On 5/15/19 1:31 PM, David Hildenbrand wrote:
> +#define DEF_VFEE(BITS)
>
Same comment wrt inline functions applies.
Here, because there's one result, writing to byte 7, I wonder if it isn't
clearer to write the loop
first_equal = n;
first_zero = n;
for (i = n - 1; i >= 0; --i) {
if (data1 == data2) {
first_equal = i;
}
if (data1 == 0) {
first_zero = i;
}
}
// As an aside, there are bit tricks for the above,
// but let's stay simple(r) for now.
if (zs) {
if (first_equal < first_zero) {
cc = (first_zero < n ? 2 : 1);
} else {
first_equal = first_zero;
cc = (first_zero < n ? 0 : 3);
}
} else {
cc = (first_equal < n ? 1 : 3);
}
s390_vec_write_element64(v1, 0, first_equal);
s390_vec_write_element64(v1, 1, 0);
Note that you don't need S390Vector tmp, since the result is written after all
of the inputs are consumed.
r~
