CUDA Managed Memory

CUDA Managed Memory
Thursday, June 18, 2020

On Thursday, June 18, 2020, NVIDIA will present part 6 of a 9-part CUDA Training Series titled “Managed Memory”.

Unified Memory creates a pool of managed memory that is shared between the CPU and GPU, bridging the CPU-GPU divide. Managed memory is accessible to both the CPU and GPU using a single pointer. The system will automatically migrate data between the host and device as needed. This module will show you how to allocate managed memory, as well as how to use it to store data structures like linked lists or C++ objects that might otherwise be difficult to manage manually in a heterogeneous system. After the presentation, there will be a hands-on 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 for remote participation. To register, please submit the form below.

If you have any questions, please contact Tom Papatheodore ([email protected]).

[tw-tab]

[tw-tab]

[tw-tab]

(slides | recording)

[tw-tab]

The example exercises for this module can be found in the exercises/hw6 directory of the following GitHub repo: https://github.com/olcf/cuda-training-series.

[tw-tab] [/tw-tab] [/tw-tabs]