Optimize text output file

[benjamin051000]

Currently the output files are on the order of 50-100MB in size. This is hugely space-inefficient. Consider a way to optimize this!

[benjamin051000]

I believe most of the records in the file are duplicates, especially since in our tests we are painting the same color (black) to the screen 90% of the time or so.

A solution to drastically reduce filesize and increase performance would be to only record signal events. This would honestly be pretty easy to do.

[benjamin051000]

Lines that have the potential to be reduced:

• 1,989,740 lines of "ns 1 1 000 000 00" (53,246,087 chars on those lines). (This is ~75% of the entire file!)
• ~300k lines of vsync/hsync (drop in the bucket, but can be optimized)

[benjamin051000]

The log would work like this: Each entry is when a signal changed.

In the simulator, only update the values when we reach the timestamp.

So what we can do is something like this:

1. Read two lines to determine the first time interval
2. Use the signals of the first one until the interval expires
3. Advance both lines to the next interval/set of signals

This requires us to know the time interval. This may need to be supplied by a CLI flag.

[benjamin051000]

Our output file is 98.4758% smaller! Let's see how difficult it is to integrate into the script

-rw-r--r--. 1 benjamin benjamin  1088541 Mar 30 13:52 newout.txt
-rw-r--r--. 1 benjamin benjamin 71415293 Mar 29 23:04 vga_output.txt

[benjamin051000]

Ideally we could just have the file output two times to define a range... but I really don't care to learn VHDL text IO functions.

[benjamin051000]

This is super close to being finished. Someone should pick this up and finish the algorithm. As displayed above, it will result in way smaller log files, and likely will result in much faster image displaying. This is being worked on on the optimize-log branch