search for: marcusma

...generated by DragonEgg and clang below. It shows that clang generates "__isoc99_fscanf", while DragonEgg gives "\01__isoc99_fscanf". We generally use DragonEgg as our compiler frontend, so it is more important. What could be the reason of "\01" issue? Thanks, - Dima. marcusmae at M17xR4:~/forge/kernelgen/tests/behavior/fileio$ cat fileio.c #include <inttypes.h> #include <stdio.h> const char* filename = "fileio.txt"; int main(int argc, char* argv[]) { if (argc != 2) { printf("Test KernelGen support for File I/O\n");...

Hello, I have just started using llvm. I was trying to debug how *clang* generates IR for very simple C testcases (few assignments and if-condsitions). To get a hold on the basic functions, I put some break points on following functions but debugger *didn't stop*: llvm::BasicBlock::Create llvm::BinaryOperator::CreateAdd (design has binary op) llvm::Value::Value llvm::BranchInst::Create

...x i8] c"FRI", [3 x i8] c"SAT", [3 x i8] c"SUN"], align 4096 $ llc -march="nvptx" test.ll -o - // // Generated by LLVM NVPTX Back-End // .version 3.0 .target sm_10, texmode_independent .address_size 32 Unexpected Constant type UNREACHABLE executed at /home/marcusmae/rpmbuild/BUILD/llvm/lib/Target/NVPTX/NVPTXAsmPrinter.cpp:1903! 0 libLLVM-3.2svn.so 0x00007f1bcb71bf0e 1 libLLVM-3.2svn.so 0x00007f1bcb71bd0a 2 libpthread.so.0 0x00007f1bca33ccb0 3 libc.so.6 0x00007f1bc9aa8445 gsignal + 53 4 libc.so.6 0x00007f1bc9aabbab abort + 379 5 libLLVM...

...quot;SUN\00"], align 16 > $ llc -march="nvptx" dayofweek.ll -o - > // > // Generated by LLVM NVPTX Back-End > // > > .version 3.0 > .target sm_10, texmode_independent > .address_size 32 > > Unexpected Constant type > UNREACHABLE executed at > /home/marcusmae/rpmbuild/BUILD/llvm/lib/Target/NVPTX/NVPTXAsmPrinter.cpp:1903! > > NVCC > ===== > > $ nvcc -c -keep dayofweek.cu > $ cat dayofweek.ptx > > .global .align 1 .b8 yweek[28] = > {0x4d,0x4f,0x4e,0x0,0x54,0x55,0x45,0x0,0x57,0x45,0x44,0x0,0x54,0x48,0x55,0x0,0x46,0x52,0x...

...ot;FRI\00", [4 x i8] c"SAT\00", [4 x i8] c"SUN\00"], align 16 $ llc -march="nvptx" dayofweek.ll -o - // // Generated by LLVM NVPTX Back-End // .version 3.0 .target sm_10, texmode_independent .address_size 32 Unexpected Constant type UNREACHABLE executed at /home/marcusmae/rpmbuild/BUILD/llvm/lib/Target/NVPTX/NVPTXAsmPrinter.cpp:1903! NVCC ===== $ nvcc -c -keep dayofweek.cu $ cat dayofweek.ptx .global .align 1 .b8 yweek[28] = {0x4d,0x4f,0x4e,0x0,0x54,0x55,0x45,0x0,0x57,0x45,0x44,0x0,0x54,0x48,0x55,0x0,0x46,0x52,0x49,0x0,0x53,0x41,0x54,0x0,0x53,0x55,0x4e,0x0}; 2...

...ch="nvptx" dayofweek.ll -o - >> // >> // Generated by LLVM NVPTX Back-End >> // >> >> .version 3.0 >> .target sm_10, texmode_independent >> .address_size 32 >> >> Unexpected Constant type >> UNREACHABLE executed at >> /home/marcusmae/rpmbuild/BUILD/llvm/lib/Target/NVPTX/NVPTXAsmPrinter.cpp:1903! >> >> NVCC >> ===== >> >> $ nvcc -c -keep dayofweek.cu >> $ cat dayofweek.ptx >> >> .global .align 1 .b8 yweek[28] = >> {0x4d,0x4f,0x4e,0x0,0x54,0x55,0x45,0x0,0x57,0x45,0x4...

