This is one of the demos that were used by Silicon Graphics and Nintendo to show the graphics features of their upcoming game console (Project Reality / Ultra 64 / Nintendo 64) during CES in 1994.
SGI used an Onyx supercomputer to run the demo. I recorded it in 1024×768 (true-color) on a much less powerful SGI O2 workstation (released in 1996). O2 is based on a graphics architecture similar to what was actually used in Nintendo 64. It was a perfect fit for an inexpensive highly-integrated computer as well as a game console.
I would rather use a different SGI computer (Octane2 was my first choice), but O2 was the only machine that was compatible with my cheap VGA-to-HDMI converter.
We prepared a small room with old computers as a part of a bigger event at work. After careful consideration, we selected Atari 800XL and Commodore Amiga 500 for the contest purposes. Atari was running Space Invaders and each visitor played with the highest difficulty (three lives, level 12). For Amiga, we chose Pinball Dreams and the goal was to get the highest score using just a single ball.
To make the room more appealing, we brought several old computer magazines and 80s props. Curious visitors could also experience ZX Spectrum+ with the Who Dares Wins game. The feedback on our room was very positive and people were talking a lot about their first computers there. However, the biggest surprise for me is that the joystick survived all the players. Some of them gave it a hard time.
Working with old Amiga computers with their original color monitors is always a painful experience. Especially here in Europe where these screens run at 50Hz. Fortunately, my Amiga 2000 is equipped with a Multivision 2000 scan-doubler card that not only doubles H-sync from 15.75kHz to 31.5kHz (to make the output compatible with VGA monitors) but it also allows a user to increase refresh rates up to 100Hz.
I wanted to show this computer on Bytefest (a vintage computer show) and I decided to show it with a CRT (LCDs always ruin the experience). That created an ideal situation to check how this “software controllable refresh rates” feature works. The manufacturer of the card (3-State) bundled a diskette with a simple program that allows you to find the optimum refresh rate using multiple sliders and the way it works is compatible even with old simple VGA screens from the early 1990s.
The picture frequency is increased when a user decreases the number of video lines. Speed of drawing of each line is still the same. The computer just generates fewer lines, which means that a single frame takes less time and the next one can start sooner.
I switched Workbench into the interlaced mode (512 lines instead of 256) and then used the bundled Sync Master tool to decrease the number of lines close to the original non-interlaced resolution. As a result, I got nice and steady 80Hz on a standard VGA screen, which allowed me to work with the computer for hours without eye strain. This is a perfect setup for office/productivity work. However, be prepared that this tool doesn’t work well with applications that open their own screens a display outside Workbench (you need to revert it to default before starting such application).
Virtual Boy was the coolest thing I saw this year during the Bytefest vintage computer show. I knew that this was a commercial failure and expected the worst. However, it was fun to play some of the games and I quite liked the red LED displays. The 3D effect was clearly visible and I didn’t experience headaches from the displays even after nearly an hour of playing.
The only drawback was the mechanical stand that required me to lean forward too much and my (already damaged enough) cervical spine suffered from it. Anyway, I’m happy that I had the opportunity to finally experience this device. I wasn’t expecting to ever see it in our small country.
These were the computers we brought to Bytefest – a Czech vintage computer show. David and I decided not to bring more than two desktop systems. Amiga 2000 was an obvious choice – we fixed it not a long time ago and I played a lot with it recently. The other computer was SGI Indy with the original set of peripherals including the Indycam camera. There are not many vintage UNIX computers to see on vintage computer shows in this country. Thus, it is my duty to bring at least one every year.
The Aritma Minigraf plotter sitting on top of the Indy was connected using one of the Indy’s serial ports though a special ARM-based module that David built. The module contained the control software that allowed it to draw faster and with better precision than the plotter was originally designed for. From time to time, there were couple of people standing in front of the plotter, being hypnotized by the smooth movement of the pen. The Indy itself was communicating with the module as a serial terminal with the ability to send HPGL files that needed to be drawn.
I’d never played that much with Indy before (aside creating the OpenGL 1.0 version of our 3D graphics benchmark) and this was a nice experience. The graphics card in our Indy is able to display no more than 256 colors (or 16 colors for double-buffered 3D), but it’s pretty fast and allows you to have a different 256-color palette for an active window and the rest of the system. Therefore, the color flickering effects are minimized in comparison with PCs set to 256-color modes. I was surprised by the visual quality of the composite input from Nintendo 64 in 256 colors.
Commodore Amiga 2000 was configured to show the capability of this platform during the late 80s (thus, Workbench 1.3 and Kickstart 1.3 only). It didn’t have any accelerator board and the only expansions were a simple hard disk controller, 2-MB fast RAM card and A2088XT PC emulator (with an 8088 and 512kB of RAM). During the show, I also added an ISA card with a serial port (for Microsoft InterLink purposes) and a VGA adapter.
The other devices that we showed were: Apple PowerBook 100 (this year with a working hard drive and full of software), Digital DECpc 325SLC (because a 386 with color LCD is cool) and HP OmniBook 900 (just a service laptop to convert the Wi-Fi Internet into a cable form for the Indy).
Digital (DEC) always made interesting computers and their PC laptops were no exception. DECpc 325SLC was a laptop with the first generation of color passive-matrix displays. The picture quality (aside the ability to display colors) was on par with monochrome passive-matrix screens with all the drawbacks they had. On the other side, the computer was equipped with a 25-MHz Intel 386SL and SVGA graphics for just $2.100 – the price where other laptops from most other brands have just monochrome screens and VGA graphics.
The 386SL is the first CPU specifically designed by Intel for use in laptops. It integrates almost a whole PC into two chips, with the main chip containing (among other things) a 386SX core (with 16-bit data bus) and 64kB of cache (16-bit as well). Intel built the platform to support then new power management functions like the sleep mode (“suspend to RAM”). The performance of this CPU is halfway between 386SX and 386DX.
Graphics chip was not integrated in the two Intel chips. Digital decided to use a chip from Western Digital with 512kB of video memory and the support for 256 colors with a resolution up to 800×600. The chip was attached using the ISA bus and had no acceleration. On the other side, the support for VESA VBE 1.2 and ~3MB/s throughput to video memory made it a good mainstream solution among ISA based cards of the time.
There was also a detachable trackball module available for the computer. It’s not hot-plug and you need to reboot the system, but it works surprisingly well even after decades. Note the mouse icons on the arrow keys and Z/X – these are for the mouse emulation. The laptop with no trackball or mouse attached transparently emulates PS/2 mouse on these keys and the result is way more usable than the Windows feature called “MouseKeys”.
I’m still preparing my computers for the upcoming Bytefest. Amiga 2000 with a A2088XT PC emulator card is the second computer I want to have there. Unlike others, I want to show the computer with Kickstart 1.3 / Workbench 1.3, original upgrades and late-80s programs.
I often see only heavily expanded Amiga systems with latest versions of Workbench, PowerPC accelerators and PCI cards. However, these machines don’t say much about this platform when it was really used by professionals.
Bytefest 2019 is coming and I have only two weeks to prepare all the machines I want to take with me. I want to the show this Indy with a Nintendo 64 game console because Indys were often used for N64 game development (after all, N64 hardware was designed by SGI). It is nice to see that Indy’s VINO interface supports progressive scanning (used by game consoles and old 8bit computers) on its composite/S-Video inputs – unlike newer SGI O2 and SGI Visual Workstation 320. Anyway, the main planned part is to connect a vintage Czechoslovakia plotter (Aritma Minigraf) using our custom interface (modified to use a serial port) and plot processed images of visitors taken using the Indy’s bundled webcam.
I’m surprised that serial ports on Indy support speeds only up to 38.4 kb/s. Pretty slow for a computer introduced in 1993. Maybe that’s one of the reasons why the serial port speed was not even mentioned in the user guide. They just didn’t care.
It took me two years to find some time to disassemble the machine and then another year to start fixing it. Repairing the PSU with the blown EMI suppression capacitor was an easy task and that’s what we started with. Worse part was the instability – machine often crashed to a guru meditation error during certain tasks and sometimes a restart resulted in a “color screen” error.
David created and programmed a small device with eight differential inputs for voltage measurement (and logging) so we could check if the electrolytic capacitors did their job properly. To our surprise, they were fine. We also tried different programs to test memory and other hardware but the computer successfully passed all the tests.
Based on the Guru error codes, we found the root cause in the faulty EPROM chip (Kickstart 3.1). It started to lose data and reading of certain regions of the chip was sometimes affected. Thus, CPU executed faulty code (like a word or long word access on an odd address boundary).
The system is now running with Kickstart 1.3 and Workbench 1.3 and all original upgrades are inside except one. The Commodore A2630 accelerator card (25MHz 68030+98882, 2MB 32-bit RAM) crashes with dark blue and green screens and I’m afraid that there is a mechanical issue.
Next steps: Fix A2630 and reprogram the EPROM chip.
I was asked by my manager to download data from multiple boxes full of floppy disks he used in the early 90s. I’m used to people at work asking me for help with old UNIX systems, but reading 5.25-inch floppies is here for the first time. He used to be a musician when he lived in Israel and used his 386 PC as a sequencer with an E-mu Proteus/1 external wavetable synthesizer.
I picked my Vienna 286 computer to read the floppies. It’s quite a high spec machine with an 8MHz CPU, math coprocessor, Hercules-compatible Graphics and 1.5meg ISA RAM card… and it’s my only computer with a 5.25-inch drive. If you ask why I removed the CRT before I started copying the files, ugly mold smell goes from it every time it’s turned on and I hate it. I rather configured the machine to work in a headless mode (straight boot into Microsoft InterLink Server) and accessed the drives from a laptop over a null-modem cable.
