diff --git a/README.md b/README.md
index 584a664..3375255 100644
--- a/README.md
+++ b/README.md
@@ -236,6 +236,43 @@ box.space.myqueue:put({ name="xxx"; data="yyy"; }, { delay = 1.5; ttl = 100 })
 | S -> D     | Task was moved from another tube to be Done                                                 |
 | S -> B     | Task was moved from another tube to be Buried                                               |
 
+## Benchmarks
+
+```bash
+❯ .rocks/bin/luabench -d 1000000x
+Tarantool version: Tarantool 2.10.7-0-g60f7e18
+Tarantool build: Darwin-arm64-RelWithDebInfo (static)
+Tarantool build flags:  -Wno-unknown-pragmas -fexceptions -funwind-tables -fno-common -Wformat -Wformat-security -Werror=format-security -fstack-protector-strong -fPIC -fmacro-prefix-map=/var/folders/8x/1m5v3n6d4mn62g9w_65vvt_r0000gn/T/tarantool_install1565297302=. -std=c11 -Wall -Wextra -Wno-strict-aliasing -Wno-char-subscripts -Wno-gnu-alignof-expression -Wno-cast-function-type
+CPU: Apple M1 @ 8
+JIT: Enabled
+JIT: fold cse dce fwd dse narrow loop abc sink fuse
+Duration: 1000000 iters
+
+--- BENCH: 001_put_take_bench::bench_producer:producer-1
+ 1000000             44046 ns/op             22703 op/s     4137 B/op   +3946.04MB
+
+--- BENCH: 001_put_take_bench::bench_producer:producer-2
+ 1000000             27519 ns/op             36339 op/s     4282 B/op   +4084.35MB
+
+--- BENCH: 001_put_take_bench::bench_producer:producer-4
+ 1000000             21510 ns/op             46491 op/s     4500 B/op   +4291.69MB
+
+--- BENCH: 001_put_take_bench::bench_producer:producer-8
+ 1000000             20804 ns/op             48068 op/s     4379 B/op   +4176.26MB
+
+--- BENCH: 001_put_take_bench::bench_producer:producer-12
+ 1000000             20383 ns/op             49062 op/s     4372 B/op   +4169.59MB
+
+--- BENCH: 001_put_take_bench::bench_producer:producer-16
+ 1000000             21495 ns/op             46523 op/s     4215 B/op   +4019.98MB
+
+--- BENCH: 001_put_take_bench::bench_producer:producer-20
+ 1000000             23676 ns/op             42238 op/s     4258 B/op   +4061.49MB
+
+--- BENCH: 001_put_take_bench::bench_producer:producer-24
+ 1000000             24456 ns/op             40891 op/s     4274 B/op   +4076.65MB
+```
+
 ## TODO
 
 * reload/upgrade and feature switch
diff --git a/benchmarks/001_put_take_bench.lua b/benchmarks/001_put_take_bench.lua
new file mode 100644
index 0000000..1c60cb7
--- /dev/null
+++ b/benchmarks/001_put_take_bench.lua
@@ -0,0 +1,103 @@
+local lb = require 'luabench'
+local fiber = require 'fiber'
+local log   = require 'log'
+local fun   = require 'fun'
+
+local xqueue = require 'xqueue'
+
+box.cfg{ memtx_memory = 2^30, log_level = 4 }
+box.schema.space.create('queue', {
+	if_not_exists = true,
+	format = {
+		{ name = 'id',     type = 'unsigned' },
+		{ name = 'status', type = 'string'   },
+	},
+})
+box.space.queue:create_index('primary', { parts = {'id'}, if_not_exists = true })
+box.space.queue:create_index('xq', { parts = {'status', 'id'}, if_not_exists = true })
+
+xqueue.upgrade(box.space.queue, {
+	fields = {
+		id = 'id',
+		status = 'status',
+	},
+	features = {
+		id = 'time64',
+		keep = false,
+		delayed = false,
+		buried = false,
+		retval = 'tuple',
+		zombie = false,
+		ttl = false,
+		ttr = false,
+	},
+	debug = false,
+	workers = 32,
+	worker = function(task)
+		-- pass (auto-ack)
+	end
+})
+
+local queue = box.space.queue --[[@as xqueue.space]]
+lb.before_all(function() queue:truncate() end)
+
+local M = {}
+
+local function new_producer_bench_f(num)
+	local start = fiber.cond()
+	local limit = 0
+	local produced = 0
+
+	local done = fiber.channel()
+
+	for _ = 1, num do
+		fiber.create(function()
+			while true do
+				start:wait()
+
+				local empty_tab = {}
+				while produced < limit do
+					produced = produced + 1
+					local ok, err = pcall(queue.put, queue, empty_tab)
+					if not ok then
+						log.error(":put() => %s", err)
+					end
+				end
+
+				-- await everything
+				while queue:len() > 0 do
+					for _, task in queue.index.xq:pairs() do
+						queue:wait(task, 0.1)
+					end
+				end
+
+				done:put(true)
+			end
+		end)
+	end
+
+	---@param sb luabench.B
+	return function(sb)
+		limit = sb.N
+		produced = 0
+		start:broadcast()
+
+		for _ = 1, num do
+			done:get() -- await all producers to finish
+		end
+	end
+end
+
+---@param b luabench.B
+function M.bench_producer_await(b)
+	b:run('producer-1', new_producer_bench_f(1))
+	b:run('producer-2', new_producer_bench_f(2))
+	b:run('producer-4', new_producer_bench_f(4))
+	b:run('producer-8', new_producer_bench_f(8))
+	b:run('producer-12', new_producer_bench_f(12))
+	b:run('producer-16', new_producer_bench_f(16))
+	b:run('producer-20', new_producer_bench_f(20))
+	b:run('producer-24', new_producer_bench_f(24))
+end
+
+return M