From cd54f50a92e3435df99dc157715848abc6a9447c Mon Sep 17 00:00:00 2001
From: Sergey Kaplun <skaplun@tarantool.org>
Date: Wed, 22 May 2024 11:11:02 +0300
Subject: [PATCH] perf: standardize column_scan benchmark
The output now contains items per second instead of time in seconds.
Also, two new options are introduced:
* The `--output` option allows you to specify the output file.
* The `--output_format` option means the format for the printed output.
The default is "console". It prints rows proceeded per second to the
stdout. The "json" format contains all the information about the
benchmark in a format similar to Google Benchmark's.
Usually, these options should be used together to dump machine-readable
results for the benchmarks.
NO_DOC=perf test
NO_CHANGELOG=perf test
NO_TEST=perf test
---
perf/lua/column_scan.lua | 49 +++++++++++++++++++++++++++++-----------
1 file changed, 36 insertions(+), 13 deletions(-)
diff --git a/perf/lua/column_scan.lua b/perf/lua/column_scan.lua
index 634f28ae06..40541a1124 100644
--- a/perf/lua/column_scan.lua
+++ b/perf/lua/column_scan.lua
@@ -1,15 +1,8 @@
--
-- The test measures run time of space full scan.
--
--- Output format:
--- <test-case> <run-time-seconds>
---
--- Options:
--- --engine <string> space engine to use for the test
--- --column_count <number> number of columns in the test space
--- --row_count <number> number of rows in the test space
--- --use_read_view use a read view
--- --use_scanner_api use the column scanner API
+-- Output format (console):
+-- <test-case> <rows-per-second>
--
-- NOTE: The test requires a C module. Set the BUILDDIR environment variable to
-- the tarantool build directory if using out-of-source build.
@@ -21,14 +14,26 @@ local fio = require('fio')
local ffi = require('ffi')
local log = require('log')
local tarantool = require('tarantool')
+local benchmark = require('benchmark')
-local params = require('internal.argparse').parse(arg, {
- {'engine', 'string'},
+local USAGE = [[
+ column_count <number, 100> - number of columns in the test space
+ engine <string, 'memtx'> - space engine to use for the test
+ row_count <number, 1000000> - number of rows in the test space
+ use_read_view <boolean, false> - use a read view
+ use_scanner_api <boolean, false> - use the column scanner API
+
+ Being run without options, this benchmark measures the run time of a full scan
+ from the space.
+]]
+
+local params = benchmark.argparse(arg, {
{'column_count', 'number'},
+ {'engine', 'string'},
{'row_count', 'number'},
{'use_read_view', 'boolean'},
{'use_scanner_api', 'boolean'},
-})
+}, USAGE)
local DEFAULT_ENGINE = 'memtx'
local DEFAULT_COLUMN_COUNT = 100
@@ -40,6 +45,8 @@ params.row_count = params.row_count or DEFAULT_ROW_COUNT
params.use_read_view = params.use_read_view or false
params.use_scanner_api = params.use_scanner_api or false
+local bench = benchmark.new(params)
+
local BUILDDIR = fio.abspath(fio.pathjoin(os.getenv('BUILDDIR') or '.'))
local MODULEPATH = fio.pathjoin(BUILDDIR, 'perf', 'lua',
'?.' .. tarantool.build.mod_format)
@@ -144,13 +151,29 @@ local TESTS = {
},
}
+local function run_test(test)
+ local func = test.func
+ local real_time_start = clock.time()
+ local cpu_time_start = clock.proc()
+ func()
+ local delta_real = clock.time() - real_time_start
+ local delta_cpu = clock.proc() - cpu_time_start
+ bench:add_result(test.name, {
+ real_time = delta_real,
+ cpu_time = delta_cpu,
+ items = params.row_count,
+ })
+end
+
test_module.init()
fiber.set_max_slice(9000)
for _, test in ipairs(TESTS) do
log.info('Running test %s...', test.name)
test.func() -- warmup
- print(string.format('%s %.3f', test.name, clock.bench(test.func)[1]))
+ run_test(test)
end
+bench:dump_results()
+
os.exit(0)
--
GitLab