-
Enhancement
-
Resolution: Fixed
-
P4
-
19
AArch64 has SVE instruction of populating incrementing indices to vectors. With this we can vectorize some loop expressions with the induction variable in C2 superword optimization.
Below are typical vectorizable loop cases on AArch64 SVE
i)
for (int i = 0; i < count; i++) {
idx[i] = i;
}
ii)
for (int i = 0; i < count; i++) {
dst[i] = src[i] * (i & 7);
}
iii)
for (int i = 0; i < count; i++) {
f[i] = i * i + 100;
}
Below are typical vectorizable loop cases on AArch64 SVE
i)
for (int i = 0; i < count; i++) {
idx[i] = i;
}
ii)
for (int i = 0; i < count; i++) {
dst[i] = src[i] * (i & 7);
}
iii)
for (int i = 0; i < count; i++) {
f[i] = i * i + 100;
}
- relates to
-
JDK-8286972 Support the new loop induction variable related PopulateIndex IR node on x86
-
- Resolved
-
-
JDK-8286125 C2: "bad AD file" with PopulateIndex on x86_64
-
- Resolved
-