update memory calcualtions

count each layer independently when deciding gpu offloading
This commit is contained in:
Michael Yang
2024-03-18 10:45:22 +01:00
parent d338d70492
commit 91b3e4d282
7 changed files with 125 additions and 89 deletions

View File

@@ -14,6 +14,9 @@ type GpuInfo struct {
// Optional variant to select (e.g. versions, cpu feature flags)
Variant string `json:"variant,omitempty"`
// MinimumMemory represents the minimum memory required to use the GPU
MinimumMemory int64 `json:"-"`
// TODO add other useful attributes about the card here for discovery information
}