RPCS3/llvm-mirror
1
0
Fork 0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-10-21 03:53:04 +02:00
Code Issues Projects Releases Wiki Activity
fe00b0505d
llvm-mirror/test/CodeGen/AMDGPU/trunc-cmp-constant.ll
Tom Stellard 24f53ac119 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

llvm-svn: 264876
2016-03-30 16:35:09 +00:00

170 lines
6.3 KiB
LLVM
Raw Blame History

; RUN: llc -march=amdgcn -mcpu=SI -verify-machineinstrs < %s | FileCheck -check-prefix=SI -check-prefix=FUNC %s
; RUN: llc -march=amdgcn -mcpu=tonga -verify-machineinstrs < %s | FileCheck -check-prefix=SI -check-prefix=FUNC %s
; FUNC-LABEL {{^}}sextload_i1_to_i32_trunc_cmp_eq_0:
; SI: buffer_load_ubyte [[LOAD:v[0-9]+]]
; SI: v_and_b32_e32 [[TMP:v[0-9]+]], 1, [[LOAD]]
; SI: v_cmp_eq_i32_e32 vcc, 0, [[TMP]]{{$}}
; SI: v_cndmask_b32_e64
; SI: buffer_store_byte
define void @sextload_i1_to_i32_trunc_cmp_eq_0(i1 addrspace(1)* %out, i1 addrspace(1)* %in) nounwind {
%load = load i1, i1 addrspace(1)* %in
%ext = sext i1 %load to i32
%cmp = icmp eq i32 %ext, 0
store i1 %cmp, i1 addrspace(1)* %out
ret void
}
; FIXME: The negate should be inverting the compare.
; FUNC-LABEL: {{^}}zextload_i1_to_i32_trunc_cmp_eq_0:
; SI: buffer_load_ubyte [[LOAD:v[0-9]+]]
; SI: v_and_b32_e32 [[TMP:v[0-9]+]], 1, [[LOAD]]
; SI: v_cmp_eq_i32_e32 vcc, 1, [[TMP]]{{$}}
; SI-NEXT: s_xor_b64 [[NEG:s\[[0-9]+:[0-9]+\]]], vcc, -1
; SI-NEXT: v_cndmask_b32_e64 [[RESULT:v[0-9]+]], 0, 1, [[NEG]]
; SI: buffer_store_byte [[RESULT]]
define void @zextload_i1_to_i32_trunc_cmp_eq_0(i1 addrspace(1)* %out, i1 addrspace(1)* %in) nounwind {
%load = load i1, i1 addrspace(1)* %in
%ext = zext i1 %load to i32
%cmp = icmp eq i32 %ext, 0
store i1 %cmp, i1 addrspace(1)* %out
ret void
}
; FUNC-LABEL: {{^}}sextload_i1_to_i32_trunc_cmp_eq_1:
; SI: v_mov_b32_e32 [[RESULT:v[0-9]+]], 0{{$}}
; SI: buffer_store_byte [[RESULT]]
define void @sextload_i1_to_i32_trunc_cmp_eq_1(i1 addrspace(1)* %out, i1 addrspace(1)* %in) nounwind {
%load = load i1, i1 addrspace(1)* %in
%ext = sext i1 %load to i32
%cmp = icmp eq i32 %ext, 1
store i1 %cmp, i1 addrspace(1)* %out
ret void
}
; FUNC-LABEL: {{^}}zextload_i1_to_i32_trunc_cmp_eq_1:
; SI: buffer_load_ubyte [[LOAD:v[0-9]+]]
; SI: v_and_b32_e32 [[RESULT:v[0-9]+]], 1, [[LOAD]]
; SI: buffer_store_byte [[RESULT]]
define void @zextload_i1_to_i32_trunc_cmp_eq_1(i1 addrspace(1)* %out, i1 addrspace(1)* %in) nounwind {
%load = load i1, i1 addrspace(1)* %in
%ext = zext i1 %load to i32
%cmp = icmp eq i32 %ext, 1
store i1 %cmp, i1 addrspace(1)* %out
ret void
}
; FUNC-LABEL: {{^}}sextload_i1_to_i32_trunc_cmp_eq_neg1:
; SI: buffer_load_ubyte [[LOAD:v[0-9]+]]
; SI: v_and_b32_e32 [[RESULT:v[0-9]+]], 1, [[LOAD]]
; SI: buffer_store_byte [[RESULT]]
define void @sextload_i1_to_i32_trunc_cmp_eq_neg1(i1 addrspace(1)* %out, i1 addrspace(1)* %in) nounwind {
%load = load i1, i1 addrspace(1)* %in
%ext = sext i1 %load to i32
%cmp = icmp eq i32 %ext, -1
store i1 %cmp, i1 addrspace(1)* %out
ret void
}
; FUNC-LABEL: {{^}}zextload_i1_to_i32_trunc_cmp_eq_neg1:
; SI: v_mov_b32_e32 [[RESULT:v[0-9]+]], 0{{$}}
; SI: buffer_store_byte [[RESULT]]
define void @zextload_i1_to_i32_trunc_cmp_eq_neg1(i1 addrspace(1)* %out, i1 addrspace(1)* %in) nounwind {
%load = load i1, i1 addrspace(1)* %in
%ext = zext i1 %load to i32
%cmp = icmp eq i32 %ext, -1
store i1 %cmp, i1 addrspace(1)* %out
ret void
