Dear NVIDIA Technical Support Team,
My name is Alex and I am a developer at a company that produces video processing software. We are currently integrating the NVIDIA Video Codec SDK into our application to leverage hardware-accelerated encoding and decoding.
Currently, our build system produces two separate libraries to maintain broad compatibility:
One library compiled against a legacy version of the SDK (pre-13.0).
A second library compiled against the latest Video Codec SDK (v13.0).
We are now planning to streamline our development and support efforts by migrating entirely to SDK 13.0 and discontinuing support for the older versions. However, before proceeding, we need to conduct a thorough impact analysis regarding hardware support.
I have reviewed the official documentation, I need to clarify which older GPUs are excluded from support in SDK 13.0.
Could you please confirm the following:
Minimum Hardware Requirements: What is the official minimum GPU architecture requirement for Video Codec SDK 13.0? Specifically, does SDK 13.0 still support first-generation Maxwell (e.g., GM107) and second-generation Maxwell (e.g., GM204) GPUs, or has the minimum requirement been raised to a newer architecture like Pascal?
Deprecation List: Is there an official list of GPU models that are no longer compatible with SDK 13.0 due to driver or feature-set deprecation? We want to ensure that dropping our legacy library will not unexpectedly break functionality for users with specific, older hardware.
Fallback Guidance: For any GPUs that are incompatible with SDK 13.0, what is NVIDIA’s recommended strategy? Should applications rely on software fallbacks, or is the expectation that users upgrade their hardware to a supported architecture?
Our goal is to ensure that by moving to SDK 13.0, we are not inadvertently dropping support for a significant portion of our user base without a clear migration path.
Thank you for your time and assistance. We look forward to your response.
Best regards, Alex.