InternPool: improve hashing performance

Key.PtrType is now an extern struct so that hashing it can be done by
reinterpreting bytes directly. It also uses the same representation for
type_pointer Tag encoding and the Key. Accessing pointer attributes now
requires packed struct access, however, many operations are now a copy
of a u32 rather than several independent fields.

This function moves the top two most used Key variants - pointer types
and pointer values - to use a single-shot hash function that branches
for small keys instead of calling memcpy.

As a result, perf against merge-base went from 1.17x ± 0.04 slower to
1.12x ± 0.04 slower. After the pointer value hashing was changed, total
CPU instructions spent in memcpy went from 4.40% to 4.08%, and after
additionally improving pointer type hashing, it further decreased to
3.72%.
This commit is contained in:
Andrew Kelley
2023-05-30 20:23:51 -07:00
parent c7d65fa368
commit 82f6f164a1
8 changed files with 527 additions and 392 deletions

View File

@@ -673,7 +673,7 @@ fn lowerParentPtr(
mod.intern_pool.typeOf(elem.base).toType().elemType2(mod).abiSize(mod))),
),
.field => |field| {
const base_type = mod.intern_pool.indexToKey(mod.intern_pool.typeOf(field.base)).ptr_type.elem_type;
const base_type = mod.intern_pool.indexToKey(mod.intern_pool.typeOf(field.base)).ptr_type.child;
return lowerParentPtr(
bin_file,
src_loc,
@@ -681,7 +681,7 @@ fn lowerParentPtr(
code,
debug_output,
reloc_info.offset(switch (mod.intern_pool.indexToKey(base_type)) {
.ptr_type => |ptr_type| switch (ptr_type.size) {
.ptr_type => |ptr_type| switch (ptr_type.flags.size) {
.One, .Many, .C => unreachable,
.Slice => switch (field.index) {
0 => 0,