archer
archer copied to clipboard
Archer, a data race detection tool for large OpenMP applications
I was able to use the Archer in tr4 branch with clang 4.0.1 to find race in this benchmark: https://github.com/LLNL/dataracebench/blob/master/micro-benchmarks/taskdependmissing-orig-yes.c But a later version of Archer with clang 3.9.1 (installed...
@jprotze I made a branch mac_osx, I tried to deal with the weak symbols and currently, it passes most of the tests except does not catch some races. One thing...
Failing tests: # The task_schedule event returns `status=task_completed` when it should be `task_other`, since the explicit task has not been scheduled yet test/OMPT/parallel/parallel_if0.c test/OMPT/parallel/serialized.c It seems this is due to...
Hi, here is a minimal example ``` #include #include int main() { int a = 2222; #pragma omp parallel for for (int i = 0; i < 10000; i++) {...
With Archer (dc4e363) build with out of source with LLVM 4.0 and OMP-TR4. Running GROMACS unit tests: ``` git init gromacs && cd gromacs git fetch https://gerrit.gromacs.org/gromacs refs/changes/48/6648/1 && git...
Given: ``` #include int main() { __m256 x = _mm256_set1_ps(1); return 0; } ``` And Archer (dc4e3632485) build with out of source with LLVM 4.0 and OMP-TR4 I get: ```...
The following commands create a binary, that is instrumented with TSan, has the Clang-TSan, LLVM-openmp and archer runtime library linked: ``` gfortran -fopenmp -sanitize=thread -c code.f clang -fopenmp -sanitize=thread -lgfortran...
I am moving this issue from #3: This has been common feedback from our early adopters. Instead of seeing individual data race instances, they want to be able to see...
Identifying duplicates in data race reports is quite expensive regarding time. It should be possible to reduce this cost by applying only logging, but no analysis for memory accesses, that...