GamerOS Windows 10 & 11 DIY Preset

Ok, a question as I am new here and haven't really looked into 'gaming specific' builds since I generally Server LTSC versions (2019/2022) for desktop) since my priorities are 1) work and 2) gaming (in whatever spare time I have after #1) and look for stability/long life as most important.

The question I have, is looking through the presets in OP it appears a lot of this is related more to 'bloat' or disk space saving than anything. Has anyone actually run a comparison as to FPS or just general PC benchmarks of the preset against a base image? Or to quantify the effects?
Dirty secret -- disabling event logs is a huge part of the perf boost.

Yes, the cost of unwanted services does add up but "random" flushing of log buffers wreaks havoc on process prioritization (since it's a kernel level task). That's the hiccup you get in FPS games, writing logs to disk and possibly consolidating data.

What I would test:
1. Load the GamerOS preset, reset every NTLite screen except for Settings / Autologger tracing & Event Viewer channels
2. Save the preset. Open in Notepad, confirm that's the only tweaks.
3. Load your LTSC and normal preset. Load the autologger-only preset, using Overwrite.
4. Build ISO and compare.
 
Has anyone actually run a comparison as to FPS or just general PC benchmarks of the preset against a base image?
Benchmarks for builds on this forum haven't been posted yet. The closest information provided is screenshots of things like Task Manager resource consumption (processes/threads/handles/memory). This ends up being acceptable for a lot of people because the general principle is that as resource usage decreases, performance will improve to some degree. On that note, it should be mentioned that tweaking Windows isn't about increasing maximum frame rates, since that's mostly a hardware limitation, but rather improving DPC latency and 1% frame rate lows.
 
Non-existent file path: LOAD POSTSETUP.PS1 FILE UNDER USER EXECUTION

What does this mean? will it affect my fresh win10 install with gamerOS preset?
 
Non-existent file path: LOAD POSTSETUP.PS1 FILE UNDER USER EXECUTION
It's a placeholder in the preset to load an unfinished script. Txmmy hasn't fixed it, so everyone gets that warning.
Just smile and re-save the preset to clear the error.
 
Is Txmmy still around? Maybe we can get Txmmy or have a mod edit the original post on page 1 to do the following:

- remove the mod links
- shrink down the screenshot size
- add an edited preset with the script removed (as an attachment to the main post)
- add the script (as an attachment to the main post)
- add a brief comment about how to use the script
- add the info about restoring Bluetooth, Microsoft Store, and MS Accounts
- add info that the XML will work on W11, but it wasn't explicitly designed for it
- make a very clear note that the preset can be modified (DIY concept)
- any other info required to answer common questions

All of this comes up so frequently, a lot of questions would be avoided if all of this was on 1 single page, and it makes the most sense for that to be the main post, instead of having people jump around between 32 pages (and growing). NTLite is making the rounds in the YouTube world right now, with very sizable audiences, so threads like these are only going to keep growing in popularity.

Please don't mistake my intent here, I'm not condoning editing people's property without their permission, but there's already plenty of mod notes on this thread so I assume some permission has been granted, and these things wouldn't be harming the integrity of the build in any way. Txmmy has mentioned he's kind of done, and this thread has clearly taken a life of its own and grown to epic proportions, so some cleanup is warranted.
 
hopefully extra views brings in extra revenue to help feed the server hamsters and pay the internet bill each month.
downside is the the extra spammers new traffic brings in, i clocked 30 plus a couple of weekends back just on my watch alone.
 
Benchmarks for builds on this forum haven't been posted yet. The closest information provided is screenshots of things like Task Manager resource consumption (processes/threads/handles/memory). This ends up being acceptable for a lot of people because the general principle is that as resource usage decreases, performance will improve to some degree. On that note, it should be mentioned that tweaking Windows isn't about increasing maximum frame rates, since that's mostly a hardware limitation, but rather improving DPC latency and 1% frame rate lows.

understood. I was mainly looking for quantitative metrics for measurement. the screenshots are not something that can be measured/compared/or reproduced/validated (or to see if they are within statistical error). The OP preset file does a lot of changes (and from looking at it, at least 50% if not more are mainly the OP's preferences which wouldn't have any effect on gaming or anything else (e.g. changing RDP port to 55555 and similar).
 
...which wouldn't have any effect on gaming or anything else (e.g. changing RDP port to 55555 and similar).
I can't explain that either, I tried Googling it and found nothing, except some tiny excerpts about virtual machines. This can be easily changed though, since presets can be customized (load a preset and uncheck a tweak before processing the image or edit the XML in notepad and delete the lines). The majority of gamers will just blindly install scripts and images though, so long as there's some notion it will help them, because a lot of people can't be bothered to learn/test this stuff, since they only want to get back to gaming as fast as possible in the hopes they now perform better.

...the screenshots are not something that can be measured/compared/or reproduced/validated...
Yes and no, so long as the major variables are the same then Windows will consume very similar amounts on different hardware. The GamerOS does have substantially lower resource consumption compared to an unmodified Windows, and that will translate to all machines. The major variables that matter in benchmarking Windows are OS (W10), edition (Home/Pro), and version (21H2). The amount of RAM installed also matters heavily because Windows adjusts resource consumption based on that.

Some tips on benchmarking an operating system:
The first thing to do is create a baseline, which in this case means installing an unmodified Windows in a non-VM, as shown here (link). Then take whichever tools you want to get results from and run those benchmarks on the baseline. Tools to use should be Task Manager, LatencyMon, plus some others that measure CPU/GPU/Disk/Memory. Some free options are the Unigine benchmarks, 7-zip, CrystalDiskMark, NovaBench, TimerBench, CineBench, and PCMark10. For internet tweaking there's tools like speedtest.net to check speed/jitter.

