Porting Matt Heffernan’s Mandlebrot Benchmark

I’ve ported the Matt’s Mandelbrot code to the homebrew cpu, this is a discussion piece on the process and some of the challenges.

Matt Heffernan’s Links

0:00 Introduction
0:58 Run it!
1:17 Port process - Original Code
4:42 My process
9:55 Fixed point Multiply
13:12 My Implementation
13:45 Sign!

[benjaminmiller]

I'd love to see a comparison of the optimized version vs the un-optimized.

[rauljvila]

I like these deeper explanations with a more casual tone, no need for flashy graphics :).
Thanks!

[rrjmdPA]

- you stopped too soon. What do you use to edit, produce and upload??

[jaguarsrevenge]

... "make a baby Turing cry" LOL
Cheers James

[crestdazoltral]

Have you considered using video RAM chips as main memory to fetch instructions through their read-only port to reduce the number of stall conditions in your pipeline to raise the IPC if not what stopped you? Would you consider them "cheating"?

[colejohnson]

I've had an idea that I would love to hear your opinion of. If you had a second clock running at twice the system clock speed, and multiplexed the program counter clock between the normal clock and the doubled clock, selected on your argument register load line, you might be able to increase your instructions per clock cycle.