I have a similar attitude to hardware, but you don’t have to settle for a blurry screen as the x200 has. My approach is to buy used or refurbished machines from a few years ago. I spend < $200 on a machine that runs Linux fast and has a high resolution screen. One big advantage is that losing a computer to theft or breakage is not a big deal.
It's interesting you had problems with the AX200 chip, I bought one for my X13 specifically because it has good kernel drivers.
On Ubuntu, there's a "hardware enablement kernel" package, linux-generic-hwe-22.04, that's just the kernel from the latest Ubuntu back-ported to the LTS version. I find it tends to fix most issues with newer hardware. Maybe PopOS has something similar?
Anyways, I found the MIMO in wifi 6 was better enough that I even upgraded another old laptop I have with an AX210 I found for $14.
That AX200 bootup problem occurs on dual-boot Windows/Linux systems. The solution is to turn off Windows "Fast Startup" so that Windows does not leave the AX200 in a "locked" state between reboots.
Also, the latest Intel iwlwifi firmare is slowly evolving to address crummy AX200 speed issues that Windows does not suffer. On one of my laptops the AX200 is all over the place speedwise, but running wavemon shows it stabilize in the 600 range of rx and tx after awhile. Here's hoping for progress with that firmware on Linux.
At the moment I have two computers in my apartment. They are identical Microsoft Surface Pro 3 machines. I got one from ebay for $150 and the other from Amazon for (I don’t remember) less than $200. The SP3 runs linux very well, especially using the surface-pro kernel. My good experience may be partially due to the fact that I don’t use a “desktop”, but use the dwm window manager. It has a 216 dpi screen with good image characteristics. Everything (touch screen, bluetooth, cameras, etc.) works.
I like having two identical machines because I keep them cloned continuously using my wifi network. If one dies I can just switch to the other with no data loss or interruption. I haven’t had to do this yet; the hardware seems reliable.
I like this type of machine because it’s a tablet with the power of a modest laptop. I can pop in in a backpack for travel, take it to the couch to read on, or use it like a laptop with a bluetooth keyboard. Having the keyboard separate from the screen is better for ergonomics; I much prefer these to laptops.
I have one problem, and it’s pretty annoying: I can’t seem to use video conferencing sites without problems. Zoom, jitsi, they all overload the system and my microphone stops working after a short time. I kept a Windows partition on one of the SP3s and boot into it just for Zoom meetings. If anyone has any suggestions, I’m eager to solve this.
How's support for these machines under the mainline kernel these days? Why are we still relying on downstream kernel forks even for this old hardware? Shouldn't support be completely upstreamed by now?
The mainline kernel works but I think the linux-surface kernel (https://github.com/linux-surface/linux-surface -- I misnamed it in my previous comment) should be used. I accidentally booted one of the SP3s with the mainline kernel and saw some WiFi driver problems, though I can’t say I’ve tested this extensively.
That's why he doesn't like Rust is so big on C and Go.
I don't buy into this mentality. It's just wasting time. Your user suffers due to bloat and slowness on commercial software (like websites full of tracking and heavy UI design) due to economical reasons (no benefit in optimizing user perf for the company) and not due to system software like one Drew writes.
It's just kind of a LARP of "old school" and "minimalistic", which is fine. Just let's not pretend there's some practical benefit.
Developers should be using hardware comparable to what their intended audience uses, at least some of the time. It's hard to effectively address the needs of your audience if you aren't experiencing their problems.
I have to disagree, devs should use hardware appropriate to their development environment. They should test on hardware appropriate to their target environment (or at least a simulated version of it). Insisting on developing on underpowered systems can block you from using some more useful development tooling like many profilers and some debuggers effectively or even entirely. You need to know that it works in the target environment, but if the target environment is a 2005-era CPU because the customer wants a "ruggedized" laptop (aka, overpriced old hardware in a very tough case), there's no reason to restrict your devs to one core and 4GB of RAM for their development tools.
<< It's just kind of a LARP of "old school" and "minimalistic", which is fine. Just let's not pretend there's some practical benefit.
Hmm. Like with most things, there are benefits and drawbacks to this approach, but outright dismissal as 'LARP' is, at best, uncharitable. Are you sure you cannot come up with at least one good practical reason?
