When I was younger, my friends and I joked about making a "Disk Fragmenter" which would enumerate through every file on the disk and split them into non-contiguous clusters to make sure that the actuator arm on your platter-drive gets a PROPER WORKOUT TO FEEL THE BURN.
andikleen2 4 hours ago [-]
Such tools really exist and they're used for testing file systems, so that you have a more realistic "aged" layout on a freshly created test file systems.
We also have memory fragmenters to do the same for memory for the same reasons.
walrus01 1 hours ago [-]
This makes me wonder at the absurdity (yet plausible implementation??!?!) of a tool to test something like ZFS filesystem limits, as they are quite gargantuan in many possible measures.
rbanffy 52 minutes ago [-]
On CoW systems, file and block deduplication and writes to the middle of a file will do that for you.
alwa 4 hours ago [-]
Would those tools be primarily proprietary, or do some live in the open-source world too? I’d be interested in the techniques they use to aim for realism there.
Admittedly I haven’t made much of a search, but various search engines and autocorrects seem really convinced that I mean “defragmenter” instead of “fragmenter,” despite myself…
andikleen2 4 hours ago [-]
It's usually custom scripts, I'm not aware of a standard tool. You fill up the file system and then do lots of operations that cause fragmentation.
Actually here is a recent paper discussing this (although the basic concept is much older):
Perhaps it's appropriate for me to link to the free game I made where you play a taxi driving through Disk Defragmenter and delivering stray blocks of data to their final locations: https://safetystoatstudios.itch.io/ddr. It was pretty fun to watch Defragmenter as a kid, but it would have been better if it was interactive, kind of like the 3D maze screensaver.
suprjami 2 hours ago [-]
AND you made it for Sega Genesis! That's so good.
singiamtel 5 hours ago [-]
There's something satisfying about defragging that has been lost to time. Like oiling a squeaky engine
101008 2 hours ago [-]
I agree. It may sound elitist but back then you need to understand a few concepts to use a computer. From simple things like a filesystem (folders, file, different file types, things my 19 years old nephew don't know), to how to install software, how to download software (!) and then things even a bit more technical: an operative system, how it works, booting, drivers, etc.
Now everything is just easier (which is better for most people!) but I miss the old days.
jlarocco 1 hours ago [-]
I've said the same thing about using Linux.
20 years ago if somebody told you they used Linux it really meant something - they weren't running WSL, they weren't running a VM in EC2, they probably didn't power on and immediately have things working, and they probably learned a ton just getting to a login screen for the first time. Spinning up a VM just doesn't give the same learning experience.
walrus01 1 hours ago [-]
That and many desktop PCs making a kind of grinding noise from the floppy drive on start up, along with the bios "BEEP", as the default boot order option was to seek booting from A: (and possibly B:) floppy drives before the hard drive.
slig 4 hours ago [-]
Degaussing too.
rbanffy 58 minutes ago [-]
I once worked with a gigantic Apple (Sony, really) monitor on my desk. Every time I turned it on, it degaussed the monitors nearby.
So many of us spent so much time staring at the barrel of a particle accelerator.
gaudystead 2 hours ago [-]
I was going to comment this same thing until I saw yours. I miss degaussing... Especially after holding a magnet to the CRT to get it to display fun colors it wasn't meant to and crossing your fingers that the degauss would fix the damage you'd wrought.
koalalimpo 3 hours ago [-]
Beuunnnnng
vessenes 5 hours ago [-]
I love this. I wanted the disk noise, and I got it. I will say that’s a lot faster than my drives were in the late 1990s though.
haykuro 4 hours ago [-]
The speed is determined by a simple 1000 / <iops> per disk (there's different disks if you click "Settings").
If you inspect and go to sources and look for "1e3 /" you'll find the line responsible, you can set a breakpoint and change e.disk.iops to 1000 which will make the timeout per row 1s instead of 2-4s, so the whole thing goes faster.
cosmic_cheese 4 hours ago [-]
I recall there being some techno-mysticism surrounding defragging regarding how often and when it was necessary, what it did and didn’t help, etc. I also remember the claim “Macs don’t need defragging” but I don’t know if HFS(+) was actually any less prone to fragmentation than FAT32 was. Companies like Symantec were happy to sell defragging tools to Mac users regardless.
suprjami 2 hours ago [-]
People used to say the same about Linux ext4, it didn't need defragmenting.
afaics all that meant was file allocations make a little extra space at the end, so if the file expands it doesn't need to allocate a new block.
If one keeps files around for a long time and constantly adds to them, then the files get fragmented on any filesystem.
The only way to avoid this is pre-allocate all files to their maximum possible size and never exceed that size.
Hobadee 5 hours ago [-]
I love it. A couple notes:
- This is about 10x a real defrag speed - WAAAY too fast
- IIRC the "Data that will not be moved" was typically lumped at the beginning of the drive, not spread out all over.
vunderba 5 hours ago [-]
I'm a stickler for accuracy too. The defragged blocks in the original Win9x program weren't solid blue - they had a hatch pattern texture to them. I remember it way too vividly.
Source: Watching a disk defrag was every millenial kids equivalent of going to the laundromat and watching laundry tumble.
here is another version i found when googling these things that seems somewhat similar, but with a few more "accurate" features. also, has the source available in a single file.