Hi Ashay, Do you need specifically llvm-gfortran that is based on gcc 4.2? Since that, DragonEgg has been introduced - a powerful plugin to gcc that makes it possible to utilize regular gcc compilers as frontends to llvm: http://dragonegg.llvm.org/ It generates Fortran90 programs for me very well. - D. 2011/9/1 Ashay Rane <ashay.rane at tacc.utexas.edu>: > Hello, > I have been

Hello, I have been trying to compile a few small programs [ http://www.esm.psu.edu/~ajm138/fortranexamples.html] using llvm-gfortran. Since I run my own pass on the generated bitcode, I have had to split the compilation process into using llvm-gfortran, opt and then llvm-ld (for linking, as I do with C/C++ programs). For now, I'll drop the `opt' statement. Here's what I see while

Hi Alexander, I had the same error: sincos: /home/marcusmae/rpmbuild/BUILD/llvm/lib/VMCore/PassManager.cpp:635: void llvm::PMTopLevelManager::schedulePass(llvm::Pass*): Assertion `PI && "Expected required passes to be initialized"' failed. The solution was to add: PassRegistry &Registry = *PassRegistry::getPassRegistry(); in...

Dear lazydev, I'm writing an instrumentation pass that depends on AliasAnalysis. My getAnalysisUsage() looks as follows: 2453 void ThreadSanitizer::getAnalysisUsage(AnalysisUsage &AU) const { 2454 AU.addRequired<TargetData>(); 2455 AU.addRequired<AliasAnalysis>(); 2456 } and the pass initialization: 2668 char ThreadSanitizer::ID = 0; 2669

...ly important. The success of parallelization utilities that are currently in a quite good shape (thanks to their developers!) nowadays heavily relies on the code quality: if IR is too rough, we can't do much about it :( Many thanks and Happy New Year! - Dima. 1) Bad LLVM IR: =============== marcusmae at M17xR4:~/forge/kernelgen/tests/behavior/demo_f$ KERNELGEN_FALLBACK=1 kernelgen-gfortran -fplugin=dragonegg.so -fplugin-arg-dragonegg-emit-ir -fplugin-arg-dragonegg-llvm-ir-optimize=0 -S -c demo_f.f90 -o - | opt -O3 -S -o - "161.i": ; preds = %...

...each of 3 arrays a unique TBAA node. But it does not seem to help: alias analysis still considers arrays as may-alias, which most likely prevents the desired optimization. Below is the sample code with TBAA metadata inserted. Could you please suggest what might be wrong with it? Many thanks, - D. marcusmae at M17xR4:~/forge/llvm$ opt -time-passes -enable-tbaa -tbaa -print-alias-sets -O3 check.ll -o - -S Alias Set Tracker: 1 alias sets for 3 pointer values. AliasSet[0x39046c0, 3] may alias, Mod/Ref Pointers: (float* inttoptr (i64 47380979712 to float*), 4), (float* %p_newGEPInst9.cloned, 4), (flo...

Dear all, Attached notunrolled.ll is a module containing reduction kernel. What I'm trying to do is to unroll it in such way, that partial reduction on unrolled iterations would be performed on register, and then stored to memory only once. Currently llvm's unroller together with all standard optimizations produce code, which stores value to memory after every unrolled iteration, which is

...Polly on any use of this array. Attachments: 1) Sample Fortran source code (to be compiled with and without -DMODULE to see failing and working version, respectively). 2) LLVM IR for both cases right before Polly analysis (initialization loop "array = 2") Below is diff for quick look: marcusmae at M17xR4:~/forge/kernelgen/tests/behavior/module_array$ diff -u array.loop.9.ll module_array.loop.9.ll --- array.loop.9.ll 2013-01-04 01:37:40.312259953 +0100 +++ module_array.loop.9.ll 2013-01-04 01:37:50.036259544 +0100 @@ -12,34 +12,39 @@ br label %"10.cloned" "16.clo...