[Polly] Add option to limit the maximal number of parallel threads.

[dpeixott]

LGTM with small test case demonstrating the command line option is respected.

[jdoerfert]

Updated, waiting for D4990 to be commited.

[jdoerfert]

Closed by commit rL219004 (authored by @jdoerfert).

lib/CodeGen/LoopGenerators.cpp

 
 #include "polly/ScopDetection.h"
 #include "polly/CodeGen/LoopGenerators.h"
 #include "llvm/Analysis/LoopInfo.h"
 #include "llvm/IR/DataLayout.h"
 #include "llvm/IR/Dominators.h"
 #include "llvm/IR/Module.h"
 #include "llvm/Transforms/Utils/BasicBlockUtils.h"
+#include "llvm/Support/CommandLine.h"
 
 using namespace llvm;
 using namespace polly;
 
+static cl::opt<int>
+    PollyNumThreads("polly-num-threads",
+                    cl::desc("Number of threads to use (0 = auto)"), cl::Hidden,
+                    cl::init(0));
+
 // We generate a loop of either of the following structures:
 //
 //              BeforeBB                      BeforeBB
 //                 |                             |
 //                 v                             v
 //              GuardBB                      PreHeaderBB
 //              /      |                         |   _____
 //     __  PreHeaderBB  |                        v  \/    |
Type *Params[] = {PointerType::getUnqual(FunctionType::get(
                           Builder.getVoidTy(), Builder.getInt8PtrTy(), false)),
                       Builder.getInt8PtrTy(), LongType, LongType, LongType,
                       LongType};
 
     FunctionType *Ty = FunctionType::get(Builder.getVoidTy(), Params, false);
     F = Function::Create(Ty, Linkage, Name, M);
   }
 
-  Value *NumberOfThreads = ConstantInt::get(LongType, 0);
+  Value *NumberOfThreads = ConstantInt::get(LongType, PollyNumThreads);
   Value *Args[] = {SubFn, SubFnParam, NumberOfThreads, LB, UB, Stride};
 
   Builder.CreateCall(F, Args);
 }
 
 Value *ParallelLoopGenerator::createCallGetWorkItem(Value *LBPtr,
                                                     Value *UBPtr) {
   const std::string Name = "GOMP_loop_runtime_next";

test/Isl/CodeGen/openmp_limit_threads.ll

+; RUN: opt %loadPolly -polly-codegen -enable-polly-openmp -S < %s | FileCheck %s --check-prefix=AUTO
+; RUN: opt %loadPolly -polly-codegen -enable-polly-openmp -polly-num-threads=1 -S < %s | FileCheck %s --check-prefix=ONE
+; RUN: opt %loadPolly -polly-codegen -enable-polly-openmp -polly-num-threads=4 -S < %s | FileCheck %s --check-prefix=FOUR
+;
+; AUTO: call void @GOMP_parallel_loop_runtime_start(void (i8*)* @jd.polly.subfn, i8* %polly.par.userContext{{[0-9]*}}, i64 0, i64 0, i64 1024, i64 1)
+; ONE: call void @GOMP_parallel_loop_runtime_start(void (i8*)* @jd.polly.subfn, i8* %polly.par.userContext{{[0-9]*}}, i64 1, i64 0, i64 1024, i64 1)
+; FOUR: call void @GOMP_parallel_loop_runtime_start(void (i8*)* @jd.polly.subfn, i8* %polly.par.userContext{{[0-9]*}}, i64 4, i64 0, i64 1024, i64 1)
+;
+;    void jd(int *A) {
+;      for (int i = 0; i < 1024; i++)
+;        A[i] = 0;
+;    }
+;
+target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
+
+define void @jd(i32* %A) {
+entry:
+  br label %for.cond
+
+for.cond:                                         ; preds = %for.inc, %entry
+  %indvars.iv = phi i64 [ %indvars.iv.next, %for.inc ], [ 0, %entry ]
+  %exitcond = icmp ne i64 %indvars.iv, 1024
+  br i1 %exitcond, label %for.body, label %for.end
+
+for.body:                                         ; preds = %for.cond
+  %arrayidx = getelementptr inbounds i32* %A, i64 %indvars.iv
+  store i32 0, i32* %arrayidx, align 4
+  br label %for.inc
+
+for.inc:                                          ; preds = %for.body
+  %indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
+  br label %for.cond
+
+for.end:                                          ; preds = %for.cond
+  ret void
+}