MATLAB Tensor Core v0.5

This release adds:

• A model of RTX 6000 (Blackwell).
• Models of matrix cores in AMD CDNA 1, 2, and 3.

MATLAB Tensor Core v0.4.1

This release adds:

• A bug fix in the model related subnormal product alignment.
• Stand-alone validation of bf16 tensor core model of the B200, on 100M inputs.
• Validation of the models has grown to 10^7 random inputs.

MATLAB Tensor Core v0.4

This release introduces the following updates and improvements:

• Support for 5th generation tensor cores: added support for the latest tensor core architecture available in NVIDIA Blackwell GPUs, with particular focus on FP8 input formats. Previously, FP8 was using internal FP16 tensor cores where FP8 input vectors would be interleaved.
• Codebase cleanup and refactoring: reduced redundancy and improved overall code readability.

MATLAB Tensor Core v0.3.1

This release contains miscellaneous bug fixes and tweaks.

MATLAB Tensor Core v0.3

This new release adds performance improvements and bug fixes.

We thank Ivan Komarov for reporting inconsistencies between the models and the hardware.

MATLAB Tensor Core v0.2

This release adds:

1. Miscellaneous updates to the models, making them more accurate.
2. Support for Octave.

MATLAB Tensor Core v0.1

Add v0.1 of the tensor core simulator

Co-authored-by: Mantas Mikaitis <m.mikaitis@leeds.ac.uk>
Co-authored-by: Faizan A. Khattak <F.A.Khattak@leeds.ac.uk>