-
Benchmarks
25 minutes ago • 0 commentsThe CPU runs at 25.175MHz, but skips every 8th clock when drawing the active area of the screen. Each instruction is either 2,3, or 4 clocks. On average it runs about 7 million instructions per second depending on the application. It's screaming fast, but doesn't have much in the way of graphics acceleration (other than the fact it doesn't stall the CPU very much). But that's ok - there's plenty of CPU time to do some graphics help as needed.
Pi spigot
- 1000 exact digits of pi with integer arithmetic only, printed as they come out.
- 69.9s
- Other systems:
- C64: 295.57s
- Atari 800XL: 181.69s
- Apple IIgs: 108.57s
Mandelbrot
- Monochrome 640 x 480 points: 81.0 s
- 4-color 320x240 points: 20.4s
- 16-color 160x120 points: 5.2s
- Gigatron: 8,243s
- C64: 160x200 points: 190s
- Atari 800: 160x100 points: 24s
- 256-color 80x60 points: 1.3s
Game of Life
- 640 x 480 cells on a torus
- 1000 generations
- 282s
- 1 million cells per second
- C64: 2,250 cells per second
-
Holding off
2 hours ago • 0 commentsI've waited to open the project up until the last minute. I wanted to get boards in hand before going public, but I want to submit to this year's retrocomputer challenge.
I wanted to verify the timing and logic of the design in VHDL before I started with chips. It's much easier to make design changes with a few lines of code than rewiring solderless breadboards. I can load the ROM as a memory initialization file. I made the top-level VHDL code structural to represent the schematic for the board. I wanted each PLD a module with only its equations inside. Some registers are just processes, but that's ok. The Basys3 is an easy board for this since it has a VGA output and plenty of LEDs/buttons to simulate functions. It has a USB port to simulate a PS/2 keyboard or mouse.
Everything is fully operational on the FPGA. Software development can continue on the FPGA for the library and demos/benchmarks. It should perform the same as the board.
Of course, the end goal is getting a board made. So I have the KiCad main board and cartridge schematics finished. The board is 95% routed, so I should be ordering boards any day now.