Please Be Patient! Old Computer at Work. forums forums Politics and other things Please Be Patient! Old Computer at Work.

Viewing 12 posts - 16 through 27 (of 27 total)
  • Author
  • #40736

    “Wearing out of solid state hard disks brought a question to my mind–what happens when solid state disks are used with Linux? Linux kernels typically use a dedicated partition on the disk, rather than a page file, as virtual memory.”

    As stated above, one can install Linux without using a swap file. But, the SSD controller should accommodate the re-writing (wear-leveling) anyway.

    I’ve used SSDs in my Linux machines for years with swap file, including a 24/7 web server that is constantly writing log files, etc. Checking wear-level indicator in S.M.A.R.T. after a couple of years of operation, I’ve still got many years of life left even using swap files. Modern SSDs have much longer lifespans than the original products – probably due to both improved wear leveling and improved flash technology.

    I invested in some of those CF to IDE adapters about ten years ago, but I’m about to throw them away. I just have no need for them anymore. These days I’m all about efficiency – saving power, smaller components, less heat and fan noise, less junk around the house. As I said above, virtual machines solve a lot of problems. Virtualbox is pretty awesome. I’ve got XP and Windows 7 VM’s on my Linux laptop to use when occasionally needed. XP is very quick. I’d guess the older Windows and DOS would be extremely fast. The fact that things can be backed up and replicated and ported to other hardware makes this even better than fooling with old power supplies and motherboards, etc. But it would be a little initial work getting them set up.


    I do enjoy using Linux and keep one or two machines in Linux to keep my hand in it. I’ve never gotten so deep under the hood with options though.

    I guess I might have better luck putting a DOS environment into a virtual box on Linux than on XP or 7, but the hangup would be that the software expects to print to LPT, and the OS has to be able to capture this and re-direct to network resources. Of course this should in theory be possible and might be more possible under Linux than under newer Windows. This of course assumes that you’re running on a motherboard that still has parallel ports, so that the OS thinks there still is LPT or COM to address.

    Sounds like a fun project for when I retire! Problem is, right now I gotta keep working and this is what’s working for now.

    You say you’ve used the CF to IDE adapters and have no use for them now. Did any of them fail on you as is being discussed here?


    I would be tempted to play with the compact flash adapters, if you don’t want them anymore. Are there any limits to the storage capacity of the flash cards that these adapters support?


    Actually, Alfredo, I see that you can buy these CF to IDE adapters for $1.24 each shipped via eBay from Malaysia or Singapore. It may not be worth the effort for me to get them to you.

    I used only of them in a long-term use project: an IP Cop firewall (Linux-based) in an old PC tower, something I retired about eight years ago (replaced with a DD-WRT router and, later, Tomato). Using the CF adapter worked fine – it was obviously faster and quieter than an ancient IDE drive but far slower than a modern SSD drive. In the case of IP Cop, once it booted, it stayed booted for months, so it wasn’t getting much of a challenge. At least I didn’t have to to fear a failing hard drive taking down my home network in case the thing power cycled or something.


    Radiogeek, serial and parallel ports are one of the more challenging issues with running an old OS in a virtual machine. There are USB to serial or USB to parallel adapters, but they are far from perfect and may require some tweaking. It is still possible to find old PC hardware with a genuine parallel port – I think I have an Intel Atom box laying around with one. Even if you don’t buy 2019 hardware, you might still find hardware only a few years old with real serial/parallel ports – much newer than the original hardware platforms, faster, more power-efficient, etc. If you have genuine parallel/serial ports, you can expose them to the virtual machine, and that is pretty likely to work.


    Radiogeek, serial and parallel ports are one of the more challenging issues with running an old OS in a virtual machine. There are USB to serial or USB to parallel adapters, but they are far from perfect and may require some tweaking

    Yup. So much easier to stick with the old OS and an older motherboard, so long as I can get reliable bootable IDE solutions. The adapters and a new CF card are cheap, and no one so far has told me they have actually failed on them, only theory about failure. I really like the ones with a female 40 pin connector, it goes directly onto the motherboard and I remove another cable!

    I’m rebuilding from scratch two more workstations with good motherboards I bought on Ebay. Ironically, lately I’m looking for old hardware that can run Win98, or newer 64bit hardware that still has a Win7pro license! I guess everything in between that I retire I should put Linux onto!


    Yeah, we just disagree about basic philosophy on this kind of stuff. I have no trouble recycling and re-using old hardware (Mom is still using my old 13 year old Dell laptop with Ubuntu Linux on it. I have two more of them that are identical – they are almost disposable, because they are almost worthless. If one breaks, she turns on one of the spares. She is basically using them as terminals.)

    But virtual machines are convenient and efficient, once you get them set up, and after that, I don’t have to worry about hunting for ancient hardware anymore. And I can have a bunch of different virtual machines on one box. It’s neat being able to set up different versions of Windows (and Linux) in different virtual machines – make snapshots to revert to if something goes wrong in something you tweak, make easy backups of your virtual machines, etc. Who wants to deal with backing up some ancient IDE drive? My virtual machine is just a big file I can back up as needed. The physical hardware becomes an abstraction.


