Difference in performance between Windows and Linux

, ,
#1

Hello,

I’ve just recently encountered a big performance difference between my Linux Desktop at home and my laptop (Windows) at work.
Now I’m thinking about what causes that performance impact. Is it the processor? Is it the OS? The window manager?
On my Desktop I had Windows installed before and it wasn’t any better.
I get the best TiddlyWiki performance on Linux. Without a dedicated graphics card.

My Linux system:

Distribution: Arch Linux
Desktop: Gnome (on Wayland)

Does anybody experience the same performance impact of operating systems?

Thank you,
Simon Huber

#2

You don’t say if you are using the same browser ( and same version thereof ) - there are too many unspecified variables here…

1 Like
#3

Oh, that’s true. I’m testing on Firefox latest stable and Chrome latest stable on both Windows and Linux.
Here’s my Firefox version on Windows: 123.0 (64-Bit)
Here’s my Chrome version on Windows: Version 122.0.6261.70 (Official Build) (64-Bit)

I’ll check my versions on Linux later when I’m at home.

Thank you,
Simon

1 Like
#4

Don’t forget to compare hardware specs too!

#5

I do not see a big difference other than the expected one because of different CPU and memory.

I mainly use FF latest with Windows 11 Pro on a core i9-9900K, 32GB Ram, Desktop with GTX1080
Ubuntu 22.04 laptop, core i7 6560U 8GB, internal graphics.

IMO since TW does not really use the GPU, the graphics card should not make too much difference.

#6

I only have a windows set up that performs well so cant help too much, but is it possible for you too describe the kind of performace impacts you are getting?

  • loading
  • buttons
  • lists
  • refresh
  • text entry
  • text selection
  • saving

Etc…

With performance in general and especialy browsers that do a lot of caching another critical parameter can be they remaing storage space. Too little and things can slow down a lot.

#7

Hello @Scott_Sauyet

I found the difference on the same machine with Linux and Windows on it, so the hardware specs are the same. But it’s only a “felt difference”, I don’t have measurements

#8

Do Browsers use the gpu for rendering?

#9

Yes, for CSS transitions, 3d transforms, WebGL and drawing to the canvas. The first 2 may be of interest, but it’s still a bit of “black magic”.

TW itself uses some animation, which is done using transitions. The TW storyviews are responsible for “insert” and “remove” animations, which we see, when we open or close tiddlers in the story.

The pop-storyview is responsible for adding and removing tags in the tags editor. The list-widget has a “storyview” parameter, which allows us to define animations.

So if we would need new animations, it’s that mechanisms that should be used for TW. – The crux for CSS animations is, if you want to stop or re-start them, while the animation is still “running”. That can be tricky.

#10

It’s likely the OS. There’s a longstanding tradition of Linux machines outperforming Windows in a variety of similar tasks. I’ve even found that Linux with WINE will sometimes have higher performance with Windows software than even a stripped down Windows on the same machine. The problem can grow worse with some background processes often forced upon users by M$. Linux tends to have better resource management, and if you’re using TW broken down into many tiddlers (lots of little files), you’ll also find NTFS is significantly worse than ext4/btrfs/etc.

1 Like
#11

Does your work machine, regardless of OS or browser, have virus, malware detection, and other corporately mandated monitoring software continuously scanning for suspicious memory, disk, and network accesses? (Not even considering the amount of memory, CPU, and the resultant context switching those consume.)

In my experience those “keep our infrastructure safe” mandates can be big, often overlooked, drains on overall responsiveness in work environments that your home computer is less likely to have, even if machine specs are otherwise identical.

4 Likes
#12

I don’t want to get into an OS war but I don’t think it is valid to make such comparisons. There are so many implementations of Linux some designed for minimalism and small devices and similarly windows tends to be used on medium to large computers and can even be on high end servers. In both cases the level of RAM, processor and storage needs to be pitted against the number of applications, system redundancy, security, level of networking and so many parameters, that comparing is like apples to pairs.

  • For example a formula one vehicle can go much faster than my Subaru outback, but my outback can go far more places than a formula one, and there is even room for things in the glove box :nerd_face:

Just to put a final nail in the coffin of the validity of pitting one OS against another both need to use browsers to use tiddlywiki, and these are another layer in the comparison. When using tiddlywiki as your primary software platform, comparisons are impacted not only by the Browser, but how it is configured in the system it is running.

  • Because browsing is an open ended activity (the internet is big) most Browsers would limit their own RAM usage so as to, not cripple activity on the Host computer, so the browser does not slow the host. With heavy browser use the performance bottleneck can be this “browser imposed limit” and have nothing to do with the OS or hardware.
  • We needed to ask when does your browser start to use storage, what kind is it, and how much is available?.

I now have a 32GB RAM, 2TB SSD “intel NUC”, however when I had only 16GB RAM and hard disk, I found giving Chrome and Firefox permission to use up to 4GB each, boosted performance, substantially.

  • They tended to avoid using “too much” memory so they were not criticized for slowing down the host.
  • However now days both Browsers tend to use what they need so more physical RAM means they have access to more resources.

The point being other decisions have more impact that the chosen OS.

1 Like
#13

Yes, sir. I appreciate your arguments. I can agree that comparisons aren’t easy to make. I’ll add that defining the boundaries of the OS is a difficult problem in itself as well; even modern browsers seem to have a lot in common with OSes themselves. You got a nice machine there, and I imagine it does an excellent job, especially with you piloting it. Thank you for your insights.

#14

Computers

  1. Hosting computer with my TW file on the C: drive, it runs Windows Server 2019 with 64GB of memory.
  2. My work laptop with Windows 10 Pro and 32GB memory. From this laptop the C: drive from the TW hosting computer is mapped as a network drive.

I’m on Windows 10 on my laptop and have my TW file on a network drive which is actually on another computer I access via Windows Remote Desktop. The hosting computer with the network drive is some type of Windows Server, I don’t know what version. But it takes me 40 seconds to save a 4.2MB TW file when I save it from my laptop. That’s a very small file. But when I’m on the local PC via Remote Desktop where the TW files is stored, and open the TW file with Brave, saving is in a few seconds.

I use the latest Brave Browser to view and edit the TW file as Firefox would not work properly for saving. Meaning when I saved the TW file the Windows Save As dialog would pop up and the default directory in the Save As dialog was always wrong.

I never could figure out what was going on here.

#15

This is normal Browser behaviour. Browsers do not want websites knowing anything about the host computer, including where the file resides.

  • However it is possible to enhance your save button, to copy the full file and path name of your wiki to the clipboard, then launch the save as dialogue, and paste (ctrl-V) the full path and filename, then save (no need to navigate to the folder, as you provide it)

If you are getting the save as dialogue, then you may be relying on the “Default download to save mechanism” and may need to research a better save mechanism.

  • It sounds like there may be limitations on what you can install.
  • I mostly use Timimi (in the browser) for local files and tiddlyhost for online wikis.
  • Clearly this is where the performance issues occur, and there may be nothing within tiddlywiki that can help, you need to troubleshoot this process or move to a different way of using your wiki.
  • I do not think your problem is tiddlywiki but other things in your environment.