Full native support for Blackwell architecture cards will be available in an upcoming MATLAB release. In the meantime, users can benefit from forward compatibility. Full native support for Blackwell cards is available starting with CUDA 12.8.
MATLAB GPU code is forward compatible to future architectures using a byte-code representation that allows it to be compiled by the graphics driver at runtime. Since R2020b, forward compatibility is an optional feature, as performance and behavior may vary depending on the specific GPU and workflow. See this documentation page for more details on how to enable forward compatibility.
If forward compatibility is enabled then when MATLAB encounters a card with an architecture the GPU code was not originally compiled for there will be a brief one-time delay for GPU computing commands whilst the byte code is compiled and cached.
If you run into a specific limitation using GPU computing in MATLAB with a Blackwell card please contact MathWorks Support.
