llvm/test/CodeGen/AMDGPU/fcopysign.f64.ll
Tom Stellard d3adac51fc AMDGPU/SI: Enable lanemask tracking in misched
Summary:
This results in higher register usage, but should make it easier for
the compiler to hide latency.

This pass is a prerequisite for some more scheduler improvements, and I
think the increase register usage with this patch is acceptable, because
when combined with the scheduler improvements, the total register usage
will decrease.

shader-db stats:

2382 shaders in 478 tests
Totals:
SGPRS: 48672 -> 49088 (0.85 %)
VGPRS: 34148 -> 34847 (2.05 %)
Code Size: 1285816 -> 1289128 (0.26 %) bytes
LDS: 28 -> 28 (0.00 %) blocks
Scratch: 492544 -> 573440 (16.42 %) bytes per wave
Max Waves: 6856 -> 6846 (-0.15 %)
Wait states: 0 -> 0 (0.00 %)

Depends on D18451

Reviewers: nhaehnle, arsenm

Subscribers: arsenm, llvm-commits

Differential Revision: http://reviews.llvm.org/D18452

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@264876 91177308-0d34-0410-b5e6-96231b3b80d8
2016-03-30 16:35:09 +00:00

41 lines
2.3 KiB
LLVM

; RUN: llc -march=amdgcn -mcpu=SI -verify-machineinstrs < %s | FileCheck -check-prefix=SI -check-prefix=GCN -check-prefix=FUNC %s
; RUN: llc -march=amdgcn -mcpu=tonga -verify-machineinstrs < %s | FileCheck -check-prefix=VI -check-prefix=GCN -check-prefix=FUNC %s
declare double @llvm.copysign.f64(double, double) nounwind readnone
declare <2 x double> @llvm.copysign.v2f64(<2 x double>, <2 x double>) nounwind readnone
declare <4 x double> @llvm.copysign.v4f64(<4 x double>, <4 x double>) nounwind readnone
; FUNC-LABEL: {{^}}test_copysign_f64:
; SI-DAG: s_load_dwordx2 s{{\[}}[[SMAG_LO:[0-9]+]]:[[SMAG_HI:[0-9]+]]{{\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0xb
; SI-DAG: s_load_dwordx2 s{{\[}}[[SSIGN_LO:[0-9]+]]:[[SSIGN_HI:[0-9]+]]{{\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0xd
; VI-DAG: s_load_dwordx2 s{{\[}}[[SMAG_LO:[0-9]+]]:[[SMAG_HI:[0-9]+]]{{\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0x2c
; VI-DAG: s_load_dwordx2 s{{\[}}[[SSIGN_LO:[0-9]+]]:[[SSIGN_HI:[0-9]+]]{{\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0x34
; GCN-DAG: v_mov_b32_e32 v[[VSIGN_HI:[0-9]+]], s[[SSIGN_HI]]
; GCN-DAG: v_mov_b32_e32 v[[VMAG_HI:[0-9]+]], s[[SMAG_HI]]
; GCN-DAG: s_mov_b32 [[SCONST:s[0-9]+]], 0x7fffffff
; GCN-DAG: v_bfi_b32 v[[VRESULT_HI:[0-9]+]], [[SCONST]], v[[VMAG_HI]], v[[VSIGN_HI]]
; GCN-DAG: v_mov_b32_e32 v[[VMAG_LO:[0-9]+]], s[[SMAG_LO]]
; GCN: buffer_store_dwordx2 v{{\[}}[[VMAG_LO]]:[[VRESULT_HI]]{{\]}}
; GCN: s_endpgm
define void @test_copysign_f64(double addrspace(1)* %out, double %mag, double %sign) nounwind {
%result = call double @llvm.copysign.f64(double %mag, double %sign)
store double %result, double addrspace(1)* %out, align 8
ret void
}
; FUNC-LABEL: {{^}}test_copysign_v2f64:
; GCN: s_endpgm
define void @test_copysign_v2f64(<2 x double> addrspace(1)* %out, <2 x double> %mag, <2 x double> %sign) nounwind {
%result = call <2 x double> @llvm.copysign.v2f64(<2 x double> %mag, <2 x double> %sign)
store <2 x double> %result, <2 x double> addrspace(1)* %out, align 8
ret void
}
; FUNC-LABEL: {{^}}test_copysign_v4f64:
; GCN: s_endpgm
define void @test_copysign_v4f64(<4 x double> addrspace(1)* %out, <4 x double> %mag, <4 x double> %sign) nounwind {
%result = call <4 x double> @llvm.copysign.v4f64(<4 x double> %mag, <4 x double> %sign)
store <4 x double> %result, <4 x double> addrspace(1)* %out, align 8
ret void
}