I'm not dismissing it. Doing stuff certain way because it makes them fun for you is OK. Myself I'm not far from Drew with my dotfiles, doing everything in CLI and being a die-hard Vimer. And I love embedded, low-powered devices too. But I'm not going to pretend there's some big practical reason to do things this way. My VSCode-using mouse-clicking friends are as productive as I am, and produce similar software.
The idea that using underpowered computer makes you write some faster or better software etc. is just silly. Especially that all the really fast and well known projects in last few years tend to come from Rust people, with their super heavy compiler that has no respect for RAM and storage usage.
It's by definition always better to have a faster box than slower one. You want to see how it works on much slower boxes? Run it with a CPU quota or in a VM. People don't do it, not because they can't, or because they have too fast computers, but because they have other priorities.
I did not take his point to be that low powered hardware was superior. Rather, he states that older hardware is fine and good enough. That has been my experience as well. You state something similar above.
<< The idea that using underpowered computer makes you write some faster or better software etc. is just silly.
I will admit that I would rather someone with more experience commented on this rather than me.
That said, I disagree on general principle.
When you are restrained ( sometimes severely restrained ) by the environment, the net result is that you use resources a lot more efficiently. I absolutely get that we are kinda spoiled now with crazy hardware that is basically a supercomputer from not so long ago and it is hard to let that convenience and power go, but that lends to the bloated software landscape we have today that manages to overuse just about every resource available.
With limited power, you are effectively forced to write better. It is not completely unlike training with weights. It is not silly. It produces real results.
Note however that (due to Intel's complacence, AMD's inability to compete in the construction vehicle series of chips era, crossing several good-enough performance lines, upgradability and the general slowing of Moore's law at least for CPUs) using a 2012 CPU in 2022 is far more viable than using a 2002 CPU in 2012.
In 2022 a Sandy Bridge (2012) computer is still a viable computer and viable for the foreseeable future. In 2012, a late Pentium III/early Pentium 4 was not yet an antique but already severely outdated.
I'd argue that's a myth. If you use peripherals from 2012, and do simple office tasks, maybe. Try watching 4K (or FHD 50 fps) video on a 4K display on Sandy Bridge with a GPU from that era. Try editing iPhone video and raw photos. Even compiling Linux is order of magnitude faster, because of 20 faster threads instead of 4.
Speaking as someone who did have a computer from 10 years ago as the main computer until a year ago, no, it's very realistic. At no point did I come across a game that was "my computer isn't powerful enough to play this game." Even compiling large projects is less helpful than you'd think: a clobber build is still context-switch-while-it-finishes, and incremental builds usually don't build enough for all of the extra threads to be worth it.
Sure, maybe my computer might have screamed at me if I tried to watch AV1 or H.265 4K video, but... my computer is 10 years old and so is my monitor, so I don't have a 4K monitor. And, quite frankly, I don't see the point upgrading from 1080 to 4K.
Probably not that many, but the parent post to mine specifically mentioned editing iPhone footage.
A lot of people editing iPhone footage are going to use a phone-native app like CapCut, but even with that coping with the horribly inefficient compression of h.264 and h.265 footage gets them up to egg-frying temperature very quickly.
Editing 4k videos/raw photos is an outlier as far as PC tasks go.
Your argument "that's a myth" is based on a task that's so small in terms of percentage it's literally a rounding error.
I'm typing this on a 2010 computer (i7 870 @ 2.93GHz, 16GB RAM), which I also use to write software (C, Lazarus, JS, Go using VSCode, Vim, Emacs), browse web, use web apps (miro, tinkercad, asciiflow, a few others).
Inkscape works acceptably, light image retouching using Gimp works acceptably.
I used to play Starcraft II, Far Cry 3/4/5 on this machine until an upgrade nuked my Wine installation and I didn't bother to fix that.
My steam library is filled with games that work on this machine.
Out of the thousands of tasks that people use their computer for, two of them can't be done acceptably on this machine.
My wife was rocking an Ivy Bridge MacBook Air until just this year and it was working totally fine on all non-browser activities. The only thing that caused her to switch (which she didn't want to do) was that her browser with the 300 tabs open was crawling to a halt.
> It's just wasting time. Your user suffers due to bloat and slowness on commercial software (like websites full of tracking and heavy UI design) due to economical reasons (no benefit in optimizing user perf for the company) and not due to system software like one Drew writes.
And if system software developers believed that line of reasoning, then that would be slow as well, and if the commercial application vendors gave half as much attention and effort then those wouldn't be so bloated and awful.
