Abstract: Matrix multiplication is a critical computational bottleneck in modern Transformer-based AI systems, particularly within the self-attention and Feed-Forward Network (FFN) layers. Among ...