git/list[1] front-page[2] threads[3] people[4] search[5] about
 

[PATCH v2 1/5] cache-tree: add perf test comparing update and prime

From
Victoria Dye via GitGitGadget <gitgitgadget@gmail.com>
Date
Nov 10, 2022, 01:57 UTC
Message-ID
<833519d87c843eb8147a45b1ec2c1fd3f3c21905.1668045438.git.gitgitgadget@gmail.com>
In-Reply-To
<pull.1411.v2.git.1668045438.gitgitgadget@gmail.com>
From: Victoria Dye <vdye@github.com>

Add a performance test comparing the execution times of 'prime_cache_tree()' and 'cache_tree_update(_, WRITE_TREE_SILENT | WRITE_TREE_REPAIR)'. The goal of comparing these two is to identify which is the faster method for rebuilding an invalid cache tree, ultimately to remove one when both are (reundantly) called in immediate succession.

Both methods are fast, so the new tests in 'p0090-cache-tree.sh' must call each tested function multiple times to ensure the reported times (to 0.01s resolution) convey the differences between them.

The tests compare the timing of a 'test-tool cache-tree' run as a no-op (to capture a baseline for the overhead associated with running the tool), 'cache_tree_update()', and 'prime_cache_tree()' on four scenarios:

- A completely valid cache tree
- A cache tree with 2 invalid paths
- A cache tree with 50 invalid paths
- A completely empty cache tree
Example results:

Test this tree ----------------------------------------------------------- 0090.2: no-op, clean 1.27(0.48+0.52) 0090.3: prime_cache_tree, clean 2.02(0.83+0.85) 0090.4: cache_tree_update, clean 1.30(0.49+0.54) 0090.5: no-op, invalidate 2 1.29(0.48+0.54) 0090.6: prime_cache_tree, invalidate 2 1.98(0.81+0.83) 0090.7: cache_tree_update, invalidate 2 2.12(0.94+0.86) 0090.8: no-op, invalidate 50 1.32(0.50+0.55) 0090.9: prime_cache_tree, invalidate 50 2.10(0.86+0.89) 0090.10: cache_tree_update, invalidate 50 2.35(1.14+0.90) 0090.11: no-op, empty 1.33(0.50+0.54) 0090.12: prime_cache_tree, empty 2.04(0.84+0.87) 0090.13: cache_tree_update, empty 2.51(1.27+0.92)

These timings show that, while 'cache_tree_update()' is faster when the cache tree is completely valid, it is equal to or slower than 'prime_cache_tree()' when there are any invalid paths. Since the redundant calls are mostly in scenarios where the cache tree will be at least partially invalid (e.g., 'git reset --hard'), 'prime_cache_tree()' will likely perform better than 'cache_tree_update()' in typical cases.