> Your user suffers due to bloat and slowness on commercial software (like websites full of tracking and heavy UI design) due to economical reasons (no benefit in optimizing user perf for the company) and not due to system software like one Drew writes.
Did you not read the article? He makes a direct comparison between two window managers and the different levels of performance they provide. The system software obviously matters.
Moreover, there's huge benefits to optimizing the performance, even from a corporate perspective. There's just often not many market driven reasons to do so, and so many companies are willfully blind to this truth.
So, you assertion seems to boil down to "everyone else does it in a crappy way because they are ignorant, so you should too, to avoid 'wasting time'." It's a bizarre take.
I agree that X200 is a good choice for someone who wants a small vintage ThinkPad for Linux. They have real keyboards, and enough CPU to run Firefox and other things. If you also use larger vintage ThinkPads from that era, they can all probably share the same 90W AC adapters (though the X200 will also work on lower-Watt ones). I have a stockpile of high-res X200 units, which I've upgraded with more RAM, SSD, Coreboot, and newer WiFi cards.
One thing you'll miss for some purposes is modern GPU. For ML models, I access my GPU server over the network (which means a beefier GPU than I could get in even a non-vintage ThinkPad, and easier to upgrade) and use cloud providers. For gaming, I use a Sony console, for compartmentalization and fewer cheaters.
I spent six years with an entry-level pentium CPU (G4400) in my main (and often only) computer. I also did not use a SSD. It really did make me appreciate programs that make do with less. Moving to a 6C/12T Ryzen with a SSD was amazing, I couldn't believe how quick everything was. Even invoking thunar was appreciably faster, I had no idea I'd been using a "slow" system before.
The Pentium went through a lot; I threw everything I was interested in as a student at it. It even did some 24+ hour encoding jobs, some NN training and some Wii emulation and Wine gaming. Moving to a faster computer with a SSD only really sped up the gaming parts. 95% of my time is spent in my text editor / browser / terminal, and they only really benefit from the SSD, not from higher clock speeds / more cores. Compiling software is much faster, but I don't do it often enough with programs large enough for it to matter.
My desktop never did overheat, though. My laptop doesn't go under 50C without an external cooler.
Hardware support for video codecs is a good reason to upgrade.
Valid point. I went from using Palemoon to using FF right about when I switched computers, so nothing much changed, except I could play 1440p videos without stutter and could actually load some major websites. No role for VSCode because I use Emacs :).
Every piece of hardware was brand new at some time. Sometimes people forget that an old machine was once someone's awesome, fancy, new device. I've known too many people who couldn't articulate why they spent thousands of dollars on something new when their old machine worked perfectly well, but that's OK - most of my hardware I got for free or next to nothing because of this.
Like Drew writes, do I need to play videos that're higher resolution than 1080p on a 17" 1920x1200 display? Nope.
I buy most of my machines from the local electronics reseller, so I'm usually 5-10 years behind what's current. It works really well for me because the latest and greatest doesn't really bring my significant benefits for my use cases. That may not be the case for everybody.
But I do think there's a lot of people who buy more compute than they need. There's nothing wrong with that if that's what you're into. It's like buying a muscle car when you don't race much.
Both my laptop and desktop are roughly 10 year old. They both work as a charm. Unfortunately, it looks like planned obsolescence will get them both in Oct 2025, when MS drops support for Win10 (and they're not supported by Win11). Or I'll move to Linux.
How much faster can you make a thing by scaling it vertically on hardware? Not that much unless it’s embarrassingly parallel and you use ASICS or gpus. Ok so scale horizontal, but now you’re running a lot of machines and connections etc. it becomes very complicated, so you start introducing abstractions like k8s.
To hell with all of it. Scale DOWN, not up or across. Get rid of all the unnecessary bits. Get rid of most IO, memory management, operating systems, abstractions.
When people say, “that’s impractical, because …” I guarantee their solution to your concerns is involving more layers of shit.
Then sit out your house and drink coffee and seethe, like I do.
My desktop is around 13 years old. Upgraded a few times. Still more than enough for my linux, firefox, movies, casual photogrpah.
However, I also work with machine learning and rust and both these tasks are a bit too heavy for the desktop (i.e. I can work, but it takes much longer than on my 64GB intel-top-of-the line PC at work).
I use old Hardware because I can use whatever OS I want on it, no worries about secure boot and things like that. Right now I have a T420 with BSD and a W541 with Linux, both were bought used and cheap. Due to Nvidia, BSD has a rough time on the W541, but runs OK.
