Here what I am wondering about: ~ $ time python -c "print( 3^100000000000 % 102 )" 91 real 0m0.030s user 0m0.026s sys 0m0.004s ~ $ time julia -e "println( big(3)^100000000000 % 102 )" ERROR: OutOfMemoryError() Stacktrace: [1] pow_ui!(x::BigInt, a::BigInt, b::UInt64) @ Base.GMP.MPZ ./gmp.jl:179…

The state-of the art is currently that it is necessary to use specialized timing utilities to compare timings. Does this make the outcome more useful? The current outcome is that the results depend … varying between Python being four times faster than Julia up to Julia being four times faster than …

[image] oOosys: The comparison below appears a much better approach to see the performance of big integer arithmetic in Julia and Python, doesn’t it? No, you’re still doing the misguided practice of subtracting separate overhead timings and timing printing, and you added more issues: went b…

[image] Benny: ask a Python forum for their go-to packages. As stevengj has said, I believe it is a simple %timeit from ipython: In [1]: %timeit pow(3,100000000000,102) 2.82 µs ± 10.4 ns per loop (mean ± std. dev. of 7 runs, 100,000 loops each) For comparison, I get this in Julia julia> @b…

[image] Elrod: a simple %timeit from ipython That is one of the kernel magics I mentioned, it’s good enough for many people in practice including me, but I personally hesitate to end a recommendation there because of the reliance on IPython. I’m sure people who don’t use Jupyter or Spyder hav…

OK … it seems that Julia is four times faster … but … doesn’t it measure which optimization parameter were used to compile the Python C-code into an executable? Or how much the used CPU is adapted to the needs of LLVM, so that with another CPU and Python executable the outcome can be fully another o…

julia> @btime powermod(3,100000000000,102) ERROR: LoadError: UndefVarError: `@btime` not defined in expression starting at REPL[1]:1 What do I need to do to be able to reproduce your result?

Four times faster needs four times less CPU instructions to achieve the same result … doesn’t it? Or it uses instructions that take longer, or it is botlenecked on memory speed, or branch mis-prediction, or any of about 10000 possibilities.

julia> using BenchmarkTools julia> @btime powermod(3,100000000000,102) 1.107 μs (0 allocations: 0 bytes) 69 ipython3 (python3.10): In [1]: %timeit pow(3,100000000000,102) 2.01 µs ± 5.43 ns per loop (mean ± std. dev. of 7 runs, 100,000 loops each) ipython3 (python3.13): In [2]: %timeit pow(3,1…

[image] adienes: how many times does this question need to be asked and answered? Until the details of the question are improved to a degree that makes an answer which provides enough information along answering the question really useful …

[image] Benny: I’m sure people who don’t use Jupyter or Spyder I just started an ipython REPL from the command line. The ipython repl is far and away better than the regular python repl. [image] oOosys: OK … it seems that Julia is four times faster … but … doesn’t it measure which optimi…

Python big integer vs. Julia big integer arithmetic

General Usage Performance

Topic		Replies	Views
Wrong results in calculation Performance	21	3358	March 12, 2019
Why is python faster than Julia Performance	14	2022	March 12, 2020
Small benchmark Performance benchmark	14	2941	November 21, 2018
How does Julia compare to shell scripting using bash and its "friends" like bc on a Linux system? New to Julia question	40	1439	June 19, 2024
Help to get my slow Julia code to run as fast as Rust/Java/Lisp Performance	100	5368	August 6, 2021

Python big integer vs. Julia big integer arithmetic

Related topics