Julia mandelbrot for benchmarking against Python/Mojo

mandel.jl

function mandelbrot_kernel(c, max_iter)
    z = c
    for i in 1:max_iter
        z = z * z + c
        if abs2(z) > 4
            return i-1
        end
    end
    return max_iter
end

function mandelbrot(;xmin = -2.25, xmax = 0.75, xn = 450, ymin = -1.25, ymax = 1.25, yn = 375, max_iter = 200)
    dx = (xmax - xmin) / xn
    dy = (ymax - ymin) / yn
    return [mandelbrot_kernel(Complex(x,y), max_iter) for y in ymin:dy:ymax, x in xmin:dx:xmax]
end

using BenchmarkTools
@benchmark mandelbrot()

# BenchmarkTools.Trial: 172 samples with 1 evaluation.
#  Range (min … max):  25.824 ms … 38.800 ms  ┊ GC (min … max): 0.00% … 0.00%
#  Time  (median):     27.185 ms              ┊ GC (median):    0.00%
#  Time  (mean ± σ):   29.104 ms ±  3.357 ms  ┊ GC (mean ± σ):  0.27% ± 1.30%
# 
#    █▆  ▄▅                                                      
#   ▅██▆███▆▃▆▄▄▃▃▃▃▃▁▃▁▃▁▃▁▃▁▁▁▁▁▃▃▆▄▃█▆▃▄▇▄▁▃▁▃▃▁▁▃▃▁▃▁▁▁▁▃▁▃ ▃
#   25.8 ms         Histogram: frequency by time        37.4 ms <

# On same machine with python 3.10.8 I get 1150ms median-ish.

# Multi-threaded version.
function mandelbrot_mt(;xmin = -2.25, xmax = 0.75, xn = 450, ymin = -1.25, ymax = 1.25, yn = 375, max_iter = 200)
    dx = (xmax - xmin) / xn
    dy = (ymax - ymin) / yn
    result = zeros(UInt32, (yn,xn))
    Threads.@threads for j = 1:yn
        y = ymin + dy * (j-1)
        for i = 1:xn
            x = xmin + dx * (i-1)
            @inbounds result[j,i] = mandelbrot_kernel(Complex(x,y), max_iter)
        end
    end
    return result
end

@benchmark mandelbrot_mt()
# Starting julia with "julia --threads=16" gets me the following benchmark result on the same machine.
# BenchmarkTools.Trial: 1010 samples with 1 evaluation.
#  Range (min … max):  4.201 ms …  14.436 ms  ┊ GC (min … max): 0.00% … 0.00%
#  Time  (median):     4.807 ms               ┊ GC (median):    0.00%
#  Time  (mean ± σ):   4.935 ms ± 575.673 μs  ┊ GC (mean ± σ):  0.70% ± 3.91%
# 
#         ▁▃█▆█▆▄▃▁▂                                             
#   ▃▃▅▅▆▇██████████▇▆▆▆▆▅▄▄▅▃▃▃▃▃▃▃▂▃▂▃▂▁▁▂▂▂▂▁▁▂▁▂▁▂▂▂▁▂▁▂▁▁▂ ▄
#   4.2 ms          Histogram: frequency by time        7.26 ms <
# 
#  Memory estimate: 667.91 KiB, allocs estimate: 99.

Do you have a SIMD version yet for comparison?