DroplessArena: Allocate objects from the end of memory chunk #77014
Conversation
Allocating from the end of memory chunk simplifies the alignment code and reduces the number of checked arithmetic operations.
|
@bors try @rust-timer queue I wanted to ask "Why does this not updated TypedArena?" but I then realized that TypedArena always increments by a constant size because it's uniformly sized, no need to worry about alignment. |
|
Awaiting bors try build completion |
|
⌛ Trying commit c7e887c with merge 022b652b3fbd002cda8094097c922fba0b3fbf53... |
|
☀️ Try build successful - checks-actions, checks-azure |
|
Queued 022b652b3fbd002cda8094097c922fba0b3fbf53 with parent e0bf356, future comparison URL. |
|
@Mark-Simulacrum another important point is that TypedArena always deallocates objects starting from the start. But the deallocation code for u8 is thankfully a no op, and not called anyways (clear is not called), so it's no big deal. |
|
Finished benchmarking try commit (022b652b3fbd002cda8094097c922fba0b3fbf53): comparison url. Benchmarking this pull request likely means that it is perf-sensitive, so we're automatically marking it as not fit for rolling up. Please note that if the perf results are neutral, you should likely undo the rollup=never given below by specifying Importantly, though, if the results of this run are non-neutral do not roll this PR up -- it will mask other regressions or improvements in the roll up. @bors rollup=never |
|
Looks like a minor one on some benchmarks. Code looks good to me, and has "less operations" so seems good -- @bors r+ |
|
📌 Commit c7e887c has been approved by |
|
☀️ Test successful - checks-actions, checks-azure |
Allocating from the end of memory chunk simplifies the alignment code
and reduces the number of checked arithmetic operations.