Fix game stutter on Win 10 1703-1809

Btw mkb, seems like this guy https://n1kobg.blogspot.com/2018_08_12_archive.html?m=1, chef-koch has been stealing your msi_util. https://github.com/CHEF-KOCH/MSI-utility/releases/tag/2.0

Why don't you have a Github/other software colab site account?

 

He has not. That`s from his MSI utility v2 you linked:

- he took my first attempt in MSI field - the PowerShell script. I discontinued to support it and discourage to use it because of danger to execute it blindly (and in batch mode).

May be some day...
I do not think that source files of my apps are worth to look or collaborate at. All those apps are very simple and of very specific usage.
Also using github requires some changes to my current practices which is against my laziness.


PS I am curious where is Wagnard...

 

Well, dang, I was about to install that, but decided to do some research on it and came across this thread.

Is a memory purging solution even necessary on 2004 for a PC with 32GB of RAM?

 

It was never a question of Win10 version. It was a question of having stutters or not. If you have stutters then install solution and test. If stutters are gone you are good. If not then you can uninstall.
First of all I trust Wagnard. Second of all there are alternatives published in this thread.

 

I mentioned my Win10 version as Wagnard's software was developed with the "Creators Update" in mind (as clearly stated on his website). The Creators Update was... 1803? Many improvements have been made since them, particularly for gaming. 1909 was probably the first version since 1803 where I didn't have any major issues with games.

As for stuttering, I don't deal with them very often, unless it's a poorly coded Unity game where the devs forgot to enable exclusive full screen mode or flip model presentation, something that can by circumvented by using SpecialK. I've seen some users mention that memory management/purging is basically useless if you have more than 16GB of system RAM, hence my question.

Finally, I also trust Wagnard (been using DDU since forever), but if one of his programs really is phoning home for no reason, it would be nice to have a proper explanation as to what's exactly going on.

 

Of all people who use ISLC one user reported strange behaviour. May be this particular user infected the exe-file with something malicious (from his own rig).

And again, there is no theoretical benefits for using ISLC. That purging standby memory solution was found as an answer to strange stutters on some rigs across several versions of Win10. But many rigs with the same Win10 versions have not demonstrated such stutters in games. So the question is very practical: if you have stutters then you try the standby memory solution, if you have no stutters then do not waste your time.

PS I had no problems in games on Win7 - Win8.1 - Win10. I strongly doubt that Windows is getting optimized for games.

 

Did you test exactly what I did for BF1? If not, you can't say you did not have this issue. And yes, it was related to specific Windows 10 versions. It started with 1703, it was not an issue with 1607.

 

That this would be a problem with my system is highly unlikely. I find it much more reasonable that people don't use a firewall that really reports much of anything, like the windows pos firewall.

 

Creators Update is Windows 10 1709, many people now claim its the "best" version of Windows due to low DPC latency, but if you must use ILSC with that version, as it seems never got the standby memory issue fix, then it's just stupid to claim its the "best". Windows 10 1607 seems to be free of standby memory problems and MS claims the issue was fixed in a later 1803 update and in 1809 and upwards. I run 1809 now and there's no difference in games with or without ISLC, I get the same frames and occasional stutters, which seem to affect my computer a lot lately.

If the standby memory issue worries you that much, get an LTSB 1607 ISO (Enteprise), install it and call it a day, that version is quick, snappy and does not have any memory issues. I may just do that, because 1809 does not convince me one bit. I expected an improvement from 1803, but it's the same or worse.

 

Where did Microsoft state they fixed it? I'm "confident" that I would only have to install an older Nvidia driver to make it occur again, like I was able some months ago.
(Well, of course I would need an Nvidia card for this again instead of trusty old Polaris, which has never shown the issue.)

 

I love your passive-aggressive style. If you are gonna reply to someone, use the "REPLY" option or don't do it at all. I am not making things up. How about you read the very first message from this thread, in its entirety? Once do you do that, I can proceed to block you once and for all. Giving you a chance was a big mistake. Same for mbk1969, you want to talk to him or not? You try to invalidate what he is saying for god knows what reason in that head of yours, yet you don't reply directly to him, instead you just "comment". It's like you want to argue at what others say and want to "challenge" them, but you make it literally impossible unless they keep checking the threads, which few rarely do.

Don't have to prove anything to you (1607 does not have standby memory issue, 1703 and 1709 do, 1803 and upwards does not for most people. Take it or leave or go debate MS engineers), as you couldn't never prove me why HPET must be disabled on the BIOS because it is a "switch" other than because you believed so. You think I don't read what you post? I also saw when you said I defend anything this Fr33thy guy says, when it's precisely the contrary so you try to gas-light on purpose or you just don't care to read what others say, so why take you seriously for that matter then?

 

Wow, nice vendetta post. Anyhow, KB4458469 doesn't explicitly mention standby memory and, if at all, it only fixed the issue partially. It certainly didn't improve anything in BF1. I btw. just confirmed what Wagnard described later in this thread...

 

Btw, what exactly did you mean by "trying to phone home"?

 

ISLC has 0 network code, it doesn't even have code to check for a new update.
The only think it has to do with the internet is the link to the PayPal donate that's all.

How did you check and where does it call (ip), screenshot?

 

Im using WFC on medium filtering, so it first asks me if an app wants to connect to internet.

 

Thanks, can you also check if this is happening with the latest version? you are on a quite old one.

 

No sorry I cannot, as this is the one which works correctly. I'm not complaining about it though as I said I'm using WFC and I get to decide if a program connects to web.

 

no worries. I'm currently testing on my side. I did see connection open too. Currently investigating what they are, but at first look I think it is related to the fact the .exe is signed with a digital certificate and it validate online its authenticity. When I compile a new .exe with no certificated applied, I don't get theses connection.
I'm not a digital certificate guru but I think this could be the reason.

Other question, what is the problem with recent version?

Edit: I can confirm that 151.139.128.14 is the OCSP from comodo (from my digital certificate) address: ocsp.comodoca.com

 

Yes I didn't check now but the connections seem to be certificate update related of the exe.

Well, the latest versions are buggy when clearing standby memory, I think if I'm not mistaken I wrote you about that months ago in here or in your website. Old version works just fine for me right now on 1709 and I plan to upgrade Windows to 2004 or 2009 version anyways.

 

Microsoft said they fixed it

• Addresses an issue that ignores the MM_DONT_ZERO_ALLOCATION flag. This issue leads to degraded performance, and, occasionally, error 0x139 appears.

Standby list uses this flag, regions of memory in use by the standby list are available for immediate overwriting with new process data.

The issue was mitigated to a high degree following this, however what remains is that the standby cacher is extremely aggressive for affected users resulting in janky memory mapping.