Why should GPU access be integrated with workload scheduling and tenant controls?

1. To permit every tenant to modify host drivers
2. To replace all application-level validation
3. To allocate finite resources and constrain cross-workload access
4. To guarantee unlimited hardware capacity