I will never by new unless I get a 100% guarantee with full refund that NetBSD runs on that hardware without the slightest issue. You can guess how well that will go over
I'm not exactly on this trend, but I noticed something going into that direction:
15 years ago, I wanted the latest everything. Now I usually want the latest minus one. Python 3.11 is out? It's finally time to use 3.10. The new phone model is ready? Let me buy the one from last year.
Because it's almost as good, but a lot more reliable.
You get all the patches, all the user report, all the ecosystem support that has accumulated since then. And this has a LOT of value.
Also I can generally pick up the previous generation used at a discount as the new stock is about to come in. This is especially good on big ticket items that people seem to crave having the newest version of. i.e phones, TVs, computers.
A couple weeks ago I realized I was still running the previous version of MacOS and I honestly don't want to put any effort into upgrading. It works fine right now. What a weird departure when I used to want to upgrade on day 1 and had to consciously wait two weeks for the big bugs to be worked out.
Apple has advantages. And everything comes with tradeoffs.
I don't use Apple's hardware and software anymore. Thats my position. I won't take those tradeoffs. Works for me TM.
Others use Apple's hardware and software. It works for them.
I own enough personal and professional hardware and boatloads of software, on owned hardware and in the cloud, that has nothing to do with Apple. Clearly there's competition.
One's own values may shift over time. Personally, I'm still as much a fan of bleeding edge[0] as when I was a young'un.
It's just personal preference and it's fine. The weird thing is making a big deal of it, IMO.
[0] As is tradition, I must mention that I do run Arch Linux, but I'm not on the 'testing' repositories... but I'm not sure I'd ever want to be that bleeding edge, even in my youth :)
I saw this happen in myself too. I think it has to do with experiencing things as a consumer first(the "ooh shinys") and then only later being able to work through how I'm actually using the tool towards a goal.
A lot of tasks that look like they need a computer in the loop, are actually a tiny layer of tech on top of a more traditional skillset. That is, you would actually be more capable of using the tech well by studying and planning it on paper and walking in the proverbial "recording studio" practiced and prepared, than by getting a device and expecting to be hands-on with it for hours and hours every day. This analogy works for music, for visual art, for writing, for programming, and many other tasks.
The cases where you can get a lot out of being hands-on tend to be artifacts of incidental incentive structures, because so much of computing is just an abstract UI to process and edit data you already have and publish the results, vs being an more intrinsic experience - it's supposed to speed you up, not make you dwell on every decision. Staying hands-on with the machine creates incentives to polish things mindlessly, because it will never say "no" to one more layer of processing, and it will always say "yes" to your requests to keep previous versions around.
Going traditional cuts out all the fluff - you have to put in ongoing effort to support a traditional workflow, so you automatically drop your level of polish to what the task calls for. It makes you think about handling information as a performance art, instead of an idealized production process. And when you bring the computer in on top of that, it focuses you around making basic adjustments and enhancements, instead of mitigating the need for a performative approach altogether.
x200 is a bit of a stretch at times, they can get quite hot and throttling depending on the tasks (especially due to modern web requiring gpu acceleration for a lot of things). x201 is already a nice change.
Yeah, I recently picked up a 2015 15" MBP with 16GB of RAM from Facebook Marketplace for pretty cheap and I am absolutely loving it. I've got a much newer and beefier Lenovo Legion for CAD work and some other heavier tasks, but as a daily driver this machine is great so far. I had a much newer MBP for a while but it sadly died a coffee-induced death; I'll probably pick up an M1 or M2 at some point since this unit can't run the latest OSX, but other than that this unit leaves me wanting for nothing.
I think that if you’re in the business of building compilers, tools, and OS level/system software you could give your users a leg up by using a low end, refurbished machine as your main.
If you can make it fast enough to be your daily driver then it will certainly be fast for the majority of your users. And you can support more users: those who can’t afford high end machines or those who don’t want to for climate/ecological reasons. There is still a ton of good hardware out there that doesn’t need to be e-waste.
I don’t know. I have an X200 and while this inspired me to hack around with it today (I ought to upgrade to a recent Coreboot) it really is just self imposed blinders with this decades hardware improvements.
And let’s face it, if someone only has $200 to spend on a “computer” that “computer” will be a Smartphone.
It is cool and I like it(old thinkpads are awesome) but the slowness would hit me every time I do something a bit more intensive(like browsing the web with hundreds tabs). Maybe it is related to the fact that now I have really fast hardware.