I love this. However, Amiga’s Reorg program had a vastly better display, with a larger, resizable window and individual pixels instead of large boxes. You could see much more of the disk at once and get a feel for what the entire operation was doing. Large holes would appear in the middle of a hunk of data, then large files would get moved into the hole, leaving new holes where their incontiguous blocks use to be. Defrag was nice. Reorg was captivating. I miss it.
pjot 5 hours ago [-]
Will need to degauss after this
stavros 5 hours ago [-]
What, you don't need the data any more? Or are you conflating screen degaussing?
birdsongs 4 hours ago [-]
We used to degauss the CRTs because it looked cool and there was nothing else to do. Very adjacent to watching the defrag window maximised.
I did it in computer lab and showed others and the teacher lost it, thought we were breaking things.
pjot 2 hours ago [-]
It also made a really satisfying “boyoyoyinnng” sound
roger_ 5 hours ago [-]
My computer's definitely gonna be faster after this.
NBJack 5 hours ago [-]
I do laugh in that in all the times I ever did this for real (even back to my massive 170MB drive), it never felt much faster.
chasil 4 hours ago [-]
At that point, you had to deal with questions of interleave.
Typically it would be getting some bits from other parts of the drive and there would be a few (that many) bits it wouldn't move because they were the swap file and DOS boot system files.
pyaamb 5 hours ago [-]
Need this but for AI models. Sleep mode where they self prune and optimize overnight for their daily purpose.
There is even a click noise when you idly tap the screen.
As great as it is, way too fast.
RetroTechie 5 hours ago [-]
4 Drives, 5.5G total. Luxury!
Per-drive individual sounds, and BSoD included. Nice touches, love it!
MarvinYork 5 hours ago [-]
The sound killed me... :D
chasil 4 hours ago [-]
Defragmenting XFS is much less dramatic.
stavros 5 hours ago [-]
I need to set up a Windows 95 computer with a hard drive and stream the defragging online, complete with live audio.
brcmthrowaway 5 hours ago [-]
What were the major filesystem technology improvements that made this redundant?
predictsoft 5 hours ago [-]
I think NTFS and SSD disks.
WalterGR 4 hours ago [-]
Fragmentation is an issue with any file system. SSDs dramatically reduce the cost of fragmentation but do not eliminate it, as non-sequential access is still slower than sequential access.
megagpt1 5 hours ago [-]
FAT32 allocates the first free block. IIRC, file order must match disk order because reasons. Ext2 prioritized allocating the middle of free space to allow fragment-free growth on both sides and it became a standard filesystem feature after that.
wild_pointer 5 hours ago [-]
Legend
pir8life4me 5 hours ago [-]
Brilliant. Whirring disk noise is nice touch.
If AI generated this is what the AI revolution is all about.
Can we get a cross platform screensaver?
josefrichter 5 hours ago [-]
bill gates, I want my time back!
retired 4 hours ago [-]
This scratched an itch for me.
I’m going to watch Johnny Castaway now.
delduca 4 hours ago [-]
My YouTube
olivierestsage 5 hours ago [-]
Hot
Rendered at 04:31:13 GMT+0000 (Coordinated Universal Time) with Vercel.
We also have memory fragmenters to do the same for memory for the same reasons.
Admittedly I haven’t made much of a search, but various search engines and autocorrects seem really convinced that I mean “defragmenter” instead of “fragmenter,” despite myself…
Actually here is a recent paper discussing this (although the basic concept is much older):
https://arxiv.org/abs/2401.08858
Now everything is just easier (which is better for most people!) but I miss the old days.
20 years ago if somebody told you they used Linux it really meant something - they weren't running WSL, they weren't running a VM in EC2, they probably didn't power on and immediately have things working, and they probably learned a ton just getting to a login screen for the first time. Spinning up a VM just doesn't give the same learning experience.
So many of us spent so much time staring at the barrel of a particle accelerator.
If you inspect and go to sources and look for "1e3 /" you'll find the line responsible, you can set a breakpoint and change e.disk.iops to 1000 which will make the timeout per row 1s instead of 2-4s, so the whole thing goes faster.
afaics all that meant was file allocations make a little extra space at the end, so if the file expands it doesn't need to allocate a new block.
If one keeps files around for a long time and constantly adds to them, then the files get fragmented on any filesystem.
The only way to avoid this is pre-allocate all files to their maximum possible size and never exceed that size.
Source: Watching a disk defrag was every millenial kids equivalent of going to the laundromat and watching laundry tumble.
https://youtu.be/dc_SDyLYq3U?t=347
https://morello.dev/blog/defrag98-rebuild
what it looked like before
https://morello.dev/blog/windows-98-defrag-simulator
a previous discussion:
https://news.ycombinator.com/item?id=40962195
here is another version i found when googling these things that seems somewhat similar, but with a few more "accurate" features. also, has the source available in a single file.
https://horatioconkerhead.github.io/win98-defrag/
if you want the real deal, you can run windows 95 in the browser quick. it took running it a few times to get it to actually defrag:
https://archive.org/details/win95_in_dosbox
I did it in computer lab and showed others and the teacher lost it, thought we were breaking things.
https://en.wikipedia.org/wiki/Interleaving_(disk_storage)
When I changed it between drives and back I lost my progress :/
Typically it would be getting some bits from other parts of the drive and there would be a few (that many) bits it wouldn't move because they were the swap file and DOS boot system files.
As great as it is, way too fast.
Per-drive individual sounds, and BSoD included. Nice touches, love it!
If AI generated this is what the AI revolution is all about.
Can we get a cross platform screensaver?
I’m going to watch Johnny Castaway now.