commit ec0f76c5996e88f61d376640bf36ed7feb2b0ea6 (tree)
parent 11489bb04f82cf91e23bd1980fb4cf49b5a6fc54
Author: Ryan Liptak <squeek502@hotmail.com>
Date: Tue, 3 Oct 2023 17:37:09 -0700
GeneralPurposeAllocator.searchBucket: check current bucket before searching the list
Follow up to #17383. This is a minor optimization that only matters when a small allocation is resized/free'd soon after it is allocated.
The only real difference I was able to observe with this was via a synthetic benchmark that allocates a full bucket and then frees all but one of the slots, over and over in a loop:
Debug build:
Benchmark 1 (9 runs): gpa-degen-master.exe
measurement mean ± σ min … max outliers delta
wall_time 575ms ± 5.19ms 569ms … 583ms 0 ( 0%) 0%
peak_rss 43.8MB ± 1.37KB 43.8MB … 43.8MB 1 (11%) 0%
Benchmark 2 (10 runs): gpa-degen-search-cur.exe
measurement mean ± σ min … max outliers delta
wall_time 532ms ± 5.55ms 520ms … 539ms 0 ( 0%) ⚡- 7.5% ± 0.9%
peak_rss 43.8MB ± 65.2KB 43.8MB … 44.0MB 1 (10%) + 0.0% ± 0.1%
ReleaseFast build:
Benchmark 1 (129 runs): gpa-degen-master-release.exe
measurement mean ± σ min … max outliers delta
wall_time 38.9ms ± 1.12ms 36.7ms … 42.4ms 8 ( 6%) 0%
peak_rss 23.2MB ± 2.39KB 23.2MB … 23.2MB 0 ( 0%) 0%
Benchmark 2 (151 runs): gpa-degen-search-cur-release.exe
measurement mean ± σ min … max outliers delta
wall_time 33.2ms ± 999us 31.9ms … 36.3ms 20 (13%) ⚡- 14.7% ± 0.6%
peak_rss 23.2MB ± 2.26KB 23.2MB … 23.2MB 0 ( 0%) + 0.0% ± 0.0%
Diffstat:
1 file changed, 14 insertions(+), 10 deletions(-)
diff --git a/lib/std/heap/general_purpose_allocator.zig b/lib/std/heap/general_purpose_allocator.zig
@@ -541,10 +541,14 @@ pub fn GeneralPurposeAllocator(comptime config: Config) type {
fn searchBucket(
buckets: *Buckets,
addr: usize,
+ current_bucket: ?*BucketHeader,
) ?*BucketHeader {
- const search_page = mem.alignBackward(usize, addr, page_size);
+ const search_page: [*]align(page_size) u8 = @ptrFromInt(mem.alignBackward(usize, addr, page_size));
+ if (current_bucket != null and current_bucket.?.page == search_page) {
+ return current_bucket;
+ }
var search_header: BucketHeader = undefined;
- search_header.page = @ptrFromInt(search_page);
+ search_header.page = search_page;
const entry = buckets.getEntryFor(&search_header);
return if (entry.node) |node| node.key else null;
}
@@ -712,7 +716,7 @@ pub fn GeneralPurposeAllocator(comptime config: Config) type {
var bucket_index = math.log2(size_class_hint);
var size_class: usize = size_class_hint;
const bucket = while (bucket_index < small_bucket_count) : (bucket_index += 1) {
- if (searchBucket(&self.buckets[bucket_index], @intFromPtr(old_mem.ptr))) |bucket| {
+ if (searchBucket(&self.buckets[bucket_index], @intFromPtr(old_mem.ptr), self.cur_buckets[bucket_index])) |bucket| {
break bucket;
}
size_class *= 2;
@@ -720,7 +724,7 @@ pub fn GeneralPurposeAllocator(comptime config: Config) type {
if (config.retain_metadata) {
if (!self.large_allocations.contains(@intFromPtr(old_mem.ptr))) {
// object not in active buckets or a large allocation, so search empty buckets
- if (searchBucket(&self.empty_buckets, @intFromPtr(old_mem.ptr))) |bucket| {
+ if (searchBucket(&self.empty_buckets, @intFromPtr(old_mem.ptr), null)) |bucket| {
// bucket is empty so is_used below will always be false and we exit there
break :blk bucket;
} else {
@@ -830,7 +834,7 @@ pub fn GeneralPurposeAllocator(comptime config: Config) type {
var bucket_index = math.log2(size_class_hint);
var size_class: usize = size_class_hint;
const bucket = while (bucket_index < small_bucket_count) : (bucket_index += 1) {
- if (searchBucket(&self.buckets[bucket_index], @intFromPtr(old_mem.ptr))) |bucket| {
+ if (searchBucket(&self.buckets[bucket_index], @intFromPtr(old_mem.ptr), self.cur_buckets[bucket_index])) |bucket| {
break bucket;
}
size_class *= 2;
@@ -838,7 +842,7 @@ pub fn GeneralPurposeAllocator(comptime config: Config) type {
if (config.retain_metadata) {
if (!self.large_allocations.contains(@intFromPtr(old_mem.ptr))) {
// object not in active buckets or a large allocation, so search empty buckets
- if (searchBucket(&self.empty_buckets, @intFromPtr(old_mem.ptr))) |bucket| {
+ if (searchBucket(&self.empty_buckets, @intFromPtr(old_mem.ptr), null)) |bucket| {
// bucket is empty so is_used below will always be false and we exit there
break :blk bucket;
} else {
@@ -1387,10 +1391,10 @@ test "double frees" {
const index: usize = 6;
const size_class: usize = @as(usize, 1) << 6;
const small = try allocator.alloc(u8, size_class);
- try std.testing.expect(GPA.searchBucket(&gpa.buckets[index], @intFromPtr(small.ptr)) != null);
+ try std.testing.expect(GPA.searchBucket(&gpa.buckets[index], @intFromPtr(small.ptr), gpa.cur_buckets[index]) != null);
allocator.free(small);
- try std.testing.expect(GPA.searchBucket(&gpa.buckets[index], @intFromPtr(small.ptr)) == null);
- try std.testing.expect(GPA.searchBucket(&gpa.empty_buckets, @intFromPtr(small.ptr)) != null);
+ try std.testing.expect(GPA.searchBucket(&gpa.buckets[index], @intFromPtr(small.ptr), gpa.cur_buckets[index]) == null);
+ try std.testing.expect(GPA.searchBucket(&gpa.empty_buckets, @intFromPtr(small.ptr), null) != null);
// detect a large allocation double free
const large = try allocator.alloc(u8, 2 * page_size);
@@ -1408,7 +1412,7 @@ test "double frees" {
// check that flushing retained metadata doesn't disturb live allocations
gpa.flushRetainedMetadata();
try std.testing.expect(gpa.empty_buckets.root == null);
- try std.testing.expect(GPA.searchBucket(&gpa.buckets[index], @intFromPtr(normal_small.ptr)) != null);
+ try std.testing.expect(GPA.searchBucket(&gpa.buckets[index], @intFromPtr(normal_small.ptr), gpa.cur_buckets[index]) != null);
try std.testing.expect(gpa.large_allocations.contains(@intFromPtr(normal_large.ptr)));
try std.testing.expect(!gpa.large_allocations.contains(@intFromPtr(large.ptr)));
}