Install a modified image, do the same tests and compare the results. Where many people get confused is by focusing on the scores and leaderboards. Instead, we need to look at the percentage difference between stock and modified. It doesn't matter if one benchmark went from 80 to 100, the number we want is the percentage increase (25%). That improvement should carry over to any machine, however the percentage will change based on hardware, since one computer may be highly CPU bound, while another isn't, but both machines still benefit to some degree.

The benchmarks that matter most are DPC latency (LatencyMon) and 1% lows (Unigine/CapFrameX) since reducing Windows overhead will primarily result in lower DPC latency, which translates into much better 1% lows, because stutters are usually caused by Windows background activity. For a visual example of this, check out a video TechYesCity did when they made a custom image and compared it to stock (link).

It should be noted that they used MSMG, and the GhostSpectre ISO as another reference, neither of which are supported on this forum. They went on to then make another video with NTLite, but haven't posted benchmark results yet. The takeaway is that their video is a great example of how much improvement can be seen, since in most games the 1% lows *doubled* in performance on a tweaked system. That's what we are primarily chasing when modifying Windows, because it will translate into better human performance while gaming.
 
Last edited:
Is Txmmy still around? Maybe we can get Txmmy or have a mod edit the original post on page 1 to do the following:
Please don't mistake my intent here, I'm not condoning editing people's property without their permission, but there's already plenty of mod notes on this thread so I assume some permission has been granted, and these things wouldn't be harming the integrity of the build in any way. Txmmy has mentioned he's kind of done, and this thread has clearly taken a life of its own and grown to epic proportions, so some cleanup is warranted.
I'm not a fan of revising history, especially if a thread has been viewed for a very long time, it should stay untouched.
[Except for that super-lite bragging guy we don't talk about.]

While not Txmmy, I don't think he would object to forking his preset, and starting a fresh thread (GamerOS Addendum) if you owned it. But be careful what you're asking for since the follow-up questions will be non-stop. We can always post a cross-link to the new thread.

There's not much in GamerOS component removals that can be improved. In fact the current version has stepped back to improve app compatibility. More work could be done on tweaks which NTLite doesn't handle natively.

The old-timers (before me) remember MT_'s Gaming Profile, which stopped after the author lost interest.
Windows 10 LTSC 1809 - Optimize Gaming/Poweruser/Runtime Profile, retaining maximum compatibility. For x64/UEFI systems

[garlin editorial] History might need to repeat itself.
 
Perhaps it's best to let this preset run its course until Microsoft or NTLite makes enough changes to naturally cause the preset to become obsolete, or until someone else wants to try their hand at this.

Chris Titus is already taking the lead on that right now, he's maintaining his own fork of this preset on his site, so we may see traffic to this thread eventually divert over there with enough time. Other popular channels, such as TechYesCity are starting to get into the NTLite scene now too. Some of the channels keep trying to host ISO downloads though, so that's something we have to be aware of on the forum when people come here asking for help, because when they post an XML it may not reflect what their image actually contains.
 
Last edited:
MT_ switched to tweaking Post Setup.
At the moment this is the best w10/w11 gamer thread we have so keep it open and see what happens. New more up to date gamer topics do appear from time to time and superseed older topics, just as this thread replaced MT_s thread so this will be replaced at some point but until then use this one.
 
Chris Titus is already taking the lead on that right now, he's maintaining his own fork of this preset on his site, so we may see traffic to this thread eventually divert over there with enough time. Other popular channels, such as TechYesCity are starting to get into the NTLite scene now too. Some of them keep trying to host ISO downloads though, so that's something we have to be aware of on the forum when people come here asking for help, because when we ask for an XML they will post one that doesn't reflect what their image actually contains.
Except Chris Titus doesn't participate here, and he's not obligated to take our feedback. Nor does he own a forum.

YouTube comments, and GitHub issue posts really don't address support questions. While Nuhi can be grateful that Chris does promote purchasing a paid license, the fact remains his community looks here for help.

If you don't step up and take ownership, others will shape it for you. Which is why an internal fork is better, rather than driven by someone else's need for YouTube traffic or Twitch traffic.
 
Dirty secret -- disabling event logs is a huge part of the perf boost.

Yes, the cost of unwanted services does add up but "random" flushing of log buffers wreaks havoc on process prioritization (since it's a kernel level task). That's the hiccup
Event Log service is needed by something that currently escapes me but something that users might want to keep running so is it enough to just remove the event log entries or disable the service to get the performance gain?
 
probably want to do your own measurements on your own system for any mentions of 'gains'. I mean event logger here having it disabled or running I haven't seen any real difference but then again in my situation I run my OS on a nvme raid array (hardware) that has 8GB of battery backed cache. So i/o, even random i/o is not really a bottleneck unless I'm dumping 10's of GiB/s.

Kind of the reason why I was looking for actual test data for all these suggestions. Theory is fine but it needs to be tested.
 
I want to enable `Microsoft-Windows-Subsystem-Linux` feature, but it's not there. Is it expected in this preset? Is there a way to put it back in without reinstalling the whole system?

I tried to add it manually from msixpackage file I've found on Microsoft pages, and after it installs I can run `wsl` command, but it still says that `Microsoft-Windows-Subsystem-Linux` feature is not found and fails to install.
 
Have you enabled it in windows features ?
it does seem like it is disabled by default within the iso
 
Back
Top