Re: [Tarantool-patches] [PATCH v1 luajit 27/41] perf: adjust scimark-2010-12-20 in LuaJIT-benches

[Sergey Bronnikov via Tarantool-patches <tarantool-patches@dev.tarantool.org>]

[-- Attachment #1: Type: text/plain, Size: 6144 bytes --]

Hi, Sergey,

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.
>
> The time for each subsequent benchmark is increased up to 4 seconds,
> accoring the defaults in the "bench" framework. The main difference
> between this test and others that will be added in next commits is
> the usage of FFI arrays instead of plain Lua tables.
> ---
>   perf/LuaJIT-benches/scimark-2010-12-20.lua | 93 +++++++++++++---------
>   1 file changed, 54 insertions(+), 39 deletions(-)
>
> diff --git a/perf/LuaJIT-benches/scimark-2010-12-20.lua b/perf/LuaJIT-benches/scimark-2010-12-20.lua
> index 353acb7c..3fb627fa 100644
> --- a/perf/LuaJIT-benches/scimark-2010-12-20.lua
> +++ b/perf/LuaJIT-benches/scimark-2010-12-20.lua
> @@ -9,25 +9,26 @@
>   local SCIMARK_VERSION = "2010-12-10"
>   local SCIMARK_COPYRIGHT = "Copyright (C) 2006-2010 Mike Pall"
>   
> -local MIN_TIME = 2.0
> +local bench = require("bench").new(arg)
> +
>   local RANDOM_SEED = 101009 -- Must be odd.
>   local SIZE_SELECT = "small"
>   
>   local benchmarks = {
>     "FFT", "SOR", "MC", "SPARSE", "LU",
>     small = {
> -    FFT		= { 1024 },
> -    SOR		= { 100 },
> -    MC		= { },
> -    SPARSE	= { 1000, 5000 },
> -    LU		= { 100 },
> +    FFT		= { params = { 1024 }, cycles = 50000, },
> +    SOR		= { params = { 100 }, cycles = 50000, },
> +    MC		= { params = { }, cycles = 15e7, },
> +    SPARSE	= { params = { 1000, 5000 }, cycles = 15e4, },
> +    LU		= { params = { 100 }, cycles = 5000, },
>     },
>     large = {
> -    FFT		= { 1048576 },
> -    SOR		= { 1000 },
> -    MC		= { },
> -    SPARSE	= { 100000, 1000000 },
> -    LU		= { 1000 },
> +    FFT		= { params = { 1048576 }, cycles = 25, },
> +    SOR		= { params = { 1000 }, cycles = 500, },
> +    MC		= { params = { }, cycles = 15e7, },
> +    SPARSE	= { params = { 100000, 1000000 }, cycles = 1500, },
> +    LU		= { params = { 1000 }, cycles = 50, },
>     },
>   }
please add a comment about chosen parameters
>   
> @@ -342,48 +343,51 @@ local function fmtparams(p1, p2)
>     return ""
>   end
>   
> -local function measure(min_time, name, ...)
> +local function measure(name, cycles, ...)
>     array_init()
>     rand_init(RANDOM_SEED)
>     local run = benchmarks[name](...)
> -  local cycles = 1
> -  repeat
> -    local tm = clock()
> -    local flops = run(cycles, ...)
> -    tm = clock() - tm
> -    if tm >= min_time then
> -      local res = flops / tm * 1.0e-6
> -      local p1, p2 = ...
> -      printf("%-7s %8.2f  %s\n", name, res, fmtparams(...))
> -      return res
> -    end
> -    cycles = cycles * 2
> -  until false
> +  local flops = run(cycles, ...)
> +  return flops
>   end
>   
> -printf("Lua SciMark %s based on SciMark 2.0a. %s.\n\n",
> -       SCIMARK_VERSION, SCIMARK_COPYRIGHT)
> +-- printf("Lua SciMark %s based on SciMark 2.0a. %s.\n\n",
> +--        SCIMARK_VERSION, SCIMARK_COPYRIGHT)
>   

I propose to move this to a comment with test description.

Something like:

The test runs the Lua version of SciMark 2.0a, which is a benchmark for 
scientific and numerical computing developed by programmers at the NIST 
(National Institute of Standards and Technology). This test is made up 
of Fast Foruier Transform, Jacobi Successive Over-relaxation, Monte 
Carlo, Sparse Matrix Multiply, and dense LU matrix factorization benchmarks.

plus description of available test-specific options (noffi, small, etc) 
or just

a command-line that will show usage: ./scimark-2010-12-20.lua help

>   while arg and arg[1] do
>     local a = table.remove(arg, 1)
> -  if a == "-noffi" then
> +  if a == "noffi" then
>       package.preload.ffi = nil
> -  elseif a == "-small" then
> +  elseif a == "small" then
>       SIZE_SELECT = "small"
> -  elseif a == "-large" then
> +  elseif a == "large" then
>       SIZE_SELECT = "large"
>     elseif benchmarks[a] then
> -    local p = benchmarks[SIZE_SELECT][a]
> -    measure(MIN_TIME, a, tonumber(arg[1]) or p[1], tonumber(arg[2]) or p[2])
> +    local cycles = benchmarks[SIZE_SELECT][a].cycles
> +    local p = benchmarks[SIZE_SELECT][a].params
> +    local b
> +    b = {
> +      name = a,
> +      -- XXX: The description of tests for each function is too
> +      -- inconvenient.
> +      skip_check = true,
> +      payload = function()
> +        local flops = measure(a, cycles, tonumber(arg[1]) or p[1],
> +                              tonumber(arg[2]) or p[2])
> +        b.items = flops
> +      end,
> +    }
> +bench:add(b)
> +bench:run_and_report()
>       return
>     else
> -    printf("Usage: scimark [-noffi] [-small|-large] [BENCH params...]\n\n")
> -    printf("BENCH   -small         -large\n")
> +    printf("Usage: scimark [noffi] [small|large] [BENCH params...]\n\n")
> +    printf("BENCH   small         large\n")
>       printf("---------------------------------------\n")
>       for _,name in ipairs(benchmarks) do
>         printf("%-7s %-13s %s\n", name,
> -	     fmtparams(unpack(benchmarks.small[name])),
> -	     fmtparams(unpack(benchmarks.large[name])))
> +	     fmtparams(unpack(benchmarks.small[name].params)),
> +	     fmtparams(unpack(benchmarks.large[name].params)))
>       end
>       printf("\n")
>       os.exit(1)
> @@ -393,8 +397,19 @@ end
>   local params = benchmarks[SIZE_SELECT]
>   local sum = 0
>   for _,name in ipairs(benchmarks) do
> -  sum = sum + measure(MIN_TIME, name, unpack(params[name]))
> +  local cycles = params[name].cycles
> +  local b
> +  b = {
> +    name = name,
> +    -- XXX: The description of tests for each function is too
> +    -- inconvenient.
> +    skip_check = true,
> +    payload = function()
> +      local flops = measure(name, cycles, unpack(params[name].params))
> +      b.items = flops
> +    end,
> +  }
> +bench:add(b)
>   end
> -printf("\nSciMark %8.2f  [%s problem sizes]\n", sum / #benchmarks, SIZE_SELECT)
> -io.flush()
>   
> +bench:run_and_report()

[-- Attachment #2: Type: text/html, Size: 6940 bytes --]