Signed-off-by: Victoria Dye <vdye@github.com>
---
 Makefile                   |  1 +
 t/helper/test-cache-tree.c | 64 ++++++++++++++++++++++++++++++++++++++
 t/helper/test-tool.c       |  1 +
 t/helper/test-tool.h       |  1 +
 t/perf/p0090-cache-tree.sh | 36 +++++++++++++++++++++
 5 files changed, 103 insertions(+)
 create mode 100644 t/helper/test-cache-tree.c
 create mode 100755 t/perf/p0090-cache-tree.sh
diff --git a/Makefile b/Makefile
index 4927379184c..3639c7c2a94 100644
--- a/Makefile
+++ b/Makefile
@@ -723,6 +723,7 @@ TEST_BUILTINS_OBJS += test-advise.o
 TEST_BUILTINS_OBJS += test-bitmap.o
 TEST_BUILTINS_OBJS += test-bloom.o
 TEST_BUILTINS_OBJS += test-bundle-uri.o
+TEST_BUILTINS_OBJS += test-cache-tree.o
 TEST_BUILTINS_OBJS += test-chmtime.o
 TEST_BUILTINS_OBJS += test-config.o
 TEST_BUILTINS_OBJS += test-crontab.o
diff --git a/t/helper/test-cache-tree.c b/t/helper/test-cache-tree.c
new file mode 100644
index 00000000000..8d06039fb5c
--- /dev/null
+++ b/t/helper/test-cache-tree.c
@@ -0,0 +1,64 @@
+#include "test-tool.h"
+#include "cache.h"
+#include "tree.h"
+#include "cache-tree.h"
+#include "parse-options.h"
+
+static char const * const test_cache_tree_usage[] = {
+	N_("test-tool cache-tree <options> (control|prime|update)"),
+	NULL
+};
+
+int cmd__cache_tree(int argc, const char **argv)
+{
+	struct object_id oid;
+	struct tree *tree;
+	int empty = 0;
+	int invalidate_qty = 0;
+	int i;
+
+	struct option options[] = {
+		OPT_BOOL(0, "empty", &empty,
+			 N_("clear the cache tree before each iteration")),
+		OPT_INTEGER_F(0, "invalidate", &invalidate_qty,
+			      N_("number of entries in the cache tree to invalidate (default 0)"),
+			      PARSE_OPT_NONEG),
+		OPT_END()
+	};
+
+	setup_git_directory();
+
+	parse_options(argc, argv, NULL, options, test_cache_tree_usage, 0);
+
+	if (read_cache() < 0)
+		die("unable to read index file");
+
+	oidcpy(&oid, &the_index.cache_tree->oid);
+	tree = parse_tree_indirect(&oid);
+	if (!tree)
+		die(_("not a tree object: %s"), oid_to_hex(&oid));
+
+	if (empty) {
+		/* clear the cache tree & allocate a new one */
+		cache_tree_free(&the_index.cache_tree);
+		the_index.cache_tree = cache_tree();
+	} else if (invalidate_qty) {
+		/* invalidate the specified number of unique paths */
+		float f_interval = (float)the_index.cache_nr / invalidate_qty;
+		int interval = f_interval < 1.0 ? 1 : (int)f_interval;
+		for (i = 0; i < invalidate_qty && i * interval < the_index.cache_nr; i++)
+			cache_tree_invalidate_path(&the_index, the_index.cache[i * interval]->name);
+	}
+
+	if (!argc)
+		die("Must specify subcommand");
+	else if (!strcmp(argv[0], "prime"))
+		prime_cache_tree(the_repository, &the_index, tree);
+	else if (!strcmp(argv[0], "update"))
+		cache_tree_update(&the_index, WRITE_TREE_SILENT | WRITE_TREE_REPAIR);
+	/* use "control" subcommand to specify no-op */
+	else if (!!strcmp(argv[0], "control"))
+		die("Unknown command %s", argv[0]);
+
+	return 0;
+}
diff --git a/t/helper/test-tool.c b/t/helper/test-tool.c
index 01cda9358df..547a3be1c8b 100644
--- a/t/helper/test-tool.c
+++ b/t/helper/test-tool.c
@@ -14,6 +14,7 @@ static struct test_cmd cmds[] = {
 	{ "bitmap", cmd__bitmap },
 	{ "bloom", cmd__bloom },
 	{ "bundle-uri", cmd__bundle_uri },
+	{ "cache-tree", cmd__cache_tree },
 	{ "chmtime", cmd__chmtime },
 	{ "config", cmd__config },
 	{ "crontab", cmd__crontab },
diff --git a/t/helper/test-tool.h b/t/helper/test-tool.h
index ca2948066fd..e44e1d896d3 100644
--- a/t/helper/test-tool.h
+++ b/t/helper/test-tool.h
@@ -8,6 +8,7 @@ int cmd__advise_if_enabled(int argc, const char **argv);
 int cmd__bitmap(int argc, const char **argv);
 int cmd__bloom(int argc, const char **argv);
 int cmd__bundle_uri(int argc, const char **argv);
+int cmd__cache_tree(int argc, const char **argv);
 int cmd__chmtime(int argc, const char **argv);
 int cmd__config(int argc, const char **argv);
 int cmd__crontab(int argc, const char **argv);
