This document proposes a further simplification of Cross-Component Linear Model (CCLM) originally proposed in [1]. This contribution consists of two parts: reduction of bit-depth of parameter α and size reduction of LUT that is used for integer division implementation. Low-bit LUT is removed completely, high-bit LUT precision is lowered up to 8 bits as well as several methods to reduce the number of its entries are considered. This proposal reportedly shows that a non-linear sampling of LUT provides the overall results of 0.01% (Y), -0.08% (U), -0.04% (V) for AI configuration and 0.01% (Y), -0.10% (U), -0.08% (V) for RA configuration. Regular LUT sampling for 32 entries is reported to provide the following results: 0.08% (Y), 0.34% (U), 0.41% (V) for AI configuration and 0.06% (Y), 0.46% (U), 0.57% (V) for RA configuration. Regular LUT sampling for 64 entries is reported to provide the following results: 0.03% (Y), 0.09% (U), 0.11% (V) for AI configuration and 0.03% (Y), 0.15% (U), 0.24% (V) for RA configuration.