CUDA: Atomics, Reductions, and Warp Shuffle

CUDA: Atomics, Reductions, and Warp Shuffle
Wednesday, May 13, 2020

On Wednesday, May 13, 2020, NVIDIA will present part 5 of a 9-part CUDA Training Series titled “Atomics, Reductions, and Warp Shuffle”.

This CUDA programming model does not enforce any order of thread execution. This requires attention when performing operations like reductions on the GPU. This module introduces atomic operations, parallel reductions, and warp shuffle, which can help you in cases where exposing parallelism is not immediately obvious due to potential thread contention. After the presentation, there will be a session where participants can complete example exercises meant to reinforce the presented concepts.

Remote Participation
Remote participants can watch the presentations via web broadcast and will have access to the training exercises, but temporary access to the compute systems will be limited as follows:

• Current NERSC users will have Cori-GPU access temporarily added to their accounts.
• Temporary Summit access will not be available for remote participants.

NOTE: Registration is required. Please submit the form below.

If you have any questions, please contact Tom Papatheodore (papatheodore@ornl.gov).