Why was file search much faster in Windows XP than in subsequent versions?
from monovergent@lemmy.ml to nostupidquestions@lemmy.world on 27 Jun 18:36
https://lemmy.ml/post/32354673
from monovergent@lemmy.ml to nostupidquestions@lemmy.world on 27 Jun 18:36
https://lemmy.ml/post/32354673
On Windows Vista and every subsequent version of Windows, if I search for a file and include the entire C:\ drive, I might very well have time to make tea or a sandwich while the search results come in. On Windows XP, using the search dialog with the animated dog, I can search the entire C:\ drive and expect it to be done in a minute or two, if not in seconds.
It can’t just be nostalgia; I can replicate these results on period-accurate hardware today. What changed with Vista to make file searching so much slower, even with indexing enabled?
#nostupidquestions
threaded - newest
Windows sucks more at every new release. Management issues. Prefer developing new shitty features instead of cleaning the bloat. Anything is possible.
10 was better than 8 IMO
8 I believe was the last in their “tick tock” strategy…
IE “ticks” were the ones where they threw a bunch of stuff at the wall to see what sticks, basically putting users as paying beta testers.
“Tocks” were when they’d basically look at the piles of complaints from their ticks, try and fix as many as they could, and impliment some of the small features people wanted.
Ticks: Win 95, Win ME + 2000, Vista, Windows 8.
Tocks: Win 3.1, Win 98, Win XP, Windows 7, "sorta windows 10, hard to really say that model was made to be the start of more or less a auto upgrading by default as the new normal)
8 was designed terribly, but the engineering was unbelievably good. It was more streamlined and stable than it had ever been, it was just skinned by a tablet-obsessed moron.
10 was a huge step back in technical quality, but it undid a lot of the stupid mistakes on the front-end.
(and they won’t even talk about 9. they act like it doesn’t exist!)
They called it 8.1
the ui was dogshit but it was extremely lightweight and fast on my potato with 512mb ram and 60gb ide hdd.
Have no idea; last version of Windows that I used was XP then my husband moved me over to Linux and I preferred it ever since
He’s a keeper.
Just get Everything Search and you’ll be able to search just as fast as you could in XP, and with no Bing spam messing up the results.
Funny¹ thing is that Everything (and similar programs like WizTree) can be that be that fast because Microsoft’s own NTFS file system has a built in file index, which is what Windows Search used back in XP; the search programs practically don’t have to do any work, NTFS has already done it for them.
Of course, though, that’ll give you the results you want, not the results Microsoft wants, which explains the change in later further enshittified versions of Windows.
1.– And by funny I mean not funny at all. Sad, in fact. Tragic, even, maybe.
Just make sure to add git and node_modules as exclusions
I’d do that, but changing to Ubuntu, but thanks.
I don’t think everything uses some kind of premade file index. Whenever I start up everything, it starts with indexing all my drives, one by one, churning them at 100% if I look at the task manager but everything even says so in the bottom left corner. it even stores hundreds of megabytes of that index in memory.
what it actually does, as I know, is that instead of going through the slower filesystem APIs, it first scans the MFT with its admin rights, and then listens for any changes through the usn journal. so it does quite some work, because afaik both of these are publicly undocumented, and then it even implenents a very quick search for the index that even supports pattern matching
I’ve noticed this too. I’ve given up on it and instruct others to look for programs in the start menu alphabetically instead of searching because even that is bad. Same with Outlook searches, I instruct people to use webmail because the searching works there.
Check your indexed folders settings. Newer windows relies much more on this, but does a great job when what you’re after is included.
You definitely should still check this, but even with proper indexing settings Windows is still garbage at search and has been since XP.
And this is coming from one of the only people who ever defends MS on Lemmy.
I fucking hate Microsoft search. Microsoft controls the OS, they control the file system, why the fuck do they suck so bad at file search?
I’ve been using a program named Everything for file search, it’s everything the built-in search should be.
I don’t think it’s a coincidence that MS fixed Windows search when Google had its Google Desktop search product and Windows Search went back to horrible when Google discontinued Desktop Search.
You can find files faster on Windows by using the command line dir command with recursion switch and watch every directory tree scroll by until it finds the file than wait for the GUI even when Indexed.
I’m talking out of my ass, but I’m assuming it’s because of indexing. Operating Systems nowadays use indexing for searching your system, and it can be fast IF the file you’re looking for was indexed. That’s why it routinely re-indexes your entire system. It might take longer if the file wasn’t included. With file systems getting larger exponentially, indexing can be more efficient. Whereas before, the OS literally just goes over all your files to find a match.
Yes, but no.
The NTFS file system does maintain an index, and software like Everything Search or WizTree can use it to produce almost instantaneous results (probably faster than back in the XP days, even with larger discs).
The problem is that Windows Search stopped using the damn index for some reason (probably to provide sponsored web results and whatnot instead of whatever you were looking for).
Everything they’ve made recently is so utterly shit that I’ve started to think they actually want to slow down and frustrated the user. Like that better engagement figures or something. I’ve never been so regularly pissed off with how fucking shit their products are.
