Page 2 of 3

Re: BLITTER RE-CREATION THOUGHTS

Posted: Tue Apr 03, 2018 9:41 pm
by exxos
Stimpy wrote:
Tue Apr 03, 2018 9:35 pm
Might be best to use the latest tools and latest series of cyclone, 4 or 5. Might cost a bit more but will give you more room to add things and the speed will be greater as the routing won't be so complex and they are made on a smaller die process. You also get nice things like DDR controllers, larger block ram etc. Also don't be put off by BGA as tqfp is dead for FPGAs, same goes for 5v logic levels. 1.0mm BGA is pretty easy to route and as you don't need much IO you can just use the outer connections and route straight out. You'll need decent decoupling as the edge rate on the IO drivers can be very fast so 0402 caps good idea.
Some good thoughts thanks.

I think myself and rodolphe did look at BGA a couple years ago.. but we had issues routing anything with them even on multilayer boards. So we kinda gave up and just stuck with TQFN.. Though like you say, with a lot of IO we would only need the outer pins.. so worth a look..

One problem I envision is TQFN is pretty large, and I have go to use IO converters as well yet. I want to try and keep it as small as possible overall.

Re: BLITTER RE-CREATION THOUGHTS

Posted: Tue Apr 03, 2018 9:52 pm
by Stimpy
I think most PCB places will do 0.125mm track and gap no problem so that's not so bad, and yes it will probably need to be 6 layer or even 8. But if you increase layers , the pcb routing becomes so much easier you can keep the board small and that will offset the cost of more layers!

I did put the 68000 in a cyclone just to see what fits in what devices. As you found out, takes about 10 minutes to import and build.

Re: BLITTER RE-CREATION THOUGHTS

Posted: Wed Apr 04, 2018 10:53 am
by Stimpy
Its a slippery slope too. You start to think of putting other cores in, improving it then wonder what you need the rest of the Atari ST for! I guess what needs to be retained is things that make the Atari ST an Atari ST. Hats a goid question in itself which i guess is a matter of opinion but there must be some common themes, YM chip, compatibility, floppy drive, MIDI, the case and keyboard. Would anyone care if the 68000 was recreated in Intel silicon?

Re: BLITTER RE-CREATION THOUGHTS

Posted: Wed Apr 04, 2018 12:04 pm
by exxos
Yes thats my concern as well.. at what point does it re-create the MiST...

I was talking to Rodolphe just yesterday about should we move to a FPGA CPU or not.. CPU will be faster, but if it isn't 100% cycle accurate then its not going to be 100% compatible. My goal is 100% backwards compatibility and options to upgrade.

If the blitter can run code, it shouldn't be a problem, timing isn't a issue as other chips are not running faster or slower with the blitter...

In terms of MMU, Troed has been working on testing the OC mods.. where we can get 200% ST-RAM speed with 16mhz CPU. Though we really need a new MMU which can run SRAM at 32mhz or higher.. So will need to try the suska core and try and find some HDL experts to tweak the code if needs be. I want to re-create the original MMU 100% as not to break anything, then have the turbo option to run at full speed, whatever speed that is, maybe 50MHz.

Re: BLITTER RE-CREATION THOUGHTS

Posted: Wed Apr 04, 2018 1:14 pm
by Stimpy
If you want SRAM then the MMU becomes a lot simpler. I used SRAM in my FPGA based Atari ST. Suska code is cycle accurate, it runs demos with overscan tricks etc. A small board with latest FPGA, space for SRAM and level shifters for about 60 maybe 80 IO could be used in all sorts of places. Or just as one board removing shifter, glue, MMU, CPU and blitter and tying into the old bus.

Re: BLITTER RE-CREATION THOUGHTS

Posted: Wed Apr 04, 2018 2:12 pm
by exxos
Theres loads of mods to be done. Like with high bandwidth to the shifter, can get higher resolutions and colours. A few of us talked about it a few months back. Though without higher RAM speed its not really possible. Though if things go to plan, which I doubt, faster MMU with SRAM with 50mhz CPU.. Then more registers can be added to the shifter for stuff.. Basically all the mods which would make the ST really cool.. if it can be done without breaking compatibility.

Re: BLITTER RE-CREATION THOUGHTS

Posted: Mon Apr 16, 2018 6:25 pm
by exxos
In light of been able to use the Suska blitter code I think trying to recreate the blitter on a gate level is probably a lot more time-consuming than just using the Suska code. Overall I assume the code is compatible but once the new PCB is done, I can program it with the blitter code and then we can test a lot of things out on it to make sure it is working perfectly.

The only slight issue is the PCBs obviously going to be a larger than the original chip, in one respect, needs I/O translators will take up a lot of space. So may not be so easy to use this board to update all variations motherboards. Though I am thinking more along the lines of my STF remake project which will be able to use this board a lot better.

Re: BLITTER RE-CREATION THOUGHTS

Posted: Thu Apr 19, 2018 10:50 am
by Smonson
You couldn't get it to work on the Cyclone-II? The EP2C5s I used for the shifter are only $25 each, and they should have way more LEs and pins than you need, and they're not BGA. Although they are fairly large.

Re: BLITTER RE-CREATION THOUGHTS

Posted: Thu Apr 19, 2018 11:23 am
by exxos
Smonson wrote:
Thu Apr 19, 2018 10:50 am
You couldn't get it to work on the Cyclone-II? The EP2C5s I used for the shifter are only $25 each, and they should have way more LEs and pins than you need, and they're not BGA. Although they are fairly large.
I went for the cheapest 3.3V one which would work. No use going for a larger more expensive device.

Re: BLITTER RE-CREATION THOUGHTS

Posted: Thu Apr 19, 2018 1:38 pm
by Cyprian
BLiTTER in FPGA? Really cool idea, I second that.

Would be cool to add one feature - hardware C2P conversion.
Now, a standard C2P - chunky160x100 to 4-bitplane 320x200 can take up to 480 000 cycles. With 160 000 cycles per frame it is 3 frames. With the BLiTTER speed it could take advantage of copy speed - 40KB per frame (In this case C2P frame has 32KB)