Computer Slowdowns

[Lee DeRaud]

A spin-off from Kev's latest adventure...

The common syndrome of late-model Windows machines getting slower as time goes by got me thinking. Every noticeable slowdown I've encountered in recent years seemed to be due to either (1) indexing or (2) real-time AV monitoring. (By 'real-time', I mean at every file access, rather than at creation/modification time or scheduled scans.)

I gave up on indexing years ago (on Vista maybe?), although it seems to be getting harder to keep it disabled...seems like I need to check the setting every six months or so to make sure they haven't reset to default. Which leads us to...

I'm as paranoid as the next person (unless the 'next person' is Kev :) ), but I swore off RT monitoring after a few incidents where disabling it resolved "false-positive" problems that simply should never have happened. (E.g. preventing upgrade installs of known major-publisher apps.)

What I don't know is whether the virus signature database ever get "pruned". I'm wondering if every scan of a file is performed against every virus/variant signature that was downloaded since the OS was installed, even if the vulnerability it exploits no longer exists, and probably not in any optimal order. Over time, that can make every file access take longer.

Thoughts?

[Kev Williams]

Ya know, one thing I haven't tried yet, and I have the perfect computer to test, my T5400 hot rod that's down to 2 cylinders ;) , is to disable Defender, reboot and see if the computer perks up...

I DO know it IS working in real time because every so often I'll get "one of those emails" with a link that I KNOW is bad, and it'll get flagged within seconds.

And FWIW, I can't remember the last time that I got something bad that DIDN'T come in an email- does that mean the firewall may actually work?

My scary moment was this one-
468278
--I actually caught it before the AV did, I had Defender scan it; it was an attachment in an email-
this was in 2017, since then I can remember 4 or 5 or so AV messages saying it was taking care of 'a threat'...

ANYWAY, when I get a minute I'll run a test :)

[Frank Pratt]

I've really fallen behind on what goes on under the hood for a couple of reasons. I worked in a small office environment & somehow became the de facto IT guy. I had to learn how to fix it because no one else could. I'm now the only person in my office, and honestly, computer problems just seem to be pretty rare anymore.

A couple of years ago both my home & office computers had mobo problems so were both rebuilt using M2 NVMe drives for the system drives. Wow! From POST to desktop in 3 - 4 seconds. Anything to do with drive access just flies. I haven't noticed that either has slowed down. I'm sure they have because of all the additional software & data they've been loaded with since, but not that I can notice in everyday use. Both computers to have indexing enabled.

Earlier this year the HD on my 13 year old laptop, which had become almost unusable, had the HD finally die & I replaced it with an SSD. It now takes about 13 seconds to boot and working performance is excellent. No need at all for a new laptop.

The moral of my long winded story? Switch to an SSD and don't worry about indexing & some of the other HD intensive background tasks. Works for me, YMMV.

[Jim Becker]

I agree with Frank's recommendation. Operating systems these days have a huge amount of I/O to the local storage device where the OS files are..."yuge". A spinning platter drive really cannot keep up with that efficiently anymore and that's why boot times can be as long as they are sometimes, too, when that's the hardware being used. SSDs have substantially faster I/O to the system which is more in line with what the OS needs to function efficiently. OSs are so large at this point that they are constantly shuffling data and pieces of code in and out of use. The processors are not really the bottleneck.

This doesn't absolve one from doing reasonable OS and computer maintenance...and with Windows that occasionally means a "catch all" rebuild on some systems. A "filled up" storage device is also not a good thing...breathing room helps keep things moving, too, so offload things periodically to other storage from the primary unit.

[Lee DeRaud]

I agree with Frank's recommendation. Operating systems these days have a huge amount of I/O to the local storage device where the OS files are..."yuge". A spinning platter drive really cannot keep up with that efficiently anymore and that's why boot times can be as long as they are sometimes, too, when that's the hardware being used. SSDs have substantially faster I/O to the system which is more in line with what the OS needs to function efficiently. OSs are so large at this point that they are constantly shuffling data and pieces of code in and out of use. The processors are not really the bottleneck.

This doesn't absolve one from doing reasonable OS and computer maintenance...and with Windows that occasionally means a "catch all" rebuild on some systems. A "filled up" storage device is also not a good thing...breathing room helps keep things moving, too, so offload things periodically to other storage from the primary unit.A perfectly reasonable answer to a question I didn't ask. :) Again, I don't even care about raw disc access speed: there hasn't been a rotating HD in any of my machines for several years now. (Well, there are those monsters in the NAS, but that's not Windows.)

