Hmm... I agree that DXVK performance often beats native DX12 on Deck, but it's definitely not because of shader precaching. The whole thing about precaching is that you're trying to solve a problem that doesn't exist when using the native API: shader recompilation.

In the case of Elden Ring, the native performance issues were due to lousy porting practices by FROM which DXVK was able to hack around with some bespoke API translation trickery, much like the kind of game-specific "optimizations" you often see packaged into GPU driver blobs on Windows.