Also it really depends on your use case. I have a VM heavy workflow which benefits from the insane amount of RAM I have in my laptop and the number of cores.
> a bit more intensive(like browsing the web with hundreds tabs).
Maybe if you have hundreds of tabs active, but I have literally over 5,000 tabs in Firefox on one of my machines and that's 5 to 10 year old hardware with I think 8GB of RAM.
(Aside: yes, I know I have a problem, but the point is that it works)
I’m still rocking my 2015 MBP. I have no issues at all. The battery says it needs to be serviced, but the battery life is still plenty good for my uses. I have no issues with slowness. And I’m not running a minimalist WM to accommodate this like Drew is, either - I’m running Monterey, and have the UX of a full DE. I can’t upgrade to Ventura, but that’s ok.
I run VS Code on it a lot, often remote over SSH to my desktop to work on a personal raytracing project - I need the nvidia GPU to do CUDA stuff, and the 16 cores in my 3700x are pretty nice for non-GPU stuff too. I also stream games from my desktop to it using Parsec and Moonlight. I just play games casually, so no need for super low latency, and over my local network the experience is great.
I’m not sure how well it’ll be holding up at the 11 year mark though (4 years from now). But I think this habit of doing most of my development over ssh will potentially extend the life of it.
For the most part I agree with Drew. Based on all of the above, it does absolutely everything I need it to do. I can’t even remotely justify the cost of a new laptop due to this. However, if I get to the point where I can’t even run KDE comfortably, or get an equivalent desktop UX, it’s time to upgrade. I don’t prefer minimalist WMs (or Linux even really, but if anyone wants to reply, I’d prefer not to fight about using Linux as a desktop - it’s just not for me) - that’s where I draw the line and decide to get some newer hardware. Hopefully when I do, that hardware will last me another 7-10 years.
What's the display quality on these 10+ year old laptops? Looking at a low res blurry display all day can't be fun even if the performance is good enough.
There really is no reason you can’t use it shouldn’t use a 2012-2014 system in 2023. Even minor upgrades like an SSD are tremendously beneficial. And to the commenters going “try watching a 4K video hur hur!”… why? No one will argue that multimedia oriented stuff /doesn’t/ need, more often than not, beefier hardware. But let’s ignore the fact that most of the machines from this era that are laptops won’t have anything remotely approaching a 4K display. So it’s a moot point.
My personal desktop is an i9-13900k and it’s awesome.
My laptop is currently a 3rd gen i5 powered dell that works beautifully for everything I need or want on the go.
Up till literally 2 years ago my daily driver laptop was a p 2-300 with 192mb of ram, 64gb ide ssd, and a 10” 800x600 lcd. You know why? Because it did everything I needed for a portable system and the rare instances I needed more I could RDP into a stronger rig.
So, yeah, artificially limiting yourself on every rig to something old I don’t see the value of. But continuing to use, even as your primary portable, an older system is 100% fine.
I try to be as simple as possible with my setups and use the defaults wherever possible but I have always bought the highest spec I can afford when purchasing a new machine to make sure they last.
I had the perfect machine 3 machines ago until i upgraded and I've never managed to replicate that level of perfection since so I stopped trying!
I've also been using a lot of services like gitpod and GitHub's codespaces recently they make the need for newer hardware unnecessary!
I'm also increasingly multi platform so I can't commit to setting up all these things!
My next mission is to look into remote desktop offerings or just get better at the command line.
83 comments
[ 4.6 ms ] story [ 156 ms ] threadOn Ubuntu, there's a "hardware enablement kernel" package, linux-generic-hwe-22.04, that's just the kernel from the latest Ubuntu back-ported to the LTS version. I find it tends to fix most issues with newer hardware. Maybe PopOS has something similar?
Anyways, I found the MIMO in wifi 6 was better enough that I even upgraded another old laptop I have with an AX210 I found for $14.
https://community.intel.com/t5/Wireless/Unstable-AX200-conne...
Also, the latest Intel iwlwifi firmare is slowly evolving to address crummy AX200 speed issues that Windows does not suffer. On one of my laptops the AX200 is all over the place speedwise, but running wavemon shows it stabilize in the 600 range of rx and tx after awhile. Here's hoping for progress with that firmware on Linux.
https://git.kernel.org/pub/scm/linux/kernel/git/iwlwifi/linu...