My (mostly curiosity) question has more to do with per-access AV checking for ever-growing numbers of threats that don't exist. The viruses it's looking for might still be circulating in the wild, but the holes they exploit may have been plugged many updates past. If every infectable file is being checked on every read against every signature in the database, the slowdown effect is going to show up sooner or later no matter how fast the storage medium is.

The answer may well be that modern AVs don't do what I just described, but I don't know, that's why I asked.

[Jim Becker]

Sorry, Lee. I guess I missed that Point and was focusing on Franks's response.

I honestly have not found any real performance impact from Defender on Windows or MalwareBytes on MacOS in their default settings.

[Lee DeRaud]

I honestly have not found any real performance impact from Defender on Windows or MalwareBytes on MacOS in their default settings.Agreed.

I'm mostly speculating on possible causes of the "old age slowdown" syndrome that seems to affect a lot of Windows machines. (At this point none of mine exhibit it.) It could well just be that people expect a "fresh" computer to be faster...and it usually is, for any number of reasons. But eventually the new wears off and expectation exceeds reality.

[Jim Becker]

I always found that Windows machines seemed to be like "trash collectors" over time; some from normal processes, logging, upgrades/updates, etc., (native processes) and some from applications. I based that on doing the "full monty reinstall" periodically which always cured things up. Honestly, that was a major reason I moved to another OS as my primary...I got tired of rebuilding even my work issued machines with unreasonable frequency to keep them running acceptably, so I went a different direction. It was BYOD and I used virtualization of Windows when I needed to run something that required Windows. I still do that, honestly, and it's down to exactly one application: my Vectric CAD/CAM software. Interestingly, the Windows VM stays lickety split quick, but I attribute that to it only being used occasionally so nothing "builds up", as it were. Win10 has had less issue with this than Win7 and previous did, at least for me.

[Lee DeRaud]

Interestingly, the Windows VM stays lickety split quick, but I attribute that to it only being used occasionally so nothing "builds up", as it were.Just curiosity again, but does a virtualized Win10 image undergo the same update regime as a native Win10 install?
Or does it behave like a "Groundhog Day" snapshot of its installed state?

[mike stenson]

Just curiosity again, but does a virtualized Win10 image undergo the same update regime as a native Win10 install?
Or does it behave like a "Groundhog Day" snapshot of its installed state?
Virtualization is only hardware simulation. So yes. It does. You can take snapshots though, and revert to those at any time.

You have to store those though.

[Kev Williams]

Don't know much about 'new' computers other than the one I do have stinks.

My slowdown problem, pretty sure MOST of it has nothing to do with Defender. And it came on like someone flipped a switch. I have 3 'most used' tool engravers in the basement shop, two are identical, one is fed by serial port, one by LPT, the 3rd is a 'newer' machine fed by USB. The identicals, once I hit the "run" button on the computer, I'd have about a 2-3 second wait for the 'data meter' to start moving, then the machine(s) start. The serial/LPT difference has never been noticeable. The USB machine had just about the same 2-3 second wait, then the machine would upload. The only real difference, the USB machine takes all engraving data near immediately, the others can take a couple of minutes to full load up the module, but that doesn't really matter as the machine runs as long as the data flows. My big engraver in the garage is just a big version of the USB machine downstairs (Gravograph IS machines), and the lag time was virtually the same.

I say "was" because one day, about a year ago, the 2-3 second wait suddenly went to 4-6 seconds, on both computers. And it's only gotten worse. Another lag-time anomaly, my 'killer' win7, when I'm in Corel and open the print driver for my Gravograph laser and click on 'properties' to set speed/power settings, used to be when I clicked on 'print' the driver came up in less than a second, and same with clicking the properties button. Now, it's 3 or more seconds wait just for the driver to show up, and 7-10 seconds for the properties menu! The old XP machine, those menus come up right now...

I just showed the wife a little while ago the difference- on my Dell Precision T5400, w/SSD drive, 32gigs ram, 8-core/twin 3.15ghz processors, took 15 seconds to load the program...
The new Dell Optiplex 790 with a just a hard drive, 4gigs of ram, 4-core 3.33ghz processor, loaded the same program in less than 5 seconds...

