[Tarantool-patches] [PATCH v1 luajit 10/41] perf: adjust fasta in LuaJIT-benches

[Sergey Kaplun]

Hi, Sergey!
Thanks for the review!
Fixed your comments and added the description for the benchmark.

On 23.12.25, Sergey Bronnikov wrote:
> Hello,
> 
> thanks for the patch! See my comments.
> 
> Sergey
> 
> On 10/24/25 13:50, Sergey Kaplun wrote:
> > This patch adjusts the aforementioned test to use the benchmark
> > framework introduced before. The default arguments are adjusted
> > according to the <PARAM_x86.txt> file. The arguments to the script still
> > can be provided in the command line run.
> >
> > Since the result output (with the different input parameter value)
> > produced by this benchmark is used in other benchmarks
> > (<k-nucleotide.lua> and <revcomp.lua>), the original script is used as a
> > library (inside the <libs/> subdirectory) with the updated default input
> > value and returns the number of items processed. The output for the
> > benchmark itself is suppressed and not checked since it is irrational to
> > store in the repository such huge files for testing.
> > ---
> >   perf/LuaJIT-benches/fasta.lua      | 120 +++++++----------------------
> >   perf/LuaJIT-benches/libs/fasta.lua |  98 +++++++++++++++++++++++
> >   2 files changed, 125 insertions(+), 93 deletions(-)
> >   create mode 100644 perf/LuaJIT-benches/libs/fasta.lua
> >
> > diff --git a/perf/LuaJIT-benches/fasta.lua b/perf/LuaJIT-benches/fasta.lua
> > index 7ce60804..d0dc005d 100644
> > --- a/perf/LuaJIT-benches/fasta.lua
> > +++ b/perf/LuaJIT-benches/fasta.lua

<snipped>

Added the comment with the short benchmark description, as we
discussed offline:

===================================================================
diff --git a/perf/LuaJIT-benches/fasta.lua b/perf/LuaJIT-benches/fasta.lua
index d0dc005d..457623b2 100644
--- a/perf/LuaJIT-benches/fasta.lua
+++ b/perf/LuaJIT-benches/fasta.lua
@@ -1,3 +1,9 @@
+-- Benchmark to check the performance of working with strings and
+-- output to the file. It generates DNA sequences by copying or
+-- weighted random selection.
+-- For details see:
+-- https://benchmarksgame-team.pages.debian.net/benchmarksgame/description/fasta.html
+
 local bench = require("bench").new(arg)
 
 local stdout = io.output()
diff --git a/perf/LuaJIT-benches/libs/fasta.lua b/perf/LuaJIT-benches/libs/fasta.lua
index e1592e77..58f59dd5 100644
--- a/perf/LuaJIT-benches/libs/fasta.lua
+++ b/perf/LuaJIT-benches/libs/fasta.lua
@@ -1,3 +1,10 @@
+-- Benchmark to check the performance of working with strings and
+-- output to the file. It generates DNA sequences by copying or
+-- weighted random selection.
+-- For details see:
+-- https://benchmarksgame-team.pages.debian.net/benchmarksgame/description/fasta.html
+-- Also, this file is used as a script to generate inputs for
+-- other benchmarks like <k-nucleotide.lua> and <revcomp.lua>.
 
 local Last = 42
 local function random(max)
===================================================================

