[Tarantool-patches] [PATCH v1 luajit 08/41] perf: adjust euler14-bit in LuaJIT-benches

[Sergey Kaplun]

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

On 13.11.25, Sergey Bronnikov wrote:
> Hi, Sergey,
> 
> thanks for the patch!
> 
> 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.
> > ---
> >   perf/LuaJIT-benches/euler14-bit.lua | 52 ++++++++++++++++++++---------
> >   1 file changed, 36 insertions(+), 16 deletions(-)
> >
> > diff --git a/perf/LuaJIT-benches/euler14-bit.lua b/perf/LuaJIT-benches/euler14-bit.lua
> > index 537f2bf3..7c521deb 100644
> > --- a/perf/LuaJIT-benches/euler14-bit.lua
> > +++ b/perf/LuaJIT-benches/euler14-bit.lua
> > @@ -1,22 +1,42 @@

Also, added the comment with the benchmark description as we discussed
offline.

===================================================================
diff --git a/perf/LuaJIT-benches/euler14-bit.lua b/perf/LuaJIT-benches/euler14-bit.lua
index 54311abe..c4ae3713 100644
--- a/perf/LuaJIT-benches/euler14-bit.lua
+++ b/perf/LuaJIT-benches/euler14-bit.lua
@@ -1,3 +1,8 @@
+-- The benchmark to check the performance of bitwise operations.
+-- It finds the longest Collatz sequence using bitwise arithmetic.
+-- For the details see:
+-- https://projecteuler.net/problem=14
+
 local bench = require("bench").new(arg)
 
 local bit = require("bit")
===================================================================

> > +local bench = require("bench").new(arg)
> >   
> >   local bit = require("bit")
> >   local bnot, bor, band = bit.bnot, bit.bor, bit.band
> >   local shl, shr = bit.lshift, bit.rshift
> >   

<snipped>

> > +bench:add({
> > +  name = "euler14_bit",
> > +  payload = function()
> > +    local cache, m, n = { 1 }, 1, 1
> > +    if drop_cache then cache = nil end
> > +    for i=2,N do
> s/2,/2, /
> > +      local j = i
> > +      for len=1,1000000000 do
> s/1,/1, /
> > +        j = bor(band(shr(j,1), band(j,1)-1), band(shl(j,1)+j+1, bnot(band(j,1)-1)))
> please add whitespaces, here and below
> > +        if cache then
> > +          local x = cache[j]; if x then j = x+len; break end
> whitespaces
> > +        elseif j == 1 then
> > +          j = len+1; break
> s/+/ + /

Refactor this part of the code to make it more clear. See the iterative
patch below:

===================================================================
diff --git a/perf/LuaJIT-benches/euler14-bit.lua b/perf/LuaJIT-benches/euler14-bit.lua
index 7c521deb..54311abe 100644
--- a/perf/LuaJIT-benches/euler14-bit.lua
+++ b/perf/LuaJIT-benches/euler14-bit.lua
@@ -13,14 +13,22 @@ bench:add({
   payload = function()
     local cache, m, n = { 1 }, 1, 1
     if drop_cache then cache = nil end
-    for i=2,N do
+    for i = 2, N do
       local j = i
-      for len=1,1000000000 do
-        j = bor(band(shr(j,1), band(j,1)-1), band(shl(j,1)+j+1, bnot(band(j,1)-1)))
+      for len = 1, 1000000000 do
+        j = bor(
+          band(shr(j, 1), band(j, 1) - 1),
+          band(shl(j, 1) + j + 1, bnot(band(j, 1) - 1))
+        )
         if cache then
-          local x = cache[j]; if x then j = x+len; break end
+          local x = cache[j]
+          if x then
+            j = x + len
+            break
+          end
         elseif j == 1 then
-          j = len+1; break
+          j = len + 1
+          break
         end
       end
       if cache then cache[i] = j end
===================================================================

> > +        end
> > +      end
> > +      if cache then cache[i] = j end
> > +      if j > m then m, n = j, i end
> > +    end
> > +    return {n = n, m = m}
> > +  end,
> > +  checker = function(res)

<snipped>

> > +bench:run_and_report()


-- 
Best regards,
Sergey Kaplun