I like having two identical machines because I keep them cloned continuously using my wifi network. If one dies I can just switch to the other with no data loss or interruption. I haven’t had to do this yet; the hardware seems reliable.
I like this type of machine because it’s a tablet with the power of a modest laptop. I can pop in in a backpack for travel, take it to the couch to read on, or use it like a laptop with a bluetooth keyboard. Having the keyboard separate from the screen is better for ergonomics; I much prefer these to laptops.
I have one problem, and it’s pretty annoying: I can’t seem to use video conferencing sites without problems. Zoom, jitsi, they all overload the system and my microphone stops working after a short time. I kept a Windows partition on one of the SP3s and boot into it just for Zoom meetings. If anyone has any suggestions, I’m eager to solve this.
I don't buy into this mentality. It's just wasting time. Your user suffers due to bloat and slowness on commercial software (like websites full of tracking and heavy UI design) due to economical reasons (no benefit in optimizing user perf for the company) and not due to system software like one Drew writes.
It's just kind of a LARP of "old school" and "minimalistic", which is fine. Just let's not pretend there's some practical benefit.
Hmm. Like with most things, there are benefits and drawbacks to this approach, but outright dismissal as 'LARP' is, at best, uncharitable. Are you sure you cannot come up with at least one good practical reason?
The idea that using underpowered computer makes you write some faster or better software etc. is just silly. Especially that all the really fast and well known projects in last few years tend to come from Rust people, with their super heavy compiler that has no respect for RAM and storage usage.
It's by definition always better to have a faster box than slower one. You want to see how it works on much slower boxes? Run it with a CPU quota or in a VM. People don't do it, not because they can't, or because they have too fast computers, but because they have other priorities.
I will admit that I would rather someone with more experience commented on this rather than me.
That said, I disagree on general principle.
When you are restrained ( sometimes severely restrained ) by the environment, the net result is that you use resources a lot more efficiently. I absolutely get that we are kinda spoiled now with crazy hardware that is basically a supercomputer from not so long ago and it is hard to let that convenience and power go, but that lends to the bloated software landscape we have today that manages to overuse just about every resource available.
With limited power, you are effectively forced to write better. It is not completely unlike training with weights. It is not silly. It produces real results.
In 2022 a Sandy Bridge (2012) computer is still a viable computer and viable for the foreseeable future. In 2012, a late Pentium III/early Pentium 4 was not yet an antique but already severely outdated.
Sure, maybe my computer might have screamed at me if I tried to watch AV1 or H.265 4K video, but... my computer is 10 years old and so is my monitor, so I don't have a 4K monitor. And, quite frankly, I don't see the point upgrading from 1080 to 4K.
It saves footage in a very inefficient format which is horrible to edit with, so it'll be slow and clunky on anything.
A lot of people editing iPhone footage are going to use a phone-native app like CapCut, but even with that coping with the horribly inefficient compression of h.264 and h.265 footage gets them up to egg-frying temperature very quickly.
Your argument "that's a myth" is based on a task that's so small in terms of percentage it's literally a rounding error.
I'm typing this on a 2010 computer (i7 870 @ 2.93GHz, 16GB RAM), which I also use to write software (C, Lazarus, JS, Go using VSCode, Vim, Emacs), browse web, use web apps (miro, tinkercad, asciiflow, a few others).
Inkscape works acceptably, light image retouching using Gimp works acceptably.
I used to play Starcraft II, Far Cry 3/4/5 on this machine until an upgrade nuked my Wine installation and I didn't bother to fix that.
My steam library is filled with games that work on this machine.
Out of the thousands of tasks that people use their computer for, two of them can't be done acceptably on this machine.
Components using planar transistors will last longer.
This does not help save your electrolytic capacitors, which are a primary limiter of an electronic device's longevity.
I still need to get around to replacing mine with tantalums…
And if system software developers believed that line of reasoning, then that would be slow as well, and if the commercial application vendors gave half as much attention and effort then those wouldn't be so bloated and awful.
How is it wasting time?
> Your user suffers due to bloat and slowness on commercial software [..] due to economical reasons
The article mentions KDE which i wouldn't categorize as commercial software.
...so? I don't understand why I should begrudge his preferences.
Did you not read the article? He makes a direct comparison between two window managers and the different levels of performance they provide. The system software obviously matters.
Moreover, there's huge benefits to optimizing the performance, even from a corporate perspective. There's just often not many market driven reasons to do so, and so many companies are willfully blind to this truth.
