Closed SaltyChiang closed 1 year ago
For CPUs without SSE or AVX, e.g. Apple M1, bitblt_int_to_flt_cpp will be called if int to float conversion is applied. fmtconv will give a wrong result because of the duplicated val * gain + add_cst.
bitblt_int_to_flt_cpp
fmtconv
val * gain + add_cst
Thank you very much for this bug fix!
For CPUs without SSE or AVX, e.g. Apple M1,
bitblt_int_to_flt_cpp
will be called if int to float conversion is applied.fmtconv
will give a wrong result because of the duplicatedval * gain + add_cst
.