Slightly optimize hash map stable hashing
I was profiling some of the `rustc-perf` benchmarks locally and noticed that quite some time is spent inside the stable hash of hashmaps. I tried to use a `SmallVec` instead of a `Vec` there, which helped very slightly.
Then I tried to remove the sorting, which was a bottleneck, and replaced it with insertion into a binary heap. Locally, it yielded nice improvements in instruction counts and RSS in several benchmarks for incremental builds. The implementation could probably be much nicer and possibly extended to other stable hashes, but first I wanted to test the perf impact properly.
Can I ask someone to do a perf run? Thank you!
Particularly for ctfe-stress-4, the hashing of byte slices as part of the
MIR Allocation is quite hot. Previously, we were falling back on byte-by-byte
copying of the slice into the SipHash buffer (64 bytes long) before hashing a 64
byte chunk, and then doing that again and again.
This should hopefully be an improvement for that code.
The PR had some unforseen perf regressions that are not as easy to find.
Revert the PR for now.
This reverts commit 6ae8912a3e, reversing
changes made to 86d6d2b738.
SipHasher128 implements short_write in an endian-independent way, yet
its write_xxx Hasher trait methods undo this endian-independence by byte
swapping the integer inputs on big-endian hardware. StableHasher then
adds endian-independence back by also byte-swapping on big-endian
hardware prior to invoking SipHasher128.
This double swap may have the appearance of being a no-op, but is in
fact by design. In particular, we really do want SipHasher128 to be
platform-dependent, in order to be consistent with the libstd SipHasher.
Try to clarify this intent. Also, add and update a couple of unit tests.
