You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Add support for using more than 1 provider for creating a post init job.
The motivation for this are users with multiple gpus on their systems. For example, a laptop with 2 gpus or a desktop with 2 discrete gpu cards and a supported gpu on the motherboard.
We'd like to enable these users to create a post init in less time and fully utilize their system gpu compute resources.
Add API method where user can provide an array of distinct system providers for creating a post job. The implementation should utilize all providers in parallel to create the post init data. Note that the gpu-post lib already supports executions on multiple providers in parallel.
Add support for using more than 1 provider for creating a post init job.
The motivation for this are users with multiple gpus on their systems. For example, a laptop with 2 gpus or a desktop with 2 discrete gpu cards and a supported gpu on the motherboard.
We'd like to enable these users to create a post init in less time and fully utilize their system gpu compute resources.
Add API method where user can provide an array of distinct system providers for creating a post job. The implementation should utilize all providers in parallel to create the post init data. Note that the gpu-post lib already supports executions on multiple providers in parallel.
@moshababo
The text was updated successfully, but these errors were encountered: