Search found 109 matches

by lb3361
Today, 16:21
Forum: Support
Topic: Nonfunctional Gigatron.
Replies: 13
Views: 304

Re: Nonfunctional Gigatron.

I wanted to test my latest toy by measuring the lag between the two clock signals in the Gigatron. I was expecting 5ns but measured close to 20ns lag. This means that when the registers latch their data, the latch enable signal is still okay despite the fact that the instruction decoder has been wor...
by lb3361
Yesterday, 00:10
Forum: Hardware and software hacking
Topic: Extending the CTRL code space
Replies: 7
Views: 411

Re: Extending the CTRL code space

The " RAM & IO Expansion v7 Dual Drive " is a hardware proof of concept for this proposal. https://github.com/lb3361/gigatron-lb/tree/v7/extension-retro This board recognizes the special ctrl code and, instead of modifying the ctrl bits, exposes the special code on a new 13 pins header...
by lb3361
16 Nov 2021, 14:39
Forum: Hardware and software hacking
Topic: 6Bit audio:
Replies: 26
Views: 8970

Re: 6Bit audio:

Cool!
by lb3361
10 Nov 2021, 20:04
Forum: Hardware and software hacking
Topic: Implementing a faster multiplication routine
Replies: 20
Views: 3369

Re: Implementing a faster multiplication routine

And here's the pixels where I am slower. I assume this is simply that because I'm more accurate (8 bits of fraction), the bulb optimisation is kicking in later on my version. If that were the case, we would only see the slower points on one side of the bulb, not both sides. My bets are on the highe...
by lb3361
10 Nov 2021, 02:09
Forum: Support
Topic: Controller issues
Replies: 3
Views: 245

Re: Controller issues

walter wrote: 05 Nov 2021, 08:23 You can try https://www.aliexpress.com/store/group/ ... 25411.html
I think these will work.
I got one of those and it works. Mine is a Type C controller with a priority encoder instead of a shift register. Not the best kind, but good enough.
by lb3361
08 Nov 2021, 17:09
Forum: Hardware and software hacking
Topic: Implementing a faster multiplication routine
Replies: 20
Views: 3369

Re: Implementing a faster multiplication routine

The invisible bug would make a difference. There is code in CalcSet that bails out when one is inside the first lobe (the buggy one) or the cardioid. If that code fails, you're good for 64 iterations of the Mandelbrot function on each pixel... Meanwhile I looked into the profiling results for the un...
by lb3361
08 Nov 2021, 00:46
Forum: Hardware and software hacking
Topic: Implementing a faster multiplication routine
Replies: 20
Views: 3369

Re: Implementing a faster multiplication routine

Turns out I had a version of GLCC's gtsim with profiling support. I added a HALT after the first screen of Mandelbrot_v1. At that point, the total number of cycles spent executing vCPU instructions in this file is 1451210751. The total number of cycles spent in MulShift7 is 1300670415, that is 89% o...
by lb3361
03 Nov 2021, 22:52
Forum: Hardware and software hacking
Topic: Implementing a faster multiplication routine
Replies: 20
Views: 3369

Re: Implementing a faster multiplication routine

Marcel's code does not compute all the bits of A * B. It seems to assume that the absolute value of A has at most 9 bits (because the initial value of variable 'bit' is 0x200) and that the absolute value of B has at most 13 bits. Then it only computes the 15 most significant bits, that is (A * B) >>...
by lb3361
29 Oct 2021, 22:16
Forum: Hardware and software hacking
Topic: SDCard Browser for the RAM & IO expansion SPI port.
Replies: 2
Views: 233

SDCard Browser for the RAM & IO expansion SPI port.

Announcing the first release of a browser targeting SDCards attached to the SPI ports of a RAM & IO Expansion Card. Any card compatible with Marcel's original design should work. Go to https://github.com/lb3361/gigatron-os/blob/master/sys1 for code and instructions. This can be compared to the p...
by lb3361
28 Oct 2021, 14:28
Forum: Hardware and software hacking
Topic: 10MHz, 12.5MHz and Beyond!
Replies: 42
Views: 24147

Re: 10MHz, 12.5MHz and Beyond!

For 50 Mhz, bit-banging would work fine. That means if you keep the resolution, you can have 7 instructions between each pixel (unless you want 56.25 - 62.50 Mhz and get 8 or 9). With the LUT-based CU, you could even add maybe 2 more index registers to make bit-banging more efficient. The line repe...