riscv time

[grigohas]

Hello, can someone explain why the loading time of a model in llama-cli in riscv with vector enabled is less than without vector extsnsion?

[slaren]

It shouldn't make a significant difference. You may be seeing other effects such as the model already being cached in the system cache.

[grigohas]

there is a a difference of 119,821.82 ms . I want to ask if there is a different quantization when the rvv is enabled