Hi, I’m new to Julia. So I used my existing MatLab code to make a counterpart for Julia. I heard Julia prefers the performance-demanding part inside a function, so I did it. However, still, the performance is far lagging behind that of MatLab. I want to know what makes my Julia code inefficient. T…

[image] Wooheon: Can you give me an example of how to use --check-bounds=no? I mean to start Julia like this: julia.exe --optimize=3 -t auto --check-bounds=no Then, include your file as: include("\path\to\file.jl") This will turn off bounds checking in the whole code. So, in debugging ph…

I haven’t grokked the code, but as a first idea here: it’s the inner loop body that should mostly be in a function. Can you try that (extract the body of the while to a function) and see if it runs faster?

[image] Wooheon: I didn’t quite get the idea of extracting the body of while loop into a function. I thought this was what I’d done. Are you talking about the second code? First one is MatLabs. I didn’t quite get the idea of extracting the body of while loop into a function. I thought this w…

The idea is: function main(…) while {condition} {result} = loop_body(…) end end function loop_body(…) … end

On first sight: You are creating global variables in the inner loop. That would be bad in any language on purely stylistic grounds, but in Julia working with non-const untyped global variables is particularly bad for performance. Have you had a chance to read Performance Tips · The Julia Language ?…

A quick guess: This line distance=maximum(TV-V); probably causes significant allocations that will be slowing down your code. TV-V would be allocating a new array each time, which would be much slower than the rest of the code combined. I.e. reserving the memory would be slower than actually running…

In Julia there is no need for a semicolon at the end of each line.

Is there a way to calculate the supnorm of two functions avoiding such a way like distance=maximum(TV-V)? I just measured with my phone, but the difference was so significant that the measurement didn’t matter. But I would like to use the package from now on, Thanks!

Thank you for the suggestion. I tried to change the code as below. #Now initialize a V array# V=zeros(n_k,n_z);i=1;k_index=Array{CartesianIndex{3}, 3} while distance>precision k_index=k_index; (TV,Tk_index,distance)=loop_body(V); println(i);println(distance); …

[image] Wooheon: (A,B)=findmax(logC::Array{Float64,3}+repeat(V*transpose(Pi_z),1,1,n_k).*beta,dims=1) I’m on holiday without my computer, so can’t be very helpful. But I can tell that this is inefficient. You never need to use repeat, and should not. This is one of the good things about broad…

And please do not put ; at the end of the lines…

My julia code is somehow much slower than the matlab code

New to Julia

Topic		Replies	Views
Julia vs Matlab - building a Jacobian matrix New to Julia	30	4523	March 28, 2019
Poor performance due to memory allocations? Performance memory-allocation	17	3065	January 15, 2019
Julia slower than Matlab & Python? No Performance economics , tensorflow , matlab , pytorch	119	18403	March 16, 2021
Why is this Julia code considerably slower than Matlab New to Julia performance	64	9037	March 5, 2017
Understanding Julia performance in simple finite difference code New to Julia	38	4809	June 7, 2018

My julia code is somehow much slower than the matlab code

Related topics