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

Re: [PATCH v5 1/4] p0004-read-tree: perf test to time read-tree

From
René Scharfe <l.s.r@web.de>
Date
Apr 6, 2017, 20:20 UTC
Message-ID
<6937b76e-9bbf-7ce5-8605-e09f556f8a26@web.de>
In-Reply-To
<20170405173809.3098-2-git@jeffhostetler.com>
Am 05.04.2017 um 19:38 schrieb git@jeffhostetler.com:
Show 38 quoted lines
> From: Jeff Hostetler <jeffhost@microsoft.com>
> 
> Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
> ---
>  t/perf/p0004-read-tree.sh | 116 ++++++++++++++++++++++++++++++++++++++++++++++
>  1 file changed, 116 insertions(+)
>  create mode 100755 t/perf/p0004-read-tree.sh
> 
> diff --git a/t/perf/p0004-read-tree.sh b/t/perf/p0004-read-tree.sh
> new file mode 100755
> index 0000000..5d8bbf5
> --- /dev/null
> +++ b/t/perf/p0004-read-tree.sh
> @@ -0,0 +1,116 @@
> +#!/bin/sh
> +
> +test_description="Tests performance of read-tree"
> +
> +. ./perf-lib.sh
> +
> +test_perf_default_repo
> +test_checkout_worktree
> +
> +## usage: dir depth width files
> +make_paths () {
> +	for f in $(seq $4)
> +	do
> +		echo $1/file$f
> +	done;
> +	if test $2 -gt 0;
> +	then
> +		for w in $(seq $3)
> +		do
> +			make_paths $1/dir$w $(($2 - 1)) $3 $4
> +		done
> +	fi
> +	return 0
> +}

"make_paths xxx_dir_xxx 5 10 9" takes more than a minute for me. Providing its results as a file would be quicker but less flexible. The following command prints the same result in less than a second.

	awk -v dir=xxx_dir_xxx -v depth=5 -v width=10 -v files=9 '
        	function make_paths(dir, depth, width, files,  i)
	        {
        	        for (i = 1; i <= files; i++) {
	                        print dir "/file" i
        	        }
	                if (depth > 0) {
        	                for (i = 1; i <= width; i++) {
	                                make_paths(dir "/dir" i, depth - 1, width, files)
        	                }
                	}
	        }
	        END {make_paths(dir, depth, width, files)}
	' </dev/null
It's faster because it avoids calling seq thousands of times.
> +
> +fill_index () {
> +	make_paths $1 $2 $3 $4 |
> +	sed "s/^/100644 $EMPTY_BLOB	/" |

You could add the prefix to the script above and avoid this sed call as well.

René
Previous: git@jeffhostetler.comNext: Jeff Hostetler
Message 12 of 16 in “read-cache: speed up add_index_entry”
  1. 0/4 read-cache: speed up add_index_entrygit@jeffhostetler.com, Apr 5, 2017
  2. 4/4 read-cache: speed up add_index_entry during checkoutgit@jeffhostetler.com, Apr 5, 2017
  3. SZEDER GáborApr 5, 2017
  4. Jeff HostetlerApr 6, 2017
  5. 3/4 test-strcmp-offset: created test for strcmp_offsetgit@jeffhostetler.com, Apr 5, 2017
  6. SZEDER GáborApr 5, 2017
  7. Johannes SchindelinApr 6, 2017
  8. Jeff HostetlerApr 6, 2017
  9. René ScharfeApr 6, 2017
  10. Jeff HostetlerApr 6, 2017
  11. 1/4 p0004-read-tree: perf test to time read-treegit@jeffhostetler.com, Apr 5, 2017
  12. René ScharfeApr 6, 2017
  13. Jeff HostetlerApr 6, 2017
  14. 2/4 read-cache: add strcmp_offset functiongit@jeffhostetler.com, Apr 5, 2017
  15. SZEDER GáborApr 6, 2017
  16. Jeff HostetlerApr 6, 2017

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.