How to accelerate typical matrix operations in a for loop? I am currently rewriting my Matlab code to Julia. However, I find it only 3.5X the speed as Matlab. Are there any ways to further improve the Julia speed? Like improving the matrix multiplication speed. function test(b) X=randn(16,12); l=…

No, it doesn’t. What is I and CCore(X’*b)? is it a function or multiplication? And put the code between two three back-tics like this: ``` code here ```

I apologize for the misunderstanding. function test(b) X=randn(16,12); l=100; local Weights for i=1:500000 CCore=X.'*X+eye(12,12)*l; Weights=CCore\(X.'*b); end return Weights end

OK., if you use StaticArrays you can get at least 2X speedup for free: julia> using StaticArrays julia> function test(b) X = @SMatrix randn(16,12) l = 100 local Weights for i = 1:500000 CCore = X'*X + I*l Weights = CCore \ (X'*b) en…

Matlab uses multi-threading internally. I’m not sure if that happens here, but you can make sure Julia uses it. First, make sure that Julia has access to several threads (on mac/linux you invoke export JULIA_NUM_THREADS=4 if you have 4 cores from the command line – not inside Julia itself.) Then yo…

[image] complexfilter: @btime test(randn(16,1)) [image] Seif_Shebl: @btime test(@SMatrix randn(16,1)) BTW, I think you actually should use randn(16), not randn(16, 1). It’s easy to stuck on the Matlabism here, but a vector has size (16,) not (16, 1).

The OP was just comparing two loops doing matrix operations, and yes MATLAB is multi-threaded by default, so enabling @threads in not unfair but may be in the future, Julia will use multi-threaded matrix operations by default (I’m not sure if it already does that currently).

OK, let me explain you the big picture to make sense of the loop. What I am doing is translating a toy image de-noising Matlab code to Julia code to see how Julia performs in generic image processing task. It only means to learn and test Julia in a fast way in the context of a typical image processi…

There are some points need to be modified: a). May could use the M=5.;N=3.; I=1:Int(fld(M,N)); I=1:Int(cld(M,N)); I=1:Int(round(M/N)); if the type of I isn’t Int; b). Could you explain each sub-modules of the Julia such as include(…). Do they recognize the functions of images processing? In ad…

Let me answer your second question first: include("PSNR.jl") # compute the peak signal-to-noise ratio of the images. include("Image_Crop.jl") # shrink the size of the image by cropping the pixels near the boundary include("Demo_Julia.jl") include("Image_Extend.jl") # extend the size of the image by…

OK, this is a lot to digest, and very little comments. It will take a lot of work and time to get into it. For your own sake and those who wish to help: Can you start by identifying bottlenecks? Look into profiling. In Matlab it’s easy (if you’re unsure how to start, just ask). In Julia it’s suppos…

How to accelerate matrix operations(multiplication, add, inverse) in a for loop?

General Usage Performance

Topic		Replies	Views
Benchmark MATLAB & Julia for Matrix Operations Performance	148	20826	October 15, 2019
Why is this Julia code considerably slower than Matlab New to Julia performance	64	8770	March 5, 2017
Julia is significantly slower (~18 x) than Matlab in vector and matrix algebra New to Julia	32	2240	June 25, 2023
Porting code from MatLab - performance tips New to Julia	18	566	June 26, 2024
Outperformed by Matlab Performance matlab , multithreading , linearalgebra , tullio , loopvectorization	54	4512	November 23, 2021

How to accelerate matrix operations(multiplication, add, inverse) in a for loop?

Related topics