llama.cpp

76d66ee0 - CUDA: faster q2_K, q3_K MMQ + int8 tensor cores (#7921)

CUDA: faster q2_K, q3_K MMQ + int8 tensor cores (#7921) * CUDA: faster q2_K, q3_K MMQ + int8 tensor cores * try CI fix * try CI fix * try CI fix * fix data race * rever q2_K precision related changes