Yes, my plans change often. Primarily due to new use cases, newly acquired hardware, and knowledge and information gained through forums and just using my retro computers.

I keep my build plans in an Excel file and review and adjust them often. I think I'm finally at a point where I am satisfied with my three main builds, and can now focus on my fun, experimental, or edge-case builds.

Throughout all of this, my guiding principle is that the builds should serve my needs (i.e.: what do I want or need them to do).

Barley wrote on Yesterday, 15:16:

Yes, my plans change often. Primarily due to new use cases, newly acquired hardware, and knowledge and information gained through forums and just using my retro computers.

Same. I think I need to stop being distracted by every shiny new thing I see, or getting discouraged by every game I find that just won't work on my build.

Barley wrote on Yesterday, 15:16:

I keep my build plans in an Excel file and review and adjust them often. I think I'm finally at a point where I am satisfied with my three main builds, and can now focus on my fun, experimental, or edge-case builds.

Yes, similar for me. But I use a note taking app on my phone. I thought I was all done with my planning, but just again I changed my mind this morning and now want to use a PII 400Mhz rather than PII 300Mhz in my Voodoo 2 / Riva 128 build.

Barley wrote on Yesterday, 15:16:

Throughout all of this, my guiding principle is that the builds should serve my needs (i.e.: what do I want or need them to do).

Yeah, I am trying to. My main goal is to be able to play any game. Some games may work on several PCs, but picky games may have quite specific requirements for what video card, driver version, or CPU they will work with.

Yeah, it's a planning matter. It's an entirely different proposition to build "something period-accurate" vs "hardware that does X, Y and Z". Plus, being aware of where you want your hardware boundaries ahead of time helps. In my case, I know a 133MHz 486 would be great fun to have, but in practical terms it's either too slow or so fast it defeats the purpose. It's a build for when I have room to just store PCs I'd be using basically once a year, if ever. It's going to get built for the hell of it at some point, it's just in a different tier.

With this sort of mindset I always reach a point where it's good enough. Thankfully, I don't have the need to chase silicon or marginal improvements, and if I get a better idea along the way I change a build, if hardware becomes available later and makes sense I swap it after it's all "finalized" etc. In my experience if the idea is fairly concrete to begin with, everything eventually arrives at a point where there's just no need for further work aside from minor tinkering.

doesn't help when you buy disappointing things

While I have always been changing parts or even changing computers all the time, I think my Pentium 4 Win XP machine is as ultimate as it will get.
I have two pentium 1s right now and the one with 2 voodoo 2s is going to remain as it, and the second one will be for testing different hardware out.

My P3 has been the same for ages and i doubt it will change unless I find a better one. And my 2 486 machines are great the way they are . Im always up for building a new machine though

I found it fairly easy to chose for the parts on my Ultimate builds. I guess similar builds have all been done before that are the peak of whatever an OS will run on. So mine are pretty similar to what other people have done. But when it comes to the authentic experience, there are so many possible choices.

You can go by year, or GPU / CPU generation. There are different sound technologies to explore, different graphics APIs etc. Also have to consider if you want to pair a GPU with what it originally would have been installed with or do you want to let the GPU achieve it's full potential by removing any CPU bottleneck?

Come to think of it, physical constraints might be the most helpful. If you have both functionally infinite space and a dragon hoard of hardware... Build them year by year, see which gets exhausted first and possibly open a museum at the other end?

such wrote on Yesterday, 16:08:

Yeah, it's a planning matter. It's an entirely different proposition to build "something period-accurate" vs "hardware that does X, Y and Z". Plus, being aware of where you want your hardware boundaries ahead of time helps. In my case, I know a 133MHz 486 would be great fun to have, but in practical terms it's either too slow or so fast it defeats the purpose. It's a build for when I have room to just store PCs I'd be using basically once a year, if ever. It's going to get built for the hell of it at some point, it's just in a different tier.

With this sort of mindset I always reach a point where it's good enough. Thankfully, I don't have the need to chase silicon or marginal improvements, and if I get a better idea along the way I change a build, if hardware becomes available later and makes sense I swap it after it's all "finalized" etc. In my experience if the idea is fairly concrete to begin with, everything eventually arrives at a point where there's just no need for further work aside from minor tinkering.

Very true statements. My chosen boundaries are based around directx version and/or Glide versions mostly. I have made a conscious choice not to get anything prior to Pentium, as I think it gets more difficult to source parts and maintain when you go before that. Instead I have been inspired by PhilsComputerLab's 136 in 1 project.

