This presentation discusses emulation techniques for matrix multiplication with floating-point entries. General Matrix-Matrix Multiplication (GEMM), supported by the Basic Linear Algebra Subprograms (BLAS), is a highly optimized routine with hardware-specific tuning. Several prior studies have proposed methods based on error-free matrix splitting. In this talk, we introduce a technique that utilizes the Chinese Remainder Theorem (CRT) in combination with GEMM, and present both the methodology and its actual performance.