1
0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2025-02-01 05:01:59 +01:00
llvm-mirror/test/CodeGen/AMDGPU/local-memory.amdgcn.ll
Jay Foad 7c4d297102 [AMDGPU] Mark s_barrier as having side effects but not accessing memory.
Summary:
This fixes poor scheduling in a function containing a barrier and a few
load instructions.

Without this fix, ScheduleDAGInstrs::buildSchedGraph adds an artificial
edge in the dependency graph from the barrier instruction to the exit
node representing live-out latency, with a latency of about 500 cycles.
Because of this it thinks the critical path through the graph also has
a latency of about 500 cycles. And because of that it does not think
that any of the load instructions are on the critical path, so it
schedules them with no regard for their (80 cycle) latency, which gives
poor results.

Reviewers: arsenm, dstuttard, tpr, nhaehnle

Subscribers: kzhuravl, jvesely, wdng, yaxunl, t-tye, hiraditya, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D67218

llvm-svn: 371192
2019-09-06 10:07:28 +00:00

81 lines
3.5 KiB
LLVM

; RUN: llc -march=amdgcn -mcpu=verde -verify-machineinstrs < %s | FileCheck -check-prefix=SI -check-prefix=GCN %s
; RUN: llc -march=amdgcn -mcpu=bonaire -verify-machineinstrs < %s | FileCheck -check-prefix=CI -check-prefix=GCN %s
@local_memory.local_mem = internal unnamed_addr addrspace(3) global [128 x i32] undef, align 4
; GCN-LABEL: {{^}}local_memory:
; GCN-NOT: s_wqm_b64
; GCN: ds_write_b32
; GCN: s_barrier
; GCN: ds_read_b32 {{v[0-9]+}},
define amdgpu_kernel void @local_memory(i32 addrspace(1)* %out) #0 {
entry:
%y.i = call i32 @llvm.amdgcn.workitem.id.x() #1
%arrayidx = getelementptr inbounds [128 x i32], [128 x i32] addrspace(3)* @local_memory.local_mem, i32 0, i32 %y.i
store i32 %y.i, i32 addrspace(3)* %arrayidx, align 4
%add = add nsw i32 %y.i, 1
%cmp = icmp eq i32 %add, 16
%.add = select i1 %cmp, i32 0, i32 %add
call void @llvm.amdgcn.s.barrier()
%arrayidx1 = getelementptr inbounds [128 x i32], [128 x i32] addrspace(3)* @local_memory.local_mem, i32 0, i32 %.add
%tmp = load i32, i32 addrspace(3)* %arrayidx1, align 4
%arrayidx2 = getelementptr inbounds i32, i32 addrspace(1)* %out, i32 %y.i
store i32 %tmp, i32 addrspace(1)* %arrayidx2, align 4
ret void
}
@local_memory_two_objects.local_mem0 = internal unnamed_addr addrspace(3) global [4 x i32] undef, align 4
@local_memory_two_objects.local_mem1 = internal unnamed_addr addrspace(3) global [4 x i32] undef, align 4
; Check that the LDS size emitted correctly
; EG: .long 166120
; EG-NEXT: .long 8
; GCN: .long 47180
; GCN-NEXT: .long 32900
; GCN-LABEL: {{^}}local_memory_two_objects:
; GCN: v_lshlrev_b32_e32 [[ADDRW:v[0-9]+]], 2, v0
; CI-DAG: v_sub_i32_e32 [[SUB:v[0-9]+]], vcc, 0, [[ADDRW]]
; CI-DAG: ds_write2_b32 [[ADDRW]], {{v[0-9]+}}, {{v[0-9]+}} offset1:4
; SI-DAG: ds_write2_b32 [[ADDRW]], {{v[0-9]+}}, {{v[0-9]+}} offset1:4
; SI-DAG: v_sub_i32_e32 [[SUB0:v[0-9]+]], vcc, 28, [[ADDRW]]
; SI-DAG: v_sub_i32_e32 [[SUB1:v[0-9]+]], vcc, 12, [[ADDRW]]
; GCN: s_barrier
; SI-DAG: ds_read_b32 v{{[0-9]+}}, [[SUB0]]
; SI-DAG: ds_read_b32 v{{[0-9]+}}, [[SUB1]]
; CI: ds_read2_b32 {{v\[[0-9]+:[0-9]+\]}}, [[SUB]] offset0:3 offset1:7
define amdgpu_kernel void @local_memory_two_objects(i32 addrspace(1)* %out) #0 {
entry:
%x.i = call i32 @llvm.amdgcn.workitem.id.x()
%arrayidx = getelementptr inbounds [4 x i32], [4 x i32] addrspace(3)* @local_memory_two_objects.local_mem0, i32 0, i32 %x.i
store i32 %x.i, i32 addrspace(3)* %arrayidx, align 4
%mul = shl nsw i32 %x.i, 1
%arrayidx1 = getelementptr inbounds [4 x i32], [4 x i32] addrspace(3)* @local_memory_two_objects.local_mem1, i32 0, i32 %x.i
store i32 %mul, i32 addrspace(3)* %arrayidx1, align 4
%sub = sub nsw i32 3, %x.i
call void @llvm.amdgcn.s.barrier()
%arrayidx2 = getelementptr inbounds [4 x i32], [4 x i32] addrspace(3)* @local_memory_two_objects.local_mem0, i32 0, i32 %sub
%tmp = load i32, i32 addrspace(3)* %arrayidx2, align 4
%arrayidx3 = getelementptr inbounds i32, i32 addrspace(1)* %out, i32 %x.i
store i32 %tmp, i32 addrspace(1)* %arrayidx3, align 4
%arrayidx4 = getelementptr inbounds [4 x i32], [4 x i32] addrspace(3)* @local_memory_two_objects.local_mem1, i32 0, i32 %sub
%tmp1 = load i32, i32 addrspace(3)* %arrayidx4, align 4
%add = add nsw i32 %x.i, 4
%arrayidx5 = getelementptr inbounds i32, i32 addrspace(1)* %out, i32 %add
store i32 %tmp1, i32 addrspace(1)* %arrayidx5, align 4
ret void
}
declare i32 @llvm.amdgcn.workitem.id.x() #1
declare void @llvm.amdgcn.s.barrier() #2
attributes #0 = { nounwind }
attributes #1 = { nounwind readnone }
attributes #2 = { convergent nounwind }