search for: pbqpregalloc

Displaying 8 results from an estimated 8 matches for "pbqpregalloc".

2010 May 08
0
[LLVMdev] Regarding LLVM Compiler
Hi Kameshwar, You're best off sending emails such as this to the LLVM developers mailing list (llvmdev at cs.uiuc.edu). You'll reach a much wider audience. The PBQPRegAlloc::buildInterferenceMatrix method constructs a PBQP cost matrix representing a single interference constraint, which doesn't sound like what you want. It is only used by the PBQP allocator (-regalloc=pbqp), whereas LLVM uses linear scan by default (-regalloc=linscan). That's why you're no...
2010 Jan 31
2
[LLVMdev] Crash in PBQP register allocator
...0 (cost = 1.875000e-02) Selected 0 for node 5 (cost = inf) Selected 1 for node 3 (cost = 0.000000e+00) Selected 1 for node 2 (cost = 0.000000e+00) Selected 1 for node 1 (cost = 0.000000e+00) llc: /home/lhames/Projects/llvm/llvm-broken-pbqp/llvm/lib/CodeGen/RegAllocPBQP.cpp:701: bool<unnamed>::PBQPRegAlloc::mapPBQPToRegAlloc(const PBQP::Solution&): Assertion `solution.getCost() != std::numeric_limits<PBQP::PBQPNum>::infinity() && "Invalid (infinite cost) solution for PBQP problem."' failed. The problem is that node 5 is being allocated an infinite cost option (which i...
2009 Oct 20
7
[LLVMdev] request for help writing a register allocator
...> lib/CodeGen/RegAllocPBQP.cpp, with supporting files in the lib/CodeGen/ > PBQP directory. > Yes - as far as working allocators go PBQP is pretty simple. If you're just interested in LLVM API details you can focus on the lower half of the RegAllocPBQP.cpp source file (everything from PBQPRegAlloc::findVRegIntervalsToAlloc()). The rest of the source (class declaration at the top of RegAllocPBQP.cpp aside) is mostly concerned with PBQP algorithm specifics, such as constructing cost matrices, or carrying out the PBQP graph reduction algorithm. > > Is there anyone who has written an LLV...
2009 Oct 20
0
[LLVMdev] request for help writing a register allocator
Hi Susan, You may find the PBQP allocator implementation useful as a reference to what's involved in adding a new allocator to LLVM. It's located in lib/CodeGen/RegAllocPBQP.cpp, with supporting files in the lib/CodeGen/ PBQP directory. I'm no expert on the LLVM register allocation interfaces, so I'll defer to those who are regarding the specifics of your questions. -Jim
2009 Oct 20
2
[LLVMdev] request for help writing a register allocator
I'm using LLVM for a compiler course, and I'd like to have my students implement a graph-coloring register allocator. I'm having a lot of trouble figuring out how this should be done. Is there anyone who has written an LLVM register allocator who would be willing to help me understand the basic ideas? I understand the algorithm, it's LLVM that I don't understand. For
2010 Feb 01
0
[LLVMdev] Crash in PBQP register allocator
...elected 0 for node 5 (cost = inf) > Selected 1 for node 3 (cost = 0.000000e+00) > Selected 1 for node 2 (cost = 0.000000e+00) > Selected 1 for node 1 (cost = 0.000000e+00) > llc: /home/lhames/Projects/llvm/llvm-broken-pbqp/llvm/lib/CodeGen/RegAllocPBQP.cpp:701: > bool<unnamed>::PBQPRegAlloc::mapPBQPToRegAlloc(const PBQP::Solution&): > Assertion `solution.getCost() != > std::numeric_limits<PBQP::PBQPNum>::infinity() && "Invalid (infinite > cost) solution for PBQP problem."' failed. > > The problem is that node 5 is being allocated an infi...
2010 Jan 28
0
[LLVMdev] Crash in PBQP register allocator
Hi Lang, I'm surprised about the fact that you omit R1/R2 reductions in some cases. Can you give a more detailed description of the bug (e.g. a PBQP dump)? Best regards, Sebastian Lang Hames wrote: > Hi Sachin, llvm-dev, > > I've just committed a new PBQP solver which, among other things, > should take care of this bug. > > Please let me know how it works out for you.
2010 Jan 26
3
[LLVMdev] Crash in PBQP register allocator
Hi Sachin, llvm-dev, I've just committed a new PBQP solver which, among other things, should take care of this bug. Please let me know how it works out for you. Cheers, Lang. On Tue, Dec 15, 2009 at 5:54 PM, Lang Hames <lhames at gmail.com> wrote: > Hi Sachin, > > Yes. Bernhard Scholz and I have just discussed a fix for this. I hope to > commit it in the next few days. I