So, you assertion seems to boil down to "everyone else does it in a crappy way because they are ignorant, so you should too, to avoid 'wasting time'." It's a bizarre take.
There is room to improvement in regards to Rust's compilation speed.
That they rather spend resources in other areas deemed more critical is another matter.
However I must admit, it has improved a lot on my humble 1215B netbook from 2009.
One thing you'll miss for some purposes is modern GPU. For ML models, I access my GPU server over the network (which means a beefier GPU than I could get in even a non-vintage ThinkPad, and easier to upgrade) and use cloud providers. For gaming, I use a Sony console, for compartmentalization and fewer cheaters.
The Pentium went through a lot; I threw everything I was interested in as a student at it. It even did some 24+ hour encoding jobs, some NN training and some Wii emulation and Wine gaming. Moving to a faster computer with a SSD only really sped up the gaming parts. 95% of my time is spent in my text editor / browser / terminal, and they only really benefit from the SSD, not from higher clock speeds / more cores. Compiling software is much faster, but I don't do it often enough with programs large enough for it to matter.
My desktop never did overheat, though. My laptop doesn't go under 50C without an external cooler.
Hardware support for video codecs is a good reason to upgrade.
Like Drew writes, do I need to play videos that're higher resolution than 1080p on a 17" 1920x1200 display? Nope.
But I do think there's a lot of people who buy more compute than they need. There's nothing wrong with that if that's what you're into. It's like buying a muscle car when you don't race much.
To hell with all of it. Scale DOWN, not up or across. Get rid of all the unnecessary bits. Get rid of most IO, memory management, operating systems, abstractions.
When people say, “that’s impractical, because …” I guarantee their solution to your concerns is involving more layers of shit.
Then sit out your house and drink coffee and seethe, like I do.
However, I also work with machine learning and rust and both these tasks are a bit too heavy for the desktop (i.e. I can work, but it takes much longer than on my 64GB intel-top-of-the line PC at work).
I will never by new unless I get a 100% guarantee with full refund that NetBSD runs on that hardware without the slightest issue. You can guess how well that will go over
That is why I run on Old hardware.
15 years ago, I wanted the latest everything. Now I usually want the latest minus one. Python 3.11 is out? It's finally time to use 3.10. The new phone model is ready? Let me buy the one from last year.
Because it's almost as good, but a lot more reliable.
You get all the patches, all the user report, all the ecosystem support that has accumulated since then. And this has a LOT of value.
> don't want to put any effort into upgrading
There is so much effort from Apple to reverse this.
They want to make it harder to stay on older versions. So much so that upgrading feels like the less painful option.
Be it hardware or software they really put in so many roadblocks, its almost like an evil masterplan.
No more Apple for me.
It would be great if there were a competitor with equal hardware and software, but there isn’t that I’ve seen.
I don't use Apple's hardware and software anymore. Thats my position. I won't take those tradeoffs. Works for me TM.
Others use Apple's hardware and software. It works for them.
I own enough personal and professional hardware and boatloads of software, on owned hardware and in the cloud, that has nothing to do with Apple. Clearly there's competition.
It's just personal preference and it's fine. The weird thing is making a big deal of it, IMO.
[0] As is tradition, I must mention that I do run Arch Linux, but I'm not on the 'testing' repositories... but I'm not sure I'd ever want to be that bleeding edge, even in my youth :)
A lot of tasks that look like they need a computer in the loop, are actually a tiny layer of tech on top of a more traditional skillset. That is, you would actually be more capable of using the tech well by studying and planning it on paper and walking in the proverbial "recording studio" practiced and prepared, than by getting a device and expecting to be hands-on with it for hours and hours every day. This analogy works for music, for visual art, for writing, for programming, and many other tasks.
The cases where you can get a lot out of being hands-on tend to be artifacts of incidental incentive structures, because so much of computing is just an abstract UI to process and edit data you already have and publish the results, vs being an more intrinsic experience - it's supposed to speed you up, not make you dwell on every decision. Staying hands-on with the machine creates incentives to polish things mindlessly, because it will never say "no" to one more layer of processing, and it will always say "yes" to your requests to keep previous versions around.
Going traditional cuts out all the fluff - you have to put in ongoing effort to support a traditional workflow, so you automatically drop your level of polish to what the task calls for. It makes you think about handling information as a performance art, instead of an idealized production process. And when you bring the computer in on top of that, it focuses you around making basic adjustments and enhancements, instead of mitigating the need for a performative approach altogether.
x220/x230 feel like a sweet spot.
