This PR aligns any object size to 16 bytes. As the smallest size class is 8 bytes, we may allocate more memory than before, and could cause slowdown. MMTk assumes the allocation size is a multiple of VM::MIN_ALIGNMENT, which is defined as 4 for Julia (https://github.com/mmtk/mmtk-julia/blob/baeffd2d4e809289ecd962795142dd02bf59f8a7/mmtk/src/lib.rs#L43). Aligning to 4 bytes should be enough for MMTk, and aligning to 16 bytes may be unnecessary unless Julia requires it.

This PR aligns any object size to 16 bytes. As the smallest size class is 8 bytes, we may allocate more memory than before, and could cause slowdown. MMTk assumes the allocation size is a multiple of VM::MIN_ALIGNMENT, which is defined as 4 for Julia (https://github.com/mmtk/mmtk-julia/blob/baeffd2d4e809289ecd962795142dd02bf59f8a7/mmtk/src/lib.rs#L43). Aligning to 4 bytes should be enough for MMTk, and aligning to 16 bytes may be unnecessary unless Julia requires it.

8-byte aligned classes are used for strings:

This PR aligns any object size to 16 bytes. As the smallest size class is 8 bytes, we may allocate more memory than before, and could cause slowdown. MMTk assumes the allocation size is a multiple of VM::MIN_ALIGNMENT, which is defined as 4 for Julia (https://github.com/mmtk/mmtk-julia/blob/baeffd2d4e809289ecd962795142dd02bf59f8a7/mmtk/src/lib.rs#L43). Aligning to 4 bytes should be enough for MMTk, and aligning to 16 bytes may be unnecessary unless Julia requires it.

8-byte aligned classes are used for strings:

julia/src/julia_internal.h

Line 371 in 5c406d9

// the 8-byte aligned pools are only used for Strings

I think everything else is aligned to 16. I'll change the code to make sure I use that alignment only for strings.

I am actually a bit confused by the comments in the code above ("the 8-byte aligned pools are only used for Strings"). They have 8 bytes aligned pools all the way up to 136 bytes. Does that mean all the objects in that size range are strings? That doesn't sound right.

Edit: Nvm. I got it.

For the current code,

• String size is aligned up to 16 bytes (but this will be fixed so string size is unchanged -- aligned to 8 bytes).
• Objects between 16 bytes to 136 bytes should have the same allocation size as before.
• Objects larger than 144 bytes may have a smaller size.

Can you get performance results for the PR (comparing with master)?

Can you get performance results for the PR (comparing with master)?

Yes, I was doing it already, without considering the latest changes, but I'll re-run everything and put the results here once I get them.