Search for: All records

Alongside the Chicxulub meteorite impact, Deccan volcanism is considered a primary trigger for the Cretaceous–Paleogene (K–Pg) mass extinction. Models suggest that volcanic outgassing of carbon and sulfur—potent environmental stressors—drove global temperature change, but the relative timing, duration, and magnitude of such change remains uncertain. Here, we use the organic paleothermometer MBT′_5meand the carbon-isotope composition of two K–Pg-spanning lignites from the western Unites States, to test models of volcanogenic air temperature change in the ~100 kyr before the mass extinction. Our records show long-term warming of ~3°C, probably driven by Deccan CO₂emissions, and reveal a transient (<10 kyr) ~5°C cooling event, coinciding with the peak of the Poladpur “pulse” of Deccan eruption ~30 kyr before the K–Pg boundary. This cooling was likely caused by the aerosolization of volcanogenic sulfur. Temperatures returned to pre-event values before the mass extinction, suggesting that, from the terrestrial perspective, volcanogenic climate change was not the primary cause of K–Pg extinction. 

Transparently checkpointing MPI for fault tolerance and load balancing is a long-standing problem in HPC. The problem has been complicated by the need to provide checkpoint-restart services for all combinations of an MPI implementation over all network interconnects. This work presents MANA (MPI-Agnostic Network-Agnostic transparent checkpointing), a single code base which supports all MPI implementation and interconnect combinations. The agnostic properties imply that one can checkpoint an MPI application under one MPI implementation and perhaps over TCP, and then restart under a second MPI implementation over InfiniBand on a cluster with a different number of CPU cores per node. This technique is based on a novel "split-process" approach, which enables two separate programs to co-exist within a single process with a single address space. This work overcomes the limitations of the two most widely adopted transparent checkpointing solutions, BLCR and DMTCP/InfiniBand, which require separate modifications to each MPI implementation and/or underlying network API. The runtime overhead is found to be insignificant both for checkpoint-restart within a single host, and when comparing a local MPI computation that was migrated to a remote cluster against an ordinary MPI computation running natively on that same remote cluster.