Setting Compute Type
Comput modes
There are two types of compute modes in Zerve
Serverless - This is the default setting. This is slower as executors would spin up when code run starts and stop once code run finishes.
Persistent - Uses a persistent execution infrastructure rather than short-lived environments like Lambda invocations. This takes around a minute to start, but execution is faster.
Serverless mode - Set default compute
When you create a new block, by default the compute type is Lambda. You can change this in "Compute Setting" on the left side as shown in the picture below.

Persistent mode - Setup
In Persistent mode, you only have an option to select Fargate compute. You can select different instance type as per your requirement. It takes about a minute to start which and the status can be seen in the tool bar in the bottom .Following video shows how this can be started and where to monitor it's status. You will see that block run will be much quicker once persistent is selected.
Set block level compute - Serverless and Persistent
In both Serverless and Persistent compute modes, you can set different compute type for different blocks in Zerve. This allows you to use cheaper Lambda for smaller operations and comparitively expensive CPU/GPU only in places you need them.
The following video shows you can set block level compute
Last updated

