search for: computehash

...lement NodeEquals() method. Please ignore this message. ------------------ Original ------------------ From: "Xu Zhongxing"<xuzhongxing at foxmail.com>; Date: Mon, Sep 26, 2011 10:11 PM To: "llvmdev"<llvmdev at cs.uiuc.edu>; Subject: [LLVMdev] [LLVMDev] The ComputeHash algorithm in FoldingSet isunsafe I use FoldingSet to save some data structures. And I found somehow it always abandon a new node. I tracked the bug into the ComputeHash() function of FoldingSet. It computes the SAME hash value for these two 32-bit unsigned integer sequences: size: 17...

I use FoldingSet to save some data structures. And I found somehow it always abandon a new node. I tracked the bug into the ComputeHash() function of FoldingSet. It computes the SAME hash value for these two 32-bit unsigned integer sequences: size: 17 4144534024, 32767, 4146895704, 3...

Hi, While profiling LLVM using my test-cases with huge MBBs, I noticed that FoldingSetNodeID operations (ComputeHash,insertion,etc) may become really inefficient for the nodes, which have very many operands. I can give you an example of what is meant by "very many". In my test-case (you can fetch it from here http://llvm.org/bugs/attachment.cgi?id=1275), which is just one HUGE MBB with tens of thou...

...c.edu> Gesendet: Donnerstag, den 24. April 2008, 02:07:55 Uhr Betreff: Re: [LLVMdev] FoldingSetNodeID operations inefficiency On Wed, 23 Apr 2008, Roman Levenstein wrote: > Hi, > > While profiling LLVM using my test-cases with huge MBBs, I noticed that > FoldingSetNodeID operations (ComputeHash,insertion,etc) may become > really inefficient for the nodes, which have very many operands. Here is a stupid idea, which is simple and probably really effective: This only affects you because you have 3200 operands to your TF node. This only affects TF because TF can take an arbitrary number...

...c.edu> Gesendet: Donnerstag, den 24. April 2008, 02:07:55 Uhr Betreff: Re: [LLVMdev] FoldingSetNodeID operations inefficiency On Wed, 23 Apr 2008, Roman Levenstein wrote: > Hi, > > While profiling LLVM using my test-cases with huge MBBs, I noticed that > FoldingSetNodeID operations (ComputeHash,insertion,etc) may become > really inefficient for the nodes, which have very many operands. Here is a stupid idea, which is simple and probably really effective: This only affects you because you have 3200 operands to your TF node. This only affects TF because TF can take an arbitrary number...