maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/polly/test/ScheduleOptimizer/rectangular-tiling.ll

62 lines
2.5 KiB
LLVM
Raw Normal View History

Make -polly-no-tiling work again llvm-svn: 234125
2015-04-06 05:52:21 +08:00
; RUN: opt %loadPolly -polly-detect-unprofitable -polly-opt-isl -analyze -polly-ast -polly-tile-sizes=256,16 < %s | FileCheck %s
Do not use negative option name Instead of -polly-no-tiling, we use -polly-tiling=false to disable tiling. llvm-svn: 245423
2015-08-19 16:22:06 +08:00
; RUN: opt %loadPolly -polly-detect-unprofitable -polly-opt-isl -analyze -polly-tiling=false -polly-ast -polly-tile-sizes=256,16 -polly-no-early-exit < %s | FileCheck %s --check-prefix=NOTILING
Add forgotten 'FileCheck' to tiling test cases These test cases did not verify the CHECK lines at all. We add the FileCheck and also fix some broken CHECK lines. Being here, we extend the checks to cover the whole loop structure. llvm-svn: 232710
2015-03-19 15:39:34 +08:00
Add support for two-level tiling By default we only use one level of tiling for loops, but in general tiling for multiple levels is trivial for us. Hence, we add a set of options that allow people to play with a second level of tiling. If this is profitable for some cases we can work on heuristics that allow us to identify these cases and use two-level tiling for them. llvm-svn: 245563
2015-08-20 21:45:02 +08:00
; RUN: opt %loadPolly -polly-detect-unprofitable -polly-opt-isl -analyze \
; RUN: -polly-2nd-level-tiling -polly-ast \
; RUN: -polly-tile-sizes=256,16 -polly-no-early-exit \
; RUN: -polly-2nd-level-tile-sizes=16,8 < %s | \
; RUN: FileCheck %s --check-prefix=TWOLEVEL
Do not scale tile loops We now generate tile loops as: for (int c1 = 0; c1 <= 47; c1 += 1) for (int c2 = 0; c2 <= 47; c2 += 1) for (int c3 = 0; c3 <= 31; c3 += 1) for (int c4 = 0; c4 <= 31; c4 += 4) #pragma simd for (int c5 = c4; c5 <= c4 + 3; c5 += 1) Stmt_for_body3(32 * c1 + c3, 32 * c2 + c5); instead of for (int c1 = 0; c1 <= 1535; c1 += 32) for (int c2 = 0; c2 <= 1535; c2 += 32) for (int c3 = 0; c3 <= 31; c3 += 1) for (int c4 = 0; c4 <= 31; c4 += 4) #pragma simd for (int c5 = c4; c5 <= c4 + 3; c5 += 1) Stmt_for_body3(c1 + c3, c2 + c5); Run-time performance-wise this makes little difference, but this gives a large reduction in compile time (10-30% on 17 LNT benchmarks). Apparently the isl AST generator is not yet very efficient in generating the latter. llvm-svn: 233675
2015-03-31 15:52:36 +08:00
; CHECK: for (int c0 = 0; c0 <= 3; c0 += 1)
; CHECK: for (int c1 = 0; c1 <= 31; c1 += 1)
Use schedule trees to perform post-scheduling transformations Replacing the old band_tree based code with code that is based on the new schedule tree [1] interface makes applying complex schedule transformations a lot more straightforward. We now do not need to reason about the meaning of flat schedules, but can use a more straightforward tree structure. We do not yet exploit this a lot in the current code, but hopefully we will be able to do so soon. This change also allows us to drop some code, as isl now provides some higher level interfaces to apply loop transformations such as tiling. This change causes some small test case changes as isl uses a slightly different way to perform loop tiling, but no significant functional changes are intended. [1] http://impact.gforge.inria.fr/impact2014/papers/impact2014-verdoolaege.pdf llvm-svn: 232911
2015-03-22 20:06:39 +08:00
; CHECK: for (int c2 = 0; c2 <= 255; c2 += 1)
; CHECK: for (int c3 = 0; c3 <= 15; c3 += 1)
Do not scale tile loops We now generate tile loops as: for (int c1 = 0; c1 <= 47; c1 += 1) for (int c2 = 0; c2 <= 47; c2 += 1) for (int c3 = 0; c3 <= 31; c3 += 1) for (int c4 = 0; c4 <= 31; c4 += 4) #pragma simd for (int c5 = c4; c5 <= c4 + 3; c5 += 1) Stmt_for_body3(32 * c1 + c3, 32 * c2 + c5); instead of for (int c1 = 0; c1 <= 1535; c1 += 32) for (int c2 = 0; c2 <= 1535; c2 += 32) for (int c3 = 0; c3 <= 31; c3 += 1) for (int c4 = 0; c4 <= 31; c4 += 4) #pragma simd for (int c5 = c4; c5 <= c4 + 3; c5 += 1) Stmt_for_body3(c1 + c3, c2 + c5); Run-time performance-wise this makes little difference, but this gives a large reduction in compile time (10-30% on 17 LNT benchmarks). Apparently the isl AST generator is not yet very efficient in generating the latter. llvm-svn: 233675
2015-03-31 15:52:36 +08:00
; CHECK: Stmt_for_body3(256 * c0 + c2, 16 * c1 + c3);
Add forgotten 'FileCheck' to tiling test cases These test cases did not verify the CHECK lines at all. We add the FileCheck and also fix some broken CHECK lines. Being here, we extend the checks to cover the whole loop structure. llvm-svn: 232710
2015-03-19 15:39:34 +08:00
Make -polly-no-tiling work again llvm-svn: 234125
2015-04-06 05:52:21 +08:00
; NOTILING: for (int c0 = 0; c0 <= 1023; c0 += 1)
; NOTILING: for (int c1 = 0; c1 <= 511; c1 += 1)
; NOTILING: Stmt_for_body3(c0, c1);
Add support for two-level tiling By default we only use one level of tiling for loops, but in general tiling for multiple levels is trivial for us. Hence, we add a set of options that allow people to play with a second level of tiling. If this is profitable for some cases we can work on heuristics that allow us to identify these cases and use two-level tiling for them. llvm-svn: 245563
2015-08-20 21:45:02 +08:00
; TWOLEVEL: for (int c0 = 0; c0 <= 3; c0 += 1)
; TWOLEVEL: for (int c1 = 0; c1 <= 31; c1 += 1)
; TWOLEVEL: for (int c2 = 0; c2 <= 15; c2 += 1)
; TWOLEVEL: for (int c3 = 0; c3 <= 1; c3 += 1)
; TWOLEVEL: for (int c4 = 0; c4 <= 15; c4 += 1)
; TWOLEVEL: for (int c5 = 0; c5 <= 7; c5 += 1)
; TWOLEVEL: Stmt_for_body3(256 * c0 + 16 * c2 + c4, 16 * c1 + 8 * c3 + c5);
Added option for n-dimensional rectangular tiling + CL-option --polly-tile-sizes=<int,...,int> The i'th value is used as a tile size for dimension i, if there is no i'th value, the value of --polly-default-tile-size is used + CL-option --polly-default-tile-size=int Used if no tile size is given for a dimension i + 3 Simple testcases llvm-svn: 209753
2014-05-29 01:21:02 +08:00
target datalayout = "e-m:e-p:32:32-i64:64-v128:64:128-n32-S64"
; Function Attrs: nounwind
define void @rect([512 x i32]* %A) {
entry:
br label %entry.split
entry.split: ; preds = %entry
br label %for.body3.lr.ph
for.body3.lr.ph: ; preds = %for.inc5, %entry.split
%i.0 = phi i32 [ 0, %entry.split ], [ %inc6, %for.inc5 ]
br label %for.body3
for.body3: ; preds = %for.body3.lr.ph, %for.body3
%j.0 = phi i32 [ 0, %for.body3.lr.ph ], [ %inc, %for.body3 ]
%mul = mul nsw i32 %j.0, %i.0
%rem = srem i32 %mul, 42
Update Polly tests to handle explicitly typed gep changes in LLVM llvm-svn: 230784
2015-02-28 03:20:19 +08:00
%arrayidx4 = getelementptr inbounds [512 x i32], [512 x i32]* %A, i32 %i.0, i32 %j.0
Added option for n-dimensional rectangular tiling + CL-option --polly-tile-sizes=<int,...,int> The i'th value is used as a tile size for dimension i, if there is no i'th value, the value of --polly-default-tile-size is used + CL-option --polly-default-tile-size=int Used if no tile size is given for a dimension i + 3 Simple testcases llvm-svn: 209753
2014-05-29 01:21:02 +08:00
store i32 %rem, i32* %arrayidx4, align 4
%inc = add nsw i32 %j.0, 1
%cmp2 = icmp slt i32 %inc, 512
br i1 %cmp2, label %for.body3, label %for.inc5
for.inc5: ; preds = %for.body3
%inc6 = add nsw i32 %i.0, 1
%cmp = icmp slt i32 %inc6, 1024
br i1 %cmp, label %for.body3.lr.ph, label %for.end7
for.end7: ; preds = %for.inc5
ret void
}