Also I have modern re-imagining of the IBM XT in the NuXT motherboard. I would love an original XT, but they are too pricy and prone to failure.

such wrote on Yesterday, 18:39:

Come to think of it, physical constraints might be the most helpful. If you have both functionally infinite space and a dragon hoard of hardware... Build them year by year, see which gets exhausted first and possibly open a museum at the other end?

Haha. I wish for infinite space. But, sadly, no. I have limited space. It all has to fit in one tiny room that also happens to be my home office.

such wrote on Yesterday, 18:39:

possibly open a museum at the other end?

It's funny that you should say that. When I sent my proposed PC specs list to ChatGPT it did tell me that it's basically a museum. Haha.

Honestly? Not in the traditional sense, it mostly tends to be an iterative process for me, rather than really changing my mind. Often i adopt a strategy of grabbing whatever upgrades I can, when they present themselves, and work my way towards the end goal.

I only researched and assembled one "ultimate" machine, which is my Windows XP/10 dual-boot monster running Titan X and an overclocked Xeon E5-1680 V2 on an X79 motherboard. The other machines are basically defined platforms that I'm working towards "maxing out" within firmly established constraints. Like my dad's old Compaq Deskpro and my rebuilt original XP Rig with an LGA 775 Board, and a hard constraint on a pure DirecX9 GPU for superior compatibility with early Windows XP games and applications.

The only two upgrades I have planned for the later is to switch my original C2D E6700 for an E8600, so I don't need to overclock (original case doesn't have the best airflow) and eventually replacing the shitty original X-Fi XtremeGamer with an X-FI Platinum Fatal1ty. (which has a much more fun I/O box than the X-Fi Titanium Fatal1ty running on my Ultimate machine).

Eventually, once i source the parts, I also plan to build a Vista period-accurate machine with 8800 Ultra's in Tri-SLI on an nForce 790i motherboard, and that's a design plan etched in stone for me. 😀

I only wanted one machine... my first PC as it was (286-16 from 1992). This was probably around 2012 and it still isn't finished 🤣

kixs wrote on Yesterday, 19:41:

I only wanted one machine... my first PC as it was (286-16 from 1992). This was probably around 2012 and it still isn't finished 🤣

My first PC was a 286 also. Loved that thing! However the games I played on it (Duke Nukem 1 + 2, SimCity 1, Wolf3D, Prince of Persia etc), all run well, or even better, on my slowed down Pentium MMX. So, although the nostalgia was strong, I could not really justify it.

The "current project" will usually not get changed much unless something turns up faulty or horrendous mismatch (mechanical or conflicts). But the planned builds 3, 4.... 10th in line with proposed configs scribbled down, usually get a lot of changes and cancellations, as they get further up it's more like massages. Then also the schedule is not really a schedule and something might appeal more to be next than what is next, so I could just pick anything to jump to front of queue.

Though also I might go on a testing binge and forget I was gonna throw something together 🤣

Personally I only have two "completed" retro PCs for covering the 1990-2000 era and a CRT monitor which I haven't touched in a while. Only thing I did was to replace the VGA in one of them because it broke.
I think it depends on how much hardware you have laying around you. If I had multiple motherboards, CPUs, PSUs, cases, memories, VGAs, etc. I'd go crazy combining them in different computers and trying different software on them nonstop! But since I don't, I feel satisfied with the PCs I already built.
Folks who have a huge collection of PC components and a test bench to play with have a source of eternal frustration (and eternal fun) on their hands searching for "the best PC experience ever".

The Perfect Computer is Never done.

It took me 10+ years to perfect my 2 retro machines, and now I never touch them because they're perfect 😅

badmojo wrote on Yesterday, 22:08:

It took me 10+ years to perfect my 2 retro machines, and now I never touch then because they're perfect 😅

When I get a retro machine just the way I like it, I start planning backup parts. It's only going to get harder to find replacement parts so I make sure I always have backups.

I came up with a list of period correct systems from 1993 to 2000. Actually doing specs was pretty easy since I just went with the best-of options for each year.

Did build systems for 1994, 1995, and a pair for 1996. But then got distracted by other types of builds along the way, especially "all-in-one" builds and experimenting with Pentium 4 setups (imho, still a vastly underrated retro platform).

Have pretty much all the parts for the rest of the period correct builds, just haven't gotten around to building them yet.