CURRENT PROTOTYPE STATUS (SEC 64MHz 68000)

[exxos]

64MHz! Sweeeeeeeeet!

IMG_6178.JPG (390.17 KiB) Viewed 4429 times

1.jpg (225.59 KiB) Viewed 4429 times

GB6 is still running in ST-RAM, So be interesting to see what speed gains are when its running in alt-ram. Probably shoot up a bit more

[exxos]

Can't believe it's been a year since the last update

I'm going to look into getting a batch of the SEC boosters made up as I simply will never get time to build them myself.

The possible options will probable be that the booster itself will be a simple plug in on the H5 platform. The H4 will likely be fine. STFM will need all the mandatory fixes done and track editing.. As such I'm not going to officially support original machines. Now the STE is a candidate now because if the STE relocator adapter board.

Now before the haters start the usual moaning about alt-ram.. @Badwolf is working on a 68k SDRAM board which should run at 16mhz speeds. It opens the door to the lowest cost option for altram. With a new STE adapter it should be simple to use the SEC booster alongside @Badwolf's SDRAM board. The H4/H5 platform should be a simple plug-in.

The H5 platform can make use of 3.3v SRAM which will be faster still but will likely be more expensive.

The H4/H5/STE could use faster still 5v SRAM but cost ramps up even more,

At the end of they day, you're going to get what you pay for with the alt-ram. But I can't get production runs of SRAM boards made as the ram is too expensive and there is never enough demand anyway. It's why I never produced any ram board as it would cost a fortune.

So the only board I may well produce is @Badwolf's SDRAM board assuming it doesn't end up crazy expensive to produce, of course parts shortages are not helping.

As for SRAM boards, I may well only stock the various PCBs. Then people who want the fastest, buy the SRAM themselves and solder it on.

Of course if @Badwolf can get his board to cope with a 50mhz CPU then it would raise the question if SRAM boards are really worth it. The SEC booster can generally push up to 64MHz and I've even had it up to 75MHz. It would need 20ns SRAM for 64MHz. I think I saw 12ns RAM at some point, but again it's the cost of the stuff which is the problem.

Mouser is about £30 per 2MB chip. So would need 4 chips for 8MB... 5 chips for 10MB.. So in RAM alone it's going to cost £150 for 5V stuff! 3.3V is about half the price. So H5 users with its 3.3V bus built on is a lot more likely to have a production run board.

Looks like these are best value for 3.3v. Would need 2 chips for 8MB..
https://www.mouser.co.uk/ProductDetail/ ... Cg1w%3D%3D

[PhilC]

Good news, that SEC board worked great on my H4. I'll have to dig it out and put it on my H5

[andresv]

I definitely want a board like this. I'm looking for a CPU accelerator with ALT RAM, so whenever there are some available i definitely want one.
Thanks and keep the awesome job you and the rest of the Atari community are doing!

[mrbombermillzy]

@exxos I'm guessing that this is a generic booster. i.e. can be used anywhere a 68k chip sits; Amiga, Mac, ST, washing machine...

[exxos]

@exxos I'm guessing that this is a generic booster. i.e. can be used anywhere a 68k chip sits; Amiga, Mac, ST, washing machine...

Not sure.. It's geared up to work with the ST's lack of proper DTACK control. It may not work out of the box on a STE even. I don't emulate the proper 68K timings either in order to gain more speed. So another chip set could cause problems as its timings will not be as expected.

[mrbombermillzy]

Fair enough.

I presume then, that a MST with IMP chipset (and mandatory fixes to try to help) is probably WORSE signal/bus noise wise, than a more standard STFM, so out of the question?

[exxos]

I presume then, that a MST with IMP chipset (and mandatory fixes to try to help) is probably WORSE signal/bus noise wise, than a more standard STFM, so out of the question?

MST are worse than STFMs yes. If the fixes are done then there is no reason why it wouldn't work. But with so many random faults with the things.. It's "suck it and see" really. I've had the SEC booster running on a fixed STFM as its how I did testing before the H4/H5 was born. It's why I say I'm not officially supporting original machines as the debugging never ends. It's why I gave up and designed a new motherboard

[Steve]

Absolutely cannot wait to play with one of these with some alt-ram one day

[exxos]

A small update. @Badwolf Was kind enough to kit me out with a proper alt-ram driver for this stuff.

The SEC booster is running 64Mhz with fast-ram & ROM. Technically the results should be 800% alt-ram speed. Though unfortunately I had to add a small wait state on the RAM access despite the chips being 10ns. This is down to parasitics on the motherboard slowing the signal transitions down Though I am currently looking at solutions for speeding it all up. I had previously benchmarked 75MHz CPU in the benchmark section. Though I had problems at that speed as well which I put down to the CPU bottling out. But I think that is technically not true as it is struggling driving capacitance on the bus fast enough. I think if I can solve that problem then the CPU might even push to 80MHz.

IMG_0456.JPG (103.8 KiB) Viewed 2217 times

IMG_0460.JPG (423.25 KiB) Viewed 2217 times