}
; FUNC-LABEL {{^}}sextload_i1_to_i32_trunc_cmp_ne_0:
; SI: buffer_load_ubyte [[LOAD:v[0-9]+]]
; SI: v_and_b32_e32 [[TMP:v[0-9]+]], 1, [[LOAD]]
; SI: buffer_store_byte [[RESULT]]
define void @sextload_i1_to_i32_trunc_cmp_ne_0(i1 addrspace(1)* %out, i1 addrspace(1)* %in) nounwind {
%load = load i1, i1 addrspace(1)* %in
%ext = sext i1 %load to i32
%cmp = icmp ne i32 %ext, 0
store i1 %cmp, i1 addrspace(1)* %out
ret void
}
; FUNC-LABEL: {{^}}zextload_i1_to_i32_trunc_cmp_ne_0:
; SI: buffer_load_ubyte [[LOAD:v[0-9]+]]
; SI: v_and_b32_e32 [[TMP:v[0-9]+]], 1, [[LOAD]]
; SI: buffer_store_byte [[RESULT]]
define void @zextload_i1_to_i32_trunc_cmp_ne_0(i1 addrspace(1)* %out, i1 addrspace(1)* %in) nounwind {
%load = load i1, i1 addrspace(1)* %in
%ext = zext i1 %load to i32
%cmp = icmp ne i32 %ext, 0
store i1 %cmp, i1 addrspace(1)* %out
ret void
}
; FUNC-LABEL: {{^}}sextload_i1_to_i32_trunc_cmp_ne_1:
; SI: v_mov_b32_e32 [[RESULT:v[0-9]+]], 1{{$}}
; SI: buffer_store_byte [[RESULT]]
define void @sextload_i1_to_i32_trunc_cmp_ne_1(i1 addrspace(1)* %out, i1 addrspace(1)* %in) nounwind {
%load = load i1, i1 addrspace(1)* %in
%ext = sext i1 %load to i32
%cmp = icmp ne i32 %ext, 1
store i1 %cmp, i1 addrspace(1)* %out
ret void
}
; FUNC-LABEL: {{^}}zextload_i1_to_i32_trunc_cmp_ne_1:
; SI: buffer_load_ubyte [[LOAD:v[0-9]+]]
; SI: v_and_b32_e32 [[TMP:v[0-9]+]], 1, [[LOAD]]
; SI: v_cmp_eq_i32_e32 vcc, 1, [[TMP]]{{$}}
; SI-NEXT: s_xor_b64 [[NEG:s\[[0-9]+:[0-9]+\]]], vcc, -1
; SI-NEXT: v_cndmask_b32_e64 [[RESULT:v[0-9]+]], 0, 1, [[NEG]]
; SI: buffer_store_byte [[RESULT]]
define void @zextload_i1_to_i32_trunc_cmp_ne_1(i1 addrspace(1)* %out, i1 addrspace(1)* %in) nounwind {
%load = load i1, i1 addrspace(1)* %in
%ext = zext i1 %load to i32
%cmp = icmp ne i32 %ext, 1
store i1 %cmp, i1 addrspace(1)* %out
ret void
}
; FIXME: This should be one compare.
; FUNC-LABEL: {{^}}sextload_i1_to_i32_trunc_cmp_ne_neg1:
; XSI: buffer_load_ubyte [[LOAD:v[0-9]+]]
; XSI: v_and_b32_e32 [[TMP:v[0-9]+]], 1, [[LOAD]]
; XSI: v_cmp_eq_i32_e64 [[CMP0:s\[[0-9]+:[0-9]+\]]], [[TMP]], 0{{$}}
; XSI-NEXT: v_cndmask_b32_e64 [[RESULT:v[0-9]+]], 0, 1, [[CMP0]]
; XSI-NEXT: buffer_store_byte [[RESULT]]
define void @sextload_i1_to_i32_trunc_cmp_ne_neg1(i1 addrspace(1)* %out, i1 addrspace(1)* %in) nounwind {
%load = load i1, i1 addrspace(1)* %in
%ext = sext i1 %load to i32
%cmp = icmp ne i32 %ext, -1
store i1 %cmp, i1 addrspace(1)* %out
ret void
}
; FUNC-LABEL: {{^}}zextload_i1_to_i32_trunc_cmp_ne_neg1:
; SI: v_mov_b32_e32 [[RESULT:v[0-9]+]], 1{{$}}
; SI: buffer_store_byte [[RESULT]]
define void @zextload_i1_to_i32_trunc_cmp_ne_neg1(i1 addrspace(1)* %out, i1 addrspace(1)* %in) nounwind {
%load = load i1, i1 addrspace(1)* %in
%ext = zext i1 %load to i32
%cmp = icmp ne i32 %ext, -1
store i1 %cmp, i1 addrspace(1)* %out
ret void
}
; FUNC-LABEL: {{^}}masked_load_i1_to_i32_trunc_cmp_ne_neg1:
; SI: buffer_load_sbyte [[LOAD:v[0-9]+]]
; SI: v_cmp_ne_i32_e32 vcc, -1, [[LOAD]]{{$}}
; SI-NEXT: v_cndmask_b32_e64
; SI: buffer_store_byte
define void @masked_load_i1_to_i32_trunc_cmp_ne_neg1(i1 addrspace(1)* %out, i8 addrspace(1)* %in) nounwind {
%load = load i8, i8 addrspace(1)* %in
%masked = and i8 %load, 255
%ext = sext i8 %masked to i32
%cmp = icmp ne i32 %ext, -1
store i1 %cmp, i1 addrspace(1)* %out
ret void
}
Reference in New Issue View Git Blame Copy Permalink