exxos's Atari, Amiga & retro forum

terriblefire wrote: 12 Oct 2019 13:34
You would be better implementing Mike Stirling's SPI VHDL here and use the K1208 drivers.

That's good idea since these are tested and working for sure :)

You would be better implementing Mike Stirling's SPI VHDL here and use the K1208 drivers.

Brilliant, thanks! I had no idea about that card or Mike Stirling's work in general.
Yes this is by far the most sensible choice, and porting parts of his Amiga driver to Atari will surely save me a bunch of time :)

Always nice when PCBs finally arrive in the mail.
It fits like a glove, no more messy IDE cable :)

Nice indeed. Very clean!

nice

8 Bit Dreams wrote: 30 Sep 2019 13:08 That's true, China is getting worse, have bought 20x FPUs -none of them is working. some of the chips have pin1 mark on the left corner - that's crazy...
20190928_160944.jpg

I have recently bought 68060 rev.6 and 68882 from China and they're genuine working chips. I guess I'm lucky, but I also selected reputable sellers from Ebay and avoided cheapest items and obvious scams.

I think I finally have a single easy-to-use FastRAM "driver" solution that I am happy with, regardless if you're running EmuTOS or Atari TOS :)

MAPROM.PRG:
It copies the contents of your TOS Roms into FastRAM and then continues on execution from there. No restart required.
You need a 68030 with MMU, and that your ROMs are 256Kb or larger (Ie: it does not work with TOS 1.xx)
When necessary, it takes care of detecting and installing TT-RAM, the DMA buffer, and the initial MMU configuration.
Tested with Atari TOS 2.06 and EmuTOS 0.9.12. Together with NVDI.

I recommend putting MAPROM.PRG as the first item in your AUTO folder.
And then NVDI for a very snappy TOS experience :)

The binaries, and sourcecode, can be downloaded from here:
https://github.com/agranlund/tftools

While I've only tested it with my TF534, I'd expect it to work fine on other accelerators as long as the FastRAM is at 0x01000000 (TT-RAM) and the CPU is a 68030 with MMU.

Wow! This looks very promising!
What do You think, will this work on STe with Kludges plcc adapter?

Here are also the latest Atari firmwares: (No changes for the TF534, but it includes an *untested* TF530r3 build)

Source is here: https://github.com/agranlund/tf534

Wow! This looks very promising!

Thanks, yeah this is a much nicer and easier setup than those individual ram-loaders I did earlier for 2.06 and EmuTOS.

What do You think, will this work on STe with Kludges plcc adapter?

I think the main obstacle would be the Blitter, so for general use it would probably not work that great.
Technically the Blitter works fine with the TF cards as long as you only blit to-and-from ST-RAM, the issue is being able to make certain that is always 100% the case..

Perhaps it would be an idea to make the TF firmware hide the Blitter so that it cannot cause issues.
But even then, if something assume it exists just because the machine is an STE then it will probably freak out.

It would be really cool if we could find a way to be able to use it though..
Either by having to run some program to unhide it before running a blitter-enabled game, or something more elaborate like letting the TF firmware act as some kind of man-in-the-middle. It could allow the blitter hardware for ST-RAM blits but generate an exception if it tries to blit to/from FastRAM, which would then perform the blit in software instead. In theory anyway... I have no idea if it would actually work :)

On the STE, I assume you probably don't want to play DMA samples from FastRAM either..

I agree with Agranlunds, there is no reason to turn the BLiTTER off.

There are some badly written applications which have a problem with TT-RAM and try to place ACSI/Sound-DMA/the BLiTTER buffer outside ST-RAM.
There is a dedicated TOS tool - "PRGFLAGS", which informs the TOS that application should have access to e.g. the ST-RAM only.


The latest EmuTOS deals with the BLiTTER and 68030 card with its RAM (actually with ROM placed inside 68030 RAM):
https://sourceforge.net/p/emutos/mailma ... nth=201909