT
; SI-DAG: BUFFER_STORE_SHORT
; SI-DAG: BUFFER_STORE_SHORT
; SI-DAG: BUFFER_STORE_SHORT
; SI-DAG: BUFFER_STORE_SHORT
; SI-DAG: BUFFER_STORE_SHORT
R600/SI: Implement getOptimalMemOpType The default guess uses i32. This needs an address space argument to really do the right thing in all cases. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214104 91177308-0d34-0410-b5e6-96231b3b80d8
2014-07-28 17:49:26 +00:00
; SI-DAG: BUFFER_STORE_SHORT
; SI: S_ENDPGM
define void @test_small_memcpy_i64_global_to_global_align2(i64 addrspace(1)* noalias %out, i64 addrspace(1)* noalias %in) nounwind {
%bcin = bitcast i64 addrspace(1)* %in to i8 addrspace(1)*
%bcout = bitcast i64 addrspace(1)* %out to i8 addrspace(1)*
call void @llvm.memcpy.p1i8.p1i8.i64(i8 addrspace(1)* %bcout, i8 addrspace(1)* %bcin, i64 32, i32 2, i1 false) nounwind
ret void
}
R600: Call EmitFunctionHeader() in the AsmPrinter to populate the ELF symbol table git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218776 91177308-0d34-0410-b5e6-96231b3b80d8
2014-10-01 17:15:17 +00:00
; FUNC-LABEL: {{^}}test_small_memcpy_i64_global_to_global_align4:
R600/SI: Implement getOptimalMemOpType The default guess uses i32. This needs an address space argument to really do the right thing in all cases. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214104 91177308-0d34-0410-b5e6-96231b3b80d8
2014-07-28 17:49:26 +00:00
; SI: BUFFER_LOAD_DWORDX4
; SI: BUFFER_LOAD_DWORDX4
; SI: BUFFER_STORE_DWORDX4
R600: Increase nearby load scheduling threshold. This partially fixes weird looking load scheduling in memcpy test. The load clustering doesn't seem particularly smart, but this method seems to be partially deprecated so it might not be worth trying to fix. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214943 91177308-0d34-0410-b5e6-96231b3b80d8
2014-08-06 00:29:49 +00:00
; SI: BUFFER_STORE_DWORDX4
R600/SI: Implement getOptimalMemOpType The default guess uses i32. This needs an address space argument to really do the right thing in all cases. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214104 91177308-0d34-0410-b5e6-96231b3b80d8
2014-07-28 17:49:26 +00:00
; SI: S_ENDPGM
define void @test_small_memcpy_i64_global_to_global_align4(i64 addrspace(1)* noalias %out, i64 addrspace(1)* noalias %in) nounwind {
%bcin = bitcast i64 addrspace(1)* %in to i8 addrspace(1)*
%bcout = bitcast i64 addrspace(1)* %out to i8 addrspace(1)*
call void @llvm.memcpy.p1i8.p1i8.i64(i8 addrspace(1)* %bcout, i8 addrspace(1)* %bcin, i64 32, i32 4, i1 false) nounwind
ret void
}
R600: Call EmitFunctionHeader() in the AsmPrinter to populate the ELF symbol table git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218776 91177308-0d34-0410-b5e6-96231b3b80d8
2014-10-01 17:15:17 +00:00
; FUNC-LABEL: {{^}}test_small_memcpy_i64_global_to_global_align8:
R600/SI: Implement getOptimalMemOpType The default guess uses i32. This needs an address space argument to really do the right thing in all cases. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214104 91177308-0d34-0410-b5e6-96231b3b80d8
2014-07-28 17:49:26 +00:00
; SI: BUFFER_LOAD_DWORDX4
; SI: BUFFER_LOAD_DWORDX4
; SI: BUFFER_STORE_DWORDX4
R600: Increase nearby load scheduling threshold. This partially fixes weird looking load scheduling in memcpy test. The load clustering doesn't seem particularly smart, but this method seems to be partially deprecated so it might not be worth trying to fix. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214943 91177308-0d34-0410-b5e6-96231b3b80d8
2014-08-06 00:29:49 +00:00
; SI: BUFFER_STORE_DWORDX4
R600/SI: Implement getOptimalMemOpType The default guess uses i32. This needs an address space argument to really do the right thing in all cases. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214104 91177308-0d34-0410-b5e6-96231b3b80d8
2014-07-28 17:49:26 +00:00
; SI: S_ENDPGM
define void @test_small_memcpy_i64_global_to_global_align8(i64 addrspace(1)* noalias %out, i64 addrspace(1)* noalias %in) nounwind {
%bcin = bitcast i64 addrspace(1)* %in to i8 addrspace(1)*
%bcout = bitcast i64 addrspace(1)* %out to i8 addrspace(1)*
call void @llvm.memcpy.p1i8.p1i8.i64(i8 addrspace(1)* %bcout, i8 addrspace(1)* %bcin, i64 32, i32 8, i1 false) nounwind
ret void
}
R600: Call EmitFunctionHeader() in the AsmPrinter to populate the ELF symbol table git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218776 91177308-0d34-0410-b5e6-96231b3b80d8
2014-10-01 17:15:17 +00:00
; FUNC-LABEL: {{^}}test_small_memcpy_i64_global_to_global_align16:
R600/SI: Implement getOptimalMemOpType The default guess uses i32. This needs an address space argument to really do the right thing in all cases. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214104 91177308-0d34-0410-b5e6-96231b3b80d8
2014-07-28 17:49:26 +00:00
; SI: BUFFER_LOAD_DWORDX4
; SI: BUFFER_LOAD_DWORDX4
; SI: BUFFER_STORE_DWORDX4
R600: Increase nearby load scheduling threshold. This partially fixes weird looking load scheduling in memcpy test. The load clustering doesn't seem particularly smart, but this method seems to be partially deprecated so it might not be worth trying to fix. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214943 91177308-0d34-0410-b5e6-96231b3b80d8
2014-08-06 00:29:49 +00:00
; SI: BUFFER_STORE_DWORDX4
R600/SI: Implement getOptimalMemOpType The default guess uses i32. This needs an address space argument to really do the right thing in all cases. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214104 91177308-0d34-0410-b5e6-96231b3b80d8
2014-07-28 17:49:26 +00:00
; SI: S_ENDPGM
define void @test_small_memcpy_i64_global_to_global_align16(i64 addrspace(1)* noalias %out, i64 addrspace(1)* noalias %in) nounwind {
%bcin = bitcast i64 addrspace(1)* %in to i8 addrspace(1)*
%bcout = bitcast i64 addrspace(1)* %out to i8 addrspace(1)*
call void @llvm.memcpy.p1i8.p1i8.i64(i8 addrspace(1)* %bcout, i8 addrspace(1)* %bcin, i64 32, i32 16, i1 false) nounwind
ret void
}
Reference in New Issue Copy Permalink