The question is basically answered now, so I’ll just drop this video here for some additional context about Microsoft’s history of trying to build a file system that solves the problem, and the challenges they faced even in the early XP days:
www.youtube.com/watch?v=h5d5H92c4Mk
tl;dw: MS tried to understand the context of each file, not just the name. Once you add dozens of pieces of metadata to each of tens of thousands of files (even 20+ years ago), the whole system became too difficult for them to properly index and manage efficiently.
Only by virtue of it not also searching the internet for shit. It still took forever to find anything on a large drive. But that could just be because we all had slow spinning platter hard drives and not SSDs. However, the fact I can expect to wait 5-10 minutes for a search of a file I know to exist on my current system, which is the same as back when I was on XP tells me that the newer software may just suck complete balls.
There’s a term for what you are describing: enshittification
Because it didn’t index file contents back then.
Also, storage wasn’t cheap so people deleted stuff back then.
Also, that shit was slow back then too. Searching a file share was awful. Now it’s a magically federated index. It used to be so slow you could buy a google appliance for indexing your data.
Back in the day, you scanned the disk. Folder by folder, file by file. If what you were looking for appeared early in your search, you were golden. It turns out, though, that scanning a filesystem is computationally very complex and takes a long time. Not something you might notice so much on a PC, but something that you would notice on a server. So, instead, you want to index the disk, slowly and over time, and then you search against the index. This works well in a server, but no so much on a workstation. Well there’s really no difference between Windows 11 and Windows Server 2024 except for some fine tuning of resource allocation. Essentially, you get the very (for desktops) ineffective server version.
The index is there (the NTFS file system maintains it automatically) and is fast (as programs like Everything Search demonstrate)… Windows Search is simply not using it anymore, probably so it can shove sponsored shit in the results, or maybe due to lost knowledge due to lay-offs.
dos had dir/s/*blah.wtf - pipe it into a txt file for the results if there’s too many. or /p
but yeah, on windows - everything. everything is the best search I’ve ever used, it updates (near as I can tell) instantly, and just freakin works great.
www.voidtools.com/support/everything/
Works well.
Not a Windows user but this was my solution when I was
For one, chances are your HDD size was significantly smaller than your current one.
Yet HDDs were also much slower than SSDs
Not really in terms of reading a massive amount of tiny files.
Of course SSDs are still much faster reading massive amounts of tiny files than HDDs are. Obviously random read speeds are much, much better, but even sequential reads of tiny files are a lot faster.
If you disagree, please provide numbers or references.
I love the “you’re wrong and if you disagree provide sources” while not providing any sources yourself lol. Amazing.
It’s fairly common knowledge that SSDs outperform HDDs in both sequential and random reads, and while the file size & number of files have an impact, it doesn’t negate this difference.
A quick search confirmed that SSDs perform better in your scenario than HDDs. I don’t care enough to spend time finding proper references, because again - this is simply common knowledge.
Still no sources. Interesting.
Moving/copying/reading/deleting tonnes of tiny files isn’t significantly faster on an ssd because the requirements for doing so are not limited by HDDs in the first place.
You mean the physical actuator moving a read/write head over a spinning platter? Which limits its traversal speed over its physical media? Which severely hampers its ability to read data from random locations?
You mean that kind of limitation? The kind of limitation that is A core part of how a hard drive works?
That?
I would highly recommend that you learn what a hard drive is before you start commenting about its its performance characteristics. 🤦🤦🤦
For everyone else in the thread, remember that arguing with an idiot is always a losing battle because they will drag you down to their level and win with experience.
Again - there’s no significant real world difference between an SSD and a HDD in the scenario I’m describing. Neither drive types are the limiting factor in speed of the operation. You act like HDDs take a long time to seek data lol.
I can pretty much guarantee you I’ve got more experience with data and drives than you do. Theoretical speeds and performance are just that - theoretical. The only way you ever get close to them are transferring a single huge file.
Your last little pot shot is ironic and hilarious.
This is like asking for a source for common sense statements.
HDDs are pretty terrible at random IO, which is what reading many small files tends to be. This is because they have a literal mechanical arm with a tiny magnet on the end that needs to move around to read sectors on a spinning platter. The physical limitations of how quickly the read right head can traverse limits it’s random I/O capabilities.
This makes hard drives, abysmal, at random I/O. And why defragmenting is a thing.
This is common knowledge for anyone in it and easy knowledge to obtain by reading a Wikipedia page.
SSDs are great at random I/O. They do not have physical components that need to move in order to read from random locations they generally perform equally as well from reading any location. Meaning their random I/O capabilities are significantly better.
The difference isn’t significant in this situation. You’re acting like HDDs are floppy disks lol. Their random IO is not “pretty terrible”.
IF you turn off searching the internet and fetching ads, file search can still be fast. However Microsoft keeps resetting that so I just gave up - winfdows search is not worth ising
i feel the same way about task manager windows 7’s task manager was much faster i swear!
switch to linux mint and all these issues poof away