Replies: 2 comments
-
FWIW, the grouping is more important than the bottom part. Especially if you are going past 64 bytes to get to them. Almost everyone has 64B caches now, so everything in the first 64B has the same access time basically. This is also a good reason to group, to avoid going past the 64B size for anything at all if possible. |
Beta Was this translation helpful? Give feedback.
0 replies
-
Right, now is probably the time to reorder struct members, it was useless when they were still changing. |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
In some other discussions the issue of space usage came up. There's a couple very quick fixes for saving some space that could be made safely prior to API freeze.
For 1) there's a lot of:
Most people are running foo/arm64 which will be a 32/64 ABI so both the above is:
rearrange this to group the ints at the bottom and you get:
pack the enumerations, change hints to uint16_t (which seems by the defines to be it's target size) and arrange again and you get:
Beta Was this translation helpful? Give feedback.
All reactions