tardis-sn/tardis: TARDIS v3.0.dev3442

Changes: b16b1eda78936fa2264684e2bd9196ea811a9888 Fixed crash when integrating in parallel. Crash was previously caused by a double free error. Removing the last call to free caused the spectrum to be computed incorrectly, indicating that the structure of the parallelism was at fault. I removed the...

Full description

Bibliographic Details
Published in:Canadian Journal of Fisheries and Aquatic Sciences
Main Authors: Kerzendorf, Wolfgang, Sim, Stuart, Vogl, Christian, Williamson, Marc, Pássaro, Ezequiel, Flörs, Andreas, Camacho, Yssa, Jančauskas, Vytautas, Harpole, Alice, Nöbauer, Ulrich, Lietzau, Stefan, Mishin, Mikhail, Tsamis, Fotis, Boyle, Aoife, Shingles, Luke, Gupta, Vaibhav, Desai, Karan, Klauser, Michael, Beaujean, Frederik, Suban-Loewen, Adam, Heringer, Epson, Barna, Barnabás, Gautam, Gaurav, Barbosa, Talytha, Patel, Maryam, Varanasi, Kaushik, Reinecke, Martin, Eweis, Youssef, Bylund, Tomas, Bentil, Laud, Livneh, Ran, Rajagopalan, Srinath, Reichenbach, John, Singhal, Jaladh, Jain, Rinkle, Singh, Sourav, Pipelines, Azure, Kowalski, Nathan, Sofiatti, Caroline, Talegaonkar, Chinmay, Selsing, Jonatan, Savel, Arjun, Nayak, Ashwin, Wahi, Ujjwal, Kumar, Atul, Mishra, Sashank, Dasgupta, Debajyoti, Gillanders, James, Sharma, Sampark, Eguren, Jordi, Gupta, Suyash, Yap, Kevin, Magee, Mark, Sarafina, Nance, Patra, Nilesh, Patel, Pratik
Format: Other/Unknown Material
Language:unknown
Published: Zenodo 2020
Subjects:
Online Access:https://doi.org/10.5281/zenodo.3842125
Description
Summary:Changes: b16b1eda78936fa2264684e2bd9196ea811a9888 Fixed crash when integrating in parallel. Crash was previously caused by a double free error. Removing the last call to free caused the spectrum to be computed incorrectly, indicating that the structure of the parallelism was at fault. I removed the firstprivate declaration from the first call to omp parallel as there was no need to make copies of the pointers for each thread (was this even a reasonable thing to do since each thread would have to reallocate the arrays anyway?). There also did not appear to be any need to maintain copies of L or exp_tau for each thread. I then moved the last call to free outside of the parallel section. So now there is no memory allocation or deallocation of L or exp_tau in a parallel block, and exp_tau is freed only once as it should be (#1159) cddd9a62f3ab25298ba312639c3020dadd995c1d add codecov to azure pipelines (#1155) d47f1bae00f58da05802ce35c45ecee5e365a4da Update .mailmap (#1157) This list of changes was auto generated .