I DO know the new computer isn't running Defender in RT, and it's virus defs are 12 years old. I still need to test the 2 work computers, going to block them from the 'net, shut down the AV's, reboot and see what happens...

Found this online, 6 years old but good read on AV system loads of various AV brands...
https://www.av-test.org/en/news/endurance-test-does-antivirus-software-slow-down-pcs/

[Lee DeRaud]

Another lag-time anomaly, my 'killer' win7, when I'm in Corel and open the print driver for my Gravograph laser and click on 'properties' to set speed/power settings, used to be when I clicked on 'print' the driver came up in less than a second, and same with clicking the properties button. Now, it's 3 or more seconds wait just for the driver to show up, and 7-10 seconds for the properties menu! The old XP machine, those menus come up right now...Same version of Corel? I'm seeing similar print driver sluggishness in Corel, but it came on all of a sudden with my last upgrade (to CD2019). It affects all print drivers, from the ancient ULS laser to the newest Canon inkjet. There are some other timing oddities with that version as well, including this one I commented on two years ago: https://sawmillcreek.org/showthread.php?276543-CorelDraw-2019-Oddity

[Jim Becker]

Just curiosity again, but does a virtualized Win10 image undergo the same update regime as a native Win10 install?
Or does it behave like a "Groundhog Day" snapshot of its installed state?

Yes, I regularly update the images. I also recently updated my VM so that it's setup properly to support the Win11 requirements when it's made available to me. (I will not be able to do that on my CNC controller PC because the hardware will not support Win11, but that's OK...it runs great on Win10.

[Kev Williams]

I always found that Windows machines seemed to be like "trash collectors" over time; some from normal processes, logging, upgrades/updates, etc., (native processes) and some from applications. I based that on doing the "full monty reinstall" periodically which always cured things up.
^THIS^ is exactly why I'm putting myself thru this nightmare: The Trash. And the only place the trash comes from is the bleeping INTERNET.
--as the joke goes: "Doc, it hurts when I do like this!" -- "Then don't DO like that!" - Can't collect trash if I don't let the trash truck in! :D


Same version of Corel? I'm seeing similar print driver sluggishness in Corel, but it came on all of a sudden with my last upgrade
YES- I'm using x3 and x4 Corel... Just tested both, they're about identical- at the moment the driver itself is pretty snappy, the properties menu is taking 5 seconds. However, I re-booted last night just before shutting down, and hibernated, so I'm still running pretty 'fresh'. Tomorrow at this time, if I don't reboot in between those times will grow. ALL lag times will grow. My garage 7 has started a new 'thing' if it's been un-re-booted for awhile, when I switch windows from Firefox to anything else, I'll get a horizontal split screen for several seconds, half of the Firefox window on top, the new window on the bottom, and the dreaded spinning blue circle from hell and the ever-present 'Not Responding' on the top ribbon. I left the computer after waiting over a full minute the other day- when I got back it had cleared its throat, but I don't know how long it took. Since it only does this with a Firefox window, I'm assuming it's THEIR "trash" that's causing it. But it's only THAT computer...

[Mike Henderson]

I run three Windows machines - well, I run two and my wife runs one, but I do all the maintenance on hers. I haven't found slowdown problems with them in the last few years. Earlier, I used to notice that a computer would run slower with time. One thing may be the size of the SSDs. My guess theory is that as you fill up a disk (or SSD) it takes longer to do I/O. The SSDs I have on all of the machines now are less than half full.

Mike

[Kev Williams]

The 1/2t SSD in my slowpoke "killer" T5400 computer is slower right now than the HDD in my new 790, at least as to how each computer acts. I don't blame the drive, for one thing the motherboard in the 5400 isn't 'high-speed' capable. But I'm still puzzling over how 32gigs of ram is still loading SSD programs noticeably slower than the exact same programs load with 4gigs ram and a hard drive-?

Here's the video I made a few months ago, of this computer trying to play a simple youtube video with no other programs running- All I have to do is move the mouse around to screw up the playback... I'm seriously thinking just moving the mouse around REALLY should hog all the resources needed to play a video! ;)


https://youtu.be/rzIfJFU74zs

[mike stenson]

How full is the ssd, which one is it? Are you using trim, how about garbage collection? What other processes, not applications, are running? Does this only happen after it's been running memory intensive applications? Some applications have memory leaks, some just suck at memory management. Does it get better with a reboot?