diff --git a/t/perf/p0090-cache-tree.sh b/t/perf/p0090-cache-tree.sh
new file mode 100755
index 00000000000..a8eabca2c4d
--- /dev/null
+++ b/t/perf/p0090-cache-tree.sh
@@ -0,0 +1,36 @@
+#!/bin/sh
+
+test_description="Tests performance of cache tree update operations"
+
+. ./perf-lib.sh
+
+test_perf_large_repo
+test_checkout_worktree
+
+count=100
+
+test_expect_success 'setup cache tree' '
+	git write-tree
+'
+
+test_cache_tree () {
+	test_perf "$1, $3" "
+		for i in \$(test_seq $count)
+		do
+			test-tool cache-tree $4 $2
+		done
+	"
+}
+
+test_cache_tree_update_functions () {
+	test_cache_tree 'no-op' 'control' "$1" "$2"
+	test_cache_tree 'prime_cache_tree' 'prime' "$1" "$2"
+	test_cache_tree 'cache_tree_update' 'update' "$1" "$2"
+}
+
+test_cache_tree_update_functions "clean" ""
+test_cache_tree_update_functions "invalidate 2" "--invalidate 2"
+test_cache_tree_update_functions "invalidate 50" "--invalidate 50"
+test_cache_tree_update_functions "empty" "--empty"
+
+test_done
-- 
gitgitgadget
Previous: Victoria Dye via GitGitGadgetNext: Victoria Dye via GitGitGadget
Message 15 of 31 in “Skip 'cache_tree_update()' when 'prime_cache_tree()' is called immediate after”
  1. 0/5 Skip 'cache_tree_update()' when 'prime_cache_tree()' is called immediate afterVictoria Dye via GitGitGadget, Nov 8, 2022
  2. 1/5 cache-tree: add perf test comparing update and primeVictoria Dye via GitGitGadget, Nov 8, 2022
  3. SZEDER GáborNov 10, 2022
  4. 3/5 reset: use 'skip_cache_tree_update' optionVictoria Dye via GitGitGadget, Nov 8, 2022
  5. 2/5 unpack-trees: add 'skip_cache_tree_update' optionVictoria Dye via GitGitGadget, Nov 8, 2022
  6. 5/5 rebase: use 'skip_cache_tree_update' optionVictoria Dye via GitGitGadget, Nov 8, 2022
  7. 4/5 read-tree: use 'skip_cache_tree_update' optionVictoria Dye via GitGitGadget, Nov 8, 2022
  8. Derrick StoleeNov 9, 2022
  9. Victoria DyeNov 9, 2022
  10. Derrick StoleeNov 10, 2022
  11. Taylor BlauNov 9, 2022
  12. 0/5 Skip 'cache_tree_update()' when 'prime_cache_tree()' is called immediate afterVictoria Dye via GitGitGadget, Nov 10, 2022
  13. 3/5 reset: use 'skip_cache_tree_update' optionVictoria Dye via GitGitGadget, Nov 10, 2022
  14. 2/5 unpack-trees: add 'skip_cache_tree_update' optionVictoria Dye via GitGitGadget, Nov 10, 2022
  15. 1/5 cache-tree: add perf test comparing update and primeVictoria Dye via GitGitGadget, Nov 10, 2022
  16. 5/5 rebase: use 'skip_cache_tree_update' optionVictoria Dye via GitGitGadget, Nov 10, 2022
  17. Phillip WoodNov 10, 2022
  18. Victoria DyeNov 10, 2022
  19. 4/5 read-tree: use 'skip_cache_tree_update' optionVictoria Dye via GitGitGadget, Nov 10, 2022
  20. Taylor BlauNov 10, 2022
  21. Derrick StoleeNov 10, 2022
  22. 0/5 Skip 'cache_tree_update()' when 'prime_cache_tree()' is called immediate afterVictoria Dye via GitGitGadget, Nov 10, 2022
  23. 1/5 cache-tree: add perf test comparing update and primeVictoria Dye via GitGitGadget, Nov 10, 2022
  24. 2/5 unpack-trees: add 'skip_cache_tree_update' optionVictoria Dye via GitGitGadget, Nov 10, 2022
  25. 3/5 reset: use 'skip_cache_tree_update' optionVictoria Dye via GitGitGadget, Nov 10, 2022
  26. 5/5 rebase: use 'skip_cache_tree_update' optionVictoria Dye via GitGitGadget, Nov 10, 2022
  27. 4/5 read-tree: use 'skip_cache_tree_update' optionVictoria Dye via GitGitGadget, Nov 10, 2022
  28. SZEDER GáborNov 10, 2022
  29. Victoria DyeNov 10, 2022
  30. Taylor BlauNov 11, 2022
  31. Derrick StoleeNov 14, 2022

Read the whole thread, see it on lore, or plain text.

$ cat FOOTERMessages come from the public archive at lore.kernel.org/git, fetched every hour. The front page is chosen and written each morning by an AI editor and can be wrong; the threads themselves are the record. About and API. For agents: an MCP server at https://gitlist.dev/mcp, and any thread, story or person page as Markdown by adding .md to its URL (or sending Accept: text/markdown). Details in /llms.txt.