    I was a bit in shock when I saw the weird situation going on with used 386SX motherboards. They list for $50-$300. I cannot imagine anybody wanting one of these that badly. Once the motherboard at, which cost me $0.00, fails, will be no more. The domain name will not be renewed. I cannot justify paying $300 for a motherboard that is nearly 30 years old.

    A few months ago, I contacted the person who, back in 1995, put the idea in my head of using an old computer as a miniature UNIX server. The 386 that she had at the time was a hand-me-down, and she was looking forward to replacing it with a Pentium. I think that she was surprised that anybody today is attempting to keep a 386 running as a Linux server.


    Yeah, we just disagree about basic philosophy on this kind of stuff.

    Nope. We agree on a great deal, just choose to work at it in slightly different ways.

    My basic philosophy of computers is … wait until the Dell leases are turned in on models a few years old. By then there will be service history and you can figure out which models to avoid. Then go to Ebay and buy them used for dirt cheap, and buy multiple copies of the same thing so you have spares. Just like what you said, except that if I have a laptop screen or mb fail, I just move the SSD to a spare and I’m up and running in minutes. That reminds me, I’ve only got three copies of my latest model laptop with Win7 on it, and all three are in use in my family. Need a spare!

    Quick research shows me that the machine I’m typing on now was released in 2007, and I’ve got Linux Mint (Cinnamon) running on it. We’re not that far apart! I look at the P4 challenge as the reason for the first time in decades I’m building working boxes from the case up.


    I don’t recall ever using one of the Tandy Color Computers. But I learned to program on my buddy’s TRS-80 Model III (black and white) one summer in junior high. His father, then a programmer himself, wisely bought his kid, my friend, one of these to learn on, and I got to learn on it too. We wrote BASIC programs only. It’s Z80 CPU machine language was too complicated for me, but when I bought my Vic-20, I found its 6502 much simpler to write machine code for.

    Andrew, my experience with Z80 and 6502 were similar. Using the Zero Page “STA ($80), X” style made for a lot of 16 bit registers. Took a bit of getting used to, but once I did, programming made sense, and was pretty fast!

    But, you missed out! The Color Computers used the Moto 6809, and that chip is just beautiful. I think it’s gotta be a contender, if not the king of 8 bit CPU’s. Tons of fun. I keep a Color Computer 3 handy to this day just for some fun here and there. It’s a joy to program in assembly language.

    Williams DEFENDER was created on a 1Mhz 6809. Crazy, for the time.


    Re: cassette

    UGH. I used cassette with the Atari and Color Computers. Fortunately, when I got an Apple //e, I also got a disk drive.

    The Atari was very slow, despite very sophisticated hardware for the task. POKEY chips contained an SIO (serial input output) subsystem that looked a little like USB does today. The OS (well, ROM) on those machines featured device independent I/O, and was capable of a whole lot more than most users experienced back in the day.

    Notably, the color computer, and I am pretty sure the Tandy TRS-80 machines, used meaningful filenames with cassette. It was possible to have a few programs on a tape and use them without too much hassle. The Color Computer had a 6 bit DAC and ADC that made for fast, pretty error tolerant cassette use. Best system I used. Got few errors, and fairly long programs were not too much of a PITA to develop on cassette.

    (The Atari was painful, got a disk as soon as I could)

    One of my favorite things to do with cassette was to make a bunch of C-5, 10, 15, and 20 minute tapes. I would get sets of those “Change your life” or “find Jesus” tapes in the thrifty stores. Almost all of them had great cases with the little screws and nice tape transport components.

    I would assemble tapes of various sizes for various tasks, or to fit programs I wanted to run low hassle. Drop the program on both sides, use closest length tape, and use normal bias, thick tape, and it all would run largely error free and fast. Insert it, FF or RW to leader, whichever was closest, press play and go.


    Re: SSD’s

    Honestly, I’ve had great results with the Samsung SSD devices, and am currently running an older OCD and one other off brand SSD.

    I no longer need to pound them to death with massive writes. Career change has brought me to a far less data heavy place for a while. (good)

    But, when I was running the crap out of them, I never got a failure, and I was writing seriously large data all the time doing proof of concept work on large enterprise software systems. (that’s become increasingly grueling work, and I needed the change)

    If you stick with the better names, and I strongly recommend Samsung, and larger drive sizes, you are unlikely to have any real problems with Linux. And that’s with a swap file active.

    For speed, using two of them, with one for swap, rocks and isn’t too much money.

    I’ve deployed SSD drives on quite a few computers over the last 6 years or so. One big failure. The rest are running nicely. These days, I will cycle them out once every 3 or so years. The older ones get used for projects, or swap, etc…

    One thing to be aware of though!

    They do not retain contents when powered down for very long periods of time. Hard drives will do that. And I know people who have written them full of data and then spin them up every so often. An SSD can just forget after a year or so.

    Power them, from time to time, and it’s all good.

    Little USB thumb drives have the same problem, just FYI.

Viewing 12 posts - 16 through 27 (of 27 total)
  • You must be logged in to reply to this topic.