Demozoo.org is a website that is a library of not only old school ’90’s era demo competition submission – but even all the recent ones as well. They have lists of current competitions and news too. An extra feature is many demos have youtube videos of the runs so you don’t have to download the binaries and run them locally.
8086 emulator is a fun Intel 8086 emulator / vm on github. It can run most of the 8086 instruction set and provides an interactive interpreter and debugger that allows you to run programs line by line. [hackaday]
The Video Game History Foundation (VHGF)’s digital archive of video game research has launched its first round of online early access to it’s library. It’s free to access anywhere in the world and intended for “anyone who wants to study video game history.”
What do they have?
There are some caveats: There are no playable games in the archive due to copyright restrictions and VGHF said it “cannot give express permission” for users to reproduce materials in the library unless explicitly stated otherwise though researchers may be exempt under fair use cases.
We’ve all seen old standup arcade games that used guns – like one of the iterations of House of the Dead or VirtuaCop. At home, who didn’t play Duck Hunt on the NES?
Unfortunately, if you try those games today – they don’t work. The mechanisms they relied on only worked on old tube-type CRTs. People have tried to make alternative mechanisms – but they relied on having to attach messy sensor or light bars around the screen. Using mechanisms like sensor bars also means you must stay in the exact same spot or re-calibrate – something that is very annoying during a long gaming session. What to do?
Andrew Sinden decided to tackle the project and came up with a brilliantly simple solution. Simply render/detect the square around the border of the game being played determine your location from that rectangle. That input is converted to mouse input/direction and voila. It works on any size display, doesn’t need recalibration, and allows for multiple players. Andrew Sinden shows off how he developed it here:
He started an IndieGoGo project, and ended up raising a whopping $4.6 million of the desired $300,000 goal.
His startup project is now a full-fledged product and he sells them over at SindenShop. They make 2 guns: with and without recoil. They run about $170/$115 respectively and come in blue, red, black, and grey if you want different guns for different numbers of players. They also now make arcade-style foot pedals for games that use those.
Acerola has a bunch of great graphics videos. In this one, he talks about why PS1 graphics looked the way it did.
I learned that PS1 actually had realtime camera distance tessellation – something that wasn’t available to desktop GPUs until the introduction of tessellation shaders.
I recently acquired not one, but 2, 8″ floppy drives. Behold a Data System’s Design DSD 440:
To get this all working in my experiments, I used the following guides
Parts and equipment
Cathode Ray Dude did a great rundown on the IBM 5140 Convertible. It was the first PC computer my dad got after the Radio Shack TRS-80 Model III and the first one that had a floppy drive. We even had the serial/parallel port and the thermal printer.
But probably the most amazing thing is that this 8088 XT, floppy based computer actually had a method to sleep and wake right where you left off. As far as we know, it was the very first instance of the technology.
Valve’s Year In Review for Steam revealed interesting statistics about its player base.
Only 15 percent of Steam players spent their time on games released in 2024. 47 percent of players devoted their time to games from the past 1-7 years (called “recent favorites”), and 37 percent played titles from eight or more years ago (“classics”).
While that seems low, it’s actually in line with historical trends. In fact, 2023 was particularly bad with players only spending 9% of time on games released in 2023. In 2022, players spent time on 17% of games released in 2022.
The ratio for ‘recent favorites’ for 2022 and 2023 were 19 percent and 52 percent, respectively, and ‘classics’ for each year were 62 percent and 38 percent.
| Year | Time spent on games released that year | Recent Favorites (1-7 years old) | Classics (8+ years old) |
| 2024 | 15% | 47% | 37% |
| 2023 | 9% | 52% | 38% |
| 2022 | 17% | 19% | 62% |
Articles:
The Sony Trinitron KX-45ED1, aka the PVM-4300, is thought to be the largest CRT TV ever sold to consumers. It has a 43-inch visible diagonal on its 45-inch tube and weighs in at almost 440 lbs. The stand alone is over 170lbs. At the time, it cost $40,000 USD in 1989 (or about $100K today, adjusted for inflation)
Long since thought gone, Shank Mods managed to save an extremely rare 43-inch Sony Trinitron KX-45ED1 from an untimely ending. It was being kept on the second floor of an Osaka noodle shop called Chikuma Soba – a building due for demolition in just a few weeks.
It was moved from the soba shop, crated up, and shipped to the US. While it worked well – it did need servicing. The alignment was off, had some tube cataracts, and the dynamic convergence amplifier circuit had failed. They worked on them all and have a very nice display.
The video describes the incredible journey and is definitely worth a watch
Articles:
Andreas from Insomniac Games made a Amiga 500 demo in 2019 as part of this work with The Black Lotus demo group. He presented not only the Eon Amiga 500 demo, but tons of great technical information about the 4 years it took to develop it.
Old demo scene programmers hold amazing amounts of wisdom. When solving the core pieces of logic, I found this is true (but when doing larger, complete system development, these don’t work)
Work backwards from desired outcome to discover your constraints. Don’t just brute force. Instead, ask, what must be in place for us to get the peak performance from the key component we’re dependent on (render, disk load, etc). Then work from that constraint.
Do everything you can at compile time, not run time. Pre-compute tons of things – even the build-up of the data structures in memory. Just run it and then save and reload that blob automatically.
Over-generalizing early is a trap many devs fall into. Solve the problem in front of you. Trust that you can delete the code and do something else if it sucks. It’s cheaper and faster than trying to anticipate things ahead of time. Do the simplest thing that will work and if it sucks come back and delete it.
If you end up with a small runtime table/code that doesn’t require runtime checks because you can prove it can’t go wrong, you’re doing something right.
When developing, the actual Amiga is super slow and limited. They took an Amiga emulator and hacked it up so they could debug on it instead. Using calltraps to trigger the emulator, they added memory protection, fast forward, trigger debug, loading symbols, cycle accurate profiling, single step, high-resolution timers, etc. Also allows perfect input playback.
Modern threading and consumer/producer components (disk loading, data transfer, decompressors, etc) often just throw things in buffers and YOLO. There’s no clear backpressure to show you where you’re wasting time/space. Running on this kind of hardware/simulator shows you how much time the design is wasting by poorly and inefficiently designed algorithms/constraints.
Presented at Handmade Cities event in Seattle at: https://handmadecities.com/