Vampire 600?

[matthey]

Guys
Majsta no longer manufactures the boards. His old list is pointless ( I was on it and never heard back as well) Kipper2k has made some (I have one see my sig) but not sure how many he has left or will continue to produce. This board is recommended for advanced users only.

There is an agreement for kipper2k to finish and support the boards for American customers (time consuming process with hundreds of boards). I don't know why majsta has not been better at keeping track of customer interest but kipper2k is reliable and honest if anyone wants to contact him about getting an accelerator. There are a couple of caveats with the accelerator.

1) The clip on part of the accelerator has a better connection on some Amiga 600s than others. Indivisions had a similar style attachment and some problems also.

2) Only the 68000 TG68 fpga core without caches works reliably. I expect this is what most owners are using and it's a big upgrade over the 68000. The TG68 68020 core with caches has problems. It's possible this core could be updated when the fpgaArcade releases it's sources. The Phoenix core for majsta's accelerator could be the fastest core but is still being worked on and doesn't fit well in the small fgpa. The 68020 TG68 core with caches or Phoenix core would need to be flashed, maybe several times to fix bugs. It's not too difficult but probably not for beginners either.

The TG68 68000 core and plenty of fast memory are enough to make the accelerator a good value. If that is the only expectation and there are no problems with the clip on part, then there is a good chance of a satisfied customer.

[matthey]

Hi thanks for the info and I am afraid probably not for beginners either knocks me out of the running and it would have been nice if majsta had sent me a quick email speaking of the problems as I did email him and receive some feedback on his web page, best wishes Brian.

I believe the Vampire still comes with the 68000 TG68 core which works out of the box, even for beginners. Your ACA620EC is pretty nice and the Vampire with this core would not offer much if any performance gains (the 68020 is much stronger than the 68000 TG68 without caches per clock). Your ACA620EC supports 68020 compatible code which is a large advantage for applications and games. The Vampire has significantly more memory which may be worthwhile if you are trying to use larger applications but offers little for games. I wouldn't bother upgrading (or should we say side-grading for now) if I was you unless a better core is released.

[matthey]

Its a crazy as having a Ferrari in the USA, right?
A fast car in a country with a speed limit around 80 miles.
But you know what, there are many people enjoying a Ferrari still.

Not all Ferraris are that crazy. My uncle has a low mileage Testarossa in California. Yea, no place to drive it or his quicker turbo Porsche 997 (still wasn't impressively quick or nimble, IMO). My RX-7 (high 12s in the 1/4 mile) is actually quicker than a stock Testarossa (mid 13s in 1/4 mile). The Testarossa may be a little faster with 180mph top speed because of more favorable gearing at the cost of gas mileage. The Testarossa looks fast and sounds cool though.

Putting a fast accelerator in an old Amiga might be more comparable to putting a BMW M3 engine in an old Volkswagen beetle (ready for the autobahn?).

[matthey]

umm unless you have a heavily modified rx7 there is no way on God's green earth that its faster than the Ferrari. Sorry

LOL. It's a stock motor, stock turbos, stock tranny 1993 twin turbo base model with bolt-ons (exhaust with cat though, intake, intercooler, pulleys, fuel injectors and pump, PFC computer) with about 14 PSI (1 atmosphere) of boost. It's about 350HP weighing ~2700lbs and did 12.9@110mph at the drag strip on street tires. A little 1.3 liter 2 rotor Wankel engine could never outperform a 4.9 liter V12 in a "Ferrari". Stock vs stock, the 3rd gen RX-7 would probably be quicker 0-60mph, the Testarossa would be a few tenths better in the 1/4 mile and the Testarossa is about 20mph better top speed. Then the RX-7 can be uncorked and the boost can be turned up while the Testarossa is already high compression.

http://en.wikipedia.org/wiki/Testarossa
http://en.wikipedia.org/wiki/RX-7

[matthey]

Is there enough logic to squeeze SATA inside?

SATA and PCIe require high speed transceivers which some fpgas have but they up the cost. The main focus of an accelerator for the 500/1000/2000/CDTV is to keep the price down. SD cards are small, low power and offer enough storage for (the new) low end Amigas.

Anyway, what causes the instabilities that is mentioned?

You mean the non-68000 CPU cores of majsta's accelerator? The TG68 was not originally designed to use a cache and the problems have to do with adapting one to it. It's obviously possible to get it working if investing time into this compatible core (the fpgaArcade has 68020+cache TG68 working reliably) but the Apollo Team (including Gunnar/BigGun) would rather work on the faster Phoenix core. The Phoenix core has bugs to work out because it's completely new. There is a good chance that the TG68 68020+caches or Phoenix core will be available this year.

[matthey]

The CD32 is small and has an external keyboard port.

And it has AGA, a cheap connector for the accelerator and goes for a reasonable price. I already made the argument to Gunnar that the CD32 would make a good cheap AGA test platform. The connectors are cheap for the 500/1000/2000/CDTV also and there are way more of these Amigas sitting around that could be given new life.

There is a low end CD32 expansion in the works over on EAB by the way:

http://eab.abime.net/showthread.php?t=72361

Hey big gun. I'll try to catch up on the irc when are you guys there? I have a couple of things to discuss from my testing with .1 and 1.0 cores as well.

Think German time for IRC. It's not exactly convenient for people in the states.

There is some information on the Vampire CPU cores in the "Amiga FPGA Accelerator Owners Club" social group over on EAB.

http://eab.abime.net/group.php?groupid=82

You may have to sign up for the social group in community -> social groups.

[matthey]

Its cool that the Apollo team is working with Phoenix now. I've been running 0.2 core for over a week now pretty stable but some memory operations crash or freeze the amiga and I get things like 435,567,078 Chip ram!  

I will try that version of Dopus. Ill have to check my version.

One of my main reasons for going vampire was I thought it would be better than the 620 aca board as i need 020+ for my applications.

I would have worked on the TG68 68020 nocache with full memory core first. The clock speed will probably be somewhat slower but it becomes more powerful per clock to make up for some of it. I suppose some people would miss the compatibility with the original 600 68000 but then it gains more software compatibility with 68020 support. The Vampire doesn't have any kind of fallback to the 68000 does it?

[matthey]

AFAIK vampire doesnt have fallback or disable to revert to 68000.

I didn't think it did but I wasn't sure. A CPU fallback with or without extra memory would have been nice as Majsta could have focused on the 68020 core rather than a compatible 68000 core.

 I had the a600 on for over 12hrs unzipping archives and ran like a champ. We need one last core for TG68 just like you said with no cache and proper memory map. For instance I was using Hoffman and Akira's PT1210 dj software and after like 20mins locked up on me like 3 times! This must be do to memory configuration ( I may go back to 0.1 core to test)  

Majsta has the memory mapped in pretty much the most compatible way. Some older software, especially games, may expect memory to be at a particular address. All these different memory locations fragment memory more making large memory allocations less likely to succeed (mostly affects newer productivity software). They should not cause a crash or lock up with properly written software though. Maybe mapping the memory all together in the 68020 core will reduce complexity. This is how more modern accelerators work and incompatibility isn't really noticeable with productivity and newer software.