What is the current status of the RAM board? I saw the news posting about it where you mentioned that you made some progress on it and you were going to update the Gerber files. Has that been done? In other words, is 1.5.0 ready for use? If so, I would like to build one of those cards.
I have adapted the VIBE editor to work with Tali Forth, but unfortunately, I cannot get it to fit in the 4K of Ram that is currently available to me on the Planck. I’m looking forward to the RAM card, so that I will have more RAM available for Forth code.
Thanks for everything you are doing on this project.
Very best wishes, Ben
I have not yet taken the time to order and build the 1.5 PCB version, but the gerbers on Banked RAM board do include the fixes I mention in this post
If you do not mind loosing a bit of performance, you can also modify the code to keep taliforth code in the ROM area (replace all
.segment "DATA" with
.segment "CODE") Taliforth will then run with the ROM wait states, but most of the RAM will be free.
Oh this is great news. I’m not worried about performance at all so running from ROM will work just fine.
As far as saving RAM, I did realize that I could disable inclusion of the assembler/disassembler code by commenting out the assembler line in main.s. This frees up another 4-5K but then you lose the ability to disassemble words, etc.
If others are interested in this approach, just put a semi-colon before
TALI_OPTIONAL_ASSEMBLER=1 at line 4 of main.s and rebuild the ROM.
I will try out running from ROM and will report back. Also, getting VIBE to work required updating the source to incorporate some of SamCoVT’s more recent patches to Tali. I will do a pull request soon so that those patches can be applied to the Planck version of Tali.
Best wishes, Ben
I’ve successfully recompiled and run the system from ROM rather than from RAM.
I did some tests and the system seems notably slower with Forth in ROM rather than RAM.
5 tmultimandel takes approximately 3 minutes and 20 seconds when running from ROM. If run from RAM, it takes 2 minutes and 6 seconds. I ran each three times and averaged. In this case, running from ROM is about 60% as fast as running from RAM. Subjectively, using the full-screen VIBE editor also seemed very slow, when run with Forth in ROM.
Of course, when running from ROM,
unused u. reports 21500 bytes free, which is great. Unfortunately, while running from RAM, I only have 2188 bytes free. Getting enough free RAM requires using
drop-mandel and/or not including assembler/disassembler support.
I think it’s probably time to order the banked RAM card.
Yes, two thirds of the speed is about what I would expect with the wait states jumper set to 1