> > +local bench = require("bench").new(arg)
> > +
> > +local stdout = io.output()
> > +
> > +local benchmark
> > +benchmark = {
> > +  name = "fasta",
> > +  -- XXX: The result file may take up to 278 Mb for the default
> > +  -- settings. To check the correctness of the script, run it as
> > +  -- is from the console.
> > +  skip_check = true,
> > +  setup = function()
> > +    io.output("/dev/null")
> > +  end,
> > +  payload = function()
> > +    -- Run the benchmark as is from the file.
> > +    local items = require("fasta")
> > +    -- Remove it from the cache to be sure the benchmark will run
> > +    -- at the next iteration.
> > +    package.loaded["fasta"] = nil
> > +    benchmark.items = items
> > +  end,
> > +  teardown = function()
> > +    io.output(stdout)
> > +  end,
> >   }
> >   
> > -local N = tonumber(arg and arg[1]) or 1000
> > -make_repeat_fasta('ONE', 'Homo sapiens alu', alu, N*2)
> > -make_random_fasta('TWO', 'IUB ambiguity codes', iub, N*3)
> > -make_random_fasta('THREE', 'Homo sapiens frequency', homosapiens, N*5)
> > +bench:add(benchmark)
> > +bench:run_and_report()
> > diff --git a/perf/LuaJIT-benches/libs/fasta.lua b/perf/LuaJIT-benches/libs/fasta.lua
> > new file mode 100644
> > index 00000000..9c72c244
> > --- /dev/null
> > +++ b/perf/LuaJIT-benches/libs/fasta.lua
> > @@ -0,0 +1,98 @@
> > +
> > +local Last = 42
> > +local function random(max)
> > +  local y = (Last * 3877 + 29573) % 139968
> > +  Last = y
> > +  return (max * y) / 139968
> > +end
> > +
> > +local function make_repeat_fasta(id, desc, s, n)
> > +  local write, sub = io.write, string.sub
> > +  write(">", id, " ", desc, "\n")
> > +  local p, sn, s2 = 1, #s, s..s
> > +  for i=60,n,60 do
> more whitespaces please
> > +    write(sub(s2, p, p + 59), "\n")
> > +    p = p + 60; if p > sn then p = p - sn end
> > +  end
> > +  local tail = n % 60
> > +  if tail > 0 then write(sub(s2, p, p + tail-1), "\n") end
> more whitespaces please. Here and below.

Reformated, as you suggested:

===================================================================
diff --git a/perf/LuaJIT-benches/libs/fasta.lua b/perf/LuaJIT-benches/libs/fasta.lua
index 9c72c244..e1592e77 100644
--- a/perf/LuaJIT-benches/libs/fasta.lua
+++ b/perf/LuaJIT-benches/libs/fasta.lua
@@ -10,12 +10,12 @@ local function make_repeat_fasta(id, desc, s, n)
   local write, sub = io.write, string.sub
   write(">", id, " ", desc, "\n")
   local p, sn, s2 = 1, #s, s..s
-  for i=60,n,60 do
+  for i = 60, n, 60 do
     write(sub(s2, p, p + 59), "\n")
     p = p + 60; if p > sn then p = p - sn end
   end
   local tail = n % 60
-  if tail > 0 then write(sub(s2, p, p + tail-1), "\n") end
+  if tail > 0 then write(sub(s2, p, p + tail - 1), "\n") end
 end
 
 local function make_random_fasta(id, desc, bs, n)
@@ -23,17 +23,17 @@ local function make_random_fasta(id, desc, bs, n)
   loadstring([=[
     local write, char, unpack, n, random = io.write, string.char, unpack, ...
     local buf, p = {}, 1
-    for i=60,n,60 do
-      for j=p,p+59 do ]=]..bs..[=[ end
-      buf[p+60] = 10; p = p + 61
+    for i = 60, n, 60 do
+      for j = p, p + 59 do ]=]..bs..[=[ end
+      buf[p + 60] = 10; p = p + 61
       if p >= 2048 then write(char(unpack(buf, 1, p-1))); p = 1 end
     end
     local tail = n % 60
     if tail > 0 then
-      for j=p,p+tail-1 do ]=]..bs..[=[ end
+      for j = p, p + tail - 1 do ]=]..bs..[=[ end
       p = p + tail; buf[p] = 10; p = p + 1
     end
-    write(char(unpack(buf, 1, p-1)))
+    write(char(unpack(buf, 1, p - 1)))
   ]=], desc)(n, random)
 end
 
@@ -41,13 +41,13 @@ local function bisect(c, p, lo, hi)
   local n = hi - lo
   if n == 0 then return "buf[j] = "..c[hi].."\n" end
   local mid = math.floor(n / 2)
-  return "if r < "..p[lo+mid].." then\n"..bisect(c, p, lo, lo+mid)..
-         "else\n"..bisect(c, p, lo+mid+1, hi).."end\n"
+  return "if r < "..p[lo + mid].." then\n"..bisect(c, p, lo, lo + mid)..
+         "else\n"..bisect(c, p, lo + mid + 1, hi).."end\n"
 end
 
 local function make_bisect(tab)
   local c, p, sum = {}, {}, 0
-  for i,row in ipairs(tab) do
+  for i, row in ipairs(tab) do
     c[i] = string.byte(row[1])
     sum = sum + row[2]
     p[i] = sum
===================================================================

> > +end
> > +

<snipped>

-- 
Best regards,
Sergey Kaplun