dbcsr
dbcsr copied to clipboard
cp2k
DBCSR: Distributed Block Compressed Sparse Row matrix library
Running cp2k test cases QS/regtest-gw/evGW_OH_PBE.inp and evGW_OH_PBE_svd.inp the dangling pointer matrix%col_i is dereferenced at dbcsr_block_access.F line 597 CALL merge_index_arrays(new_row_i, new_col_i, new_blk_p, nblks_new, & old_row_i, matrix%col_i, matrix%blk_p, nblks_old, & For first...
https://github.com/cp2k/dbcsr/blob/21dae0fbf6027aaa935eaa53cb024fc9c42573eb/src/acc/libsmm_acc/libsmm_acc_init.cpp#L37-L41 https://github.com/cp2k/dbcsr/blob/21dae0fbf6027aaa935eaa53cb024fc9c42573eb/src/acc/libsmm_acc/libsmm_acc_init.cpp#L80-L85 ... if the init/finalize is meant to be called from inside a parallel region, then guarding by `omp_in_parallel()` might be a good idea, but if called from outside...
Running many test cases of cp2k, among them xTB/regtest-3/h2o_rtp.inp, in dbcsr_mm_multrec.F line 518 CALL dbcsr_mm_csr_multiply(this%csr, left, right, & ... use_eps=this%use_eps, row_max_epss=this%row_max_epss, & the dangling pointer this%row_max_epss is referenced. The pointer...
There are various variables declared as `SAVE`, for example here: https://github.com/cp2k/dbcsr/blob/0a1ce2e81143f5a36929f07e47eb18cc7b6db6bf/src/work/dbcsr_work_operations.F#L1412 The AMD flang compiler for example does not like it (even though it should be standard) and since those...
Since about around PR #370 , the CI test `dbcsr_tensor_unittest` fails with a segfault, when compiled with the Cray compiler. I've run tests and could determine that the fail is...
Remove check for maximum kernel size in dbcsr_acc_operations.F, take maximum size from libcusmm.
DBCSR implements a limit to the amount of data MPI can allocate in one go [here](https://github.com/cp2k/dbcsr/blob/develop/src/mpi/dbcsr_mpiwrap.F#L6082-L6083). Seemingly, this limit is due to the fact that MPI functions take an int...
@juerghutter did some benchmarking with CP2K comparing same-number of MPI ranks with same number of OpenMP threads and it seems there are some functions (especially iterators) which are completely OpenMP...
Hi, I have a few questions, regarding how to batch tensors. In my program, tensor contractions are batched in chunks, similarly to how it is done in CP2K, [here](https://github.com/cp2k/cp2k/blob/master/src/hfx_ri.F). The...