It runs KDE just fine though, it is from after Intel GPUs totally suck (but I won't be playing AAA games from after 2010 or so on it)
If you can make it fast enough to be your daily driver then it will certainly be fast for the majority of your users. And you can support more users: those who can’t afford high end machines or those who don’t want to for climate/ecological reasons. There is still a ton of good hardware out there that doesn’t need to be e-waste.
And let’s face it, if someone only has $200 to spend on a “computer” that “computer” will be a Smartphone.
It is cool and I like it(old thinkpads are awesome) but the slowness would hit me every time I do something a bit more intensive(like browsing the web with hundreds tabs). Maybe it is related to the fact that now I have really fast hardware.
Also it really depends on your use case. I have a VM heavy workflow which benefits from the insane amount of RAM I have in my laptop and the number of cores.
Anything later will forcibly shut down after 30 minutes if (at least a fragment of) Intel's closed & bug-ridden monitoring code is not present.
I ran me_cleaner on a few of these systems, and I do all my finances with them running OpenBSD (usually on q9550s).
Yes, this effort to run old hardware is worth it for me. Below are the bios images that I was able to produce:
https://github.com/corna/me_cleaner/issues/233
Maybe if you have hundreds of tabs active, but I have literally over 5,000 tabs in Firefox on one of my machines and that's 5 to 10 year old hardware with I think 8GB of RAM.
(Aside: yes, I know I have a problem, but the point is that it works)
I run VS Code on it a lot, often remote over SSH to my desktop to work on a personal raytracing project - I need the nvidia GPU to do CUDA stuff, and the 16 cores in my 3700x are pretty nice for non-GPU stuff too. I also stream games from my desktop to it using Parsec and Moonlight. I just play games casually, so no need for super low latency, and over my local network the experience is great.
I’m not sure how well it’ll be holding up at the 11 year mark though (4 years from now). But I think this habit of doing most of my development over ssh will potentially extend the life of it.
For the most part I agree with Drew. Based on all of the above, it does absolutely everything I need it to do. I can’t even remotely justify the cost of a new laptop due to this. However, if I get to the point where I can’t even run KDE comfortably, or get an equivalent desktop UX, it’s time to upgrade. I don’t prefer minimalist WMs (or Linux even really, but if anyone wants to reply, I’d prefer not to fight about using Linux as a desktop - it’s just not for me) - that’s where I draw the line and decide to get some newer hardware. Hopefully when I do, that hardware will last me another 7-10 years.
I7 + 16 gigs of ram and an aftermarket OWC Aura Pro X2 SSD.
I suspect it’s the Firebase emulator running in Java?
https://dortania.github.io/OpenCore-Legacy-Patcher/MODELS.ht...
They do their job, even for the GPGPU stuff I do, and will get replaced when they eventually die.
At work, the 5 year replacement cycle is also good enough.
There really is no reason you can’t use it shouldn’t use a 2012-2014 system in 2023. Even minor upgrades like an SSD are tremendously beneficial. And to the commenters going “try watching a 4K video hur hur!”… why? No one will argue that multimedia oriented stuff /doesn’t/ need, more often than not, beefier hardware. But let’s ignore the fact that most of the machines from this era that are laptops won’t have anything remotely approaching a 4K display. So it’s a moot point.
My personal desktop is an i9-13900k and it’s awesome.
My laptop is currently a 3rd gen i5 powered dell that works beautifully for everything I need or want on the go.
Up till literally 2 years ago my daily driver laptop was a p 2-300 with 192mb of ram, 64gb ide ssd, and a 10” 800x600 lcd. You know why? Because it did everything I needed for a portable system and the rare instances I needed more I could RDP into a stronger rig.
So, yeah, artificially limiting yourself on every rig to something old I don’t see the value of. But continuing to use, even as your primary portable, an older system is 100% fine.
“But they can’t edit iPhone videos!”
… smh
I had the perfect machine 3 machines ago until i upgraded and I've never managed to replicate that level of perfection since so I stopped trying!
I've also been using a lot of services like gitpod and GitHub's codespaces recently they make the need for newer hardware unnecessary!
I'm also increasingly multi platform so I can't commit to setting up all these things!
My next mission is to look into remote desktop offerings or just get better at the command line.