[llvm-commits] [llvm] r162222 - /llvm/trunk/lib/Target/ARM/ARMISelLowering.cpp

[Sean Silva]

Derp, well, I mean where the small storage is searched by hash vs.
linear. I'm guessing that a SmallDenseSet would not be too hard to
write up based on SmallDenseMap.

--Sean Silva

On Mon, Aug 20, 2012 at 11:19 PM, Sean Silva <silvas at purdue.edu> wrote:
> Would SmallDenseMap be more efficient here since SmallPtrSet does a
> linear scan? There's also a SmallPtrSet of size 64 in the DAG combiner
> that I think would significantly benefit from becoming a SmallDenseMap
> (64 means that when it is pretty large but still in its small storage
> it does a linear scan over up to 8 cache lines).
>
> More generally, in what situations is a SmallPtrSet going to be
> preferable to a SmallDenseMap?
>
> --Sean Silva
>
> On Mon, Aug 20, 2012 at 4:52 PM, Jakob Stoklund Olesen <stoklund at 2pi.dk> wrote:
>> Author: stoklund
>> Date: Mon Aug 20 15:52:03 2012
>> New Revision: 162222
>>
>> URL: http://llvm.org/viewvc/llvm-project?rev=162222&view=rev
>> Log:
>> Use a SmallPtrSet to dedup successors in EmitSjLjDispatchBlock.
>>
>> The test case ARM/2011-05-04-MultipleLandingPadSuccs.ll was creating
>> duplicate successor list entries.
>>
>> Modified:
>>     llvm/trunk/lib/Target/ARM/ARMISelLowering.cpp
>>
>> Modified: llvm/trunk/lib/Target/ARM/ARMISelLowering.cpp
>> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/ARM/ARMISelLowering.cpp?rev=162222&r1=162221&r2=162222&view=diff
>> ==============================================================================
>> --- llvm/trunk/lib/Target/ARM/ARMISelLowering.cpp (original)
>> +++ llvm/trunk/lib/Target/ARM/ARMISelLowering.cpp Mon Aug 20 15:52:03 2012
>> @@ -6151,13 +6151,12 @@
>>    }
>>
>>    // Add the jump table entries as successors to the MBB.
>> -  MachineBasicBlock *PrevMBB = 0;
>> +  SmallPtrSet<MachineBasicBlock*, 8> SeenMBBs;
>>    for (std::vector<MachineBasicBlock*>::iterator
>>           I = LPadList.begin(), E = LPadList.end(); I != E; ++I) {
>>      MachineBasicBlock *CurMBB = *I;
>> -    if (PrevMBB != CurMBB)
>> +    if (SeenMBBs.insert(CurMBB))
>>        DispContBB->addSuccessor(CurMBB);
>> -    PrevMBB = CurMBB;
>>    }
>>
>>    // N.B. the order the invoke BBs are processed in doesn't matter here.
>>
>>
>> _______________________________________________
>> llvm-commits mailing list
>> llvm-commits at cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits