Apollo Team has new Cyclone 5 FPGA accellerator cards

[ptyerman]

Originally Posted by Gunnar

Two questions:

1) We could sell the card as pure CPU upgrade earlier.
But then all what the users will get is very fast 68K CPu plus 128 fast memory.

Would this be an option for the people?


2) Would people be interested in something like blog which we create.
We could add an update every other day or post some pictures.
Are people interested in this?

Yes to both questions from me. The early adopters could even be used as beta testers, giving feedback on how it works with different set-ups.

[Overmann]

Yes to both questions. I would buy this for my A2000 in an instant! Especially if there was some sort of OCS/ECS/AGA passthrough or similar so that I wouldn't need two monitors.

I recently bought an ACA1232, but if this was released for the A1200 I would buy another 1200 just to use it with this board (I have an unpopulated A1200-tower just sitting here with a Zorro-busboard in it).

[Gunnar]

To prevent any misunderstanding.

Phoenix -> http://www.apollo-core.com/index.htm?page=releases

Phoenix is 100% 68000 compatible.
This means all code from the 68000 will run on it.
Phoenix support many 68020 and up instructions like 32bit mul, or 32bit Div or 32bit branch...
Phoenix also supports some new stuff.
For example there is a Chunky 2 Planar instruction
which converts 8 pixels in a single ASM instruction in 1 cycle.
Phoenix removes some addressing limits of older 68000-68060 Cores.
And Phoenix supports more registers...

To sum up:
Phoenix is fast - faster than 68060 in our tests.
Phoenix is in some ways easier to program than old 68k as some limitations are moved.
Phoenix has more registers which allows to easier implement some complex functions fast.
Phoenix has some new killer instructions which allow to do the work of many ( up to 100) old 68K instruction in a single instruction.

But of of today we do NOT guarantee 100% 68020 compatibility.
As a matter of fact right now no softcore does guarante this.
This means some 68020 will run some wont.

There might come out an upgrade which will fix this.
But for the early/first release - this is not guaranteed.

[tnt23]

Originally Posted by Gunnar

2) Would people be interested in something like blog which we create.
We could add an update every other day or post some pictures.
Are people interested in this?

Absolutely!

[voyager]

Originally Posted by tnt23

Absolutely!

Yes please.

[bubbob42]

Originally Posted by Gunnar

Two questions:

1) We could sell the card as pure CPU upgrade earlier.
But then all what the users will get is very fast 68K CPu plus 128 fast memory.

There are lots of compareable options on the market, but people needing an accelerator now might be interested.

I'm waiting for the RTG/HDMI option. That's what's really interesting here.

[tolkien]

I think that there aren't a compareable thing for the Amiga. It will be the fastest Amiga 68k you can see.
I would buy it in Its cpu upgrade and could wait for the other features.

[wawa]

Originally Posted by Samurai_Crow

My current Blizzard1230mk4 has 16 megs of Fast RAM, SCSI2 adapter, FPU and full MMU installed. I might be tempted to sell it off and replace it with the FPGA board if ThoR could make a version of MuForce, MuGuardianAngel, Mungwall emulation from debug mode on AROS (Does that even use an MMU?), etc. that would run on it. It need not have a 68030 or even a 68060 compatible MMU since those are the only MMU utilities that would need to work with it. Those utilities are great for development and the main reason ACA accelerators don't sufficiently satisfy me or convince me to switch (not to mention that ACA doesn't go higher than an 'EC030 anyway).

looks like aros68k is not able to offer features available with the mutools on genuine amiga os, that means backtrace given non stripped binaries. at least its how i understood toni, since i would love this functionality and just asked him about that yesterday.

[wawa]

Originally Posted by Gunnar

Two questions:

1) We could sell the card as pure CPU upgrade earlier.
But then all what the users will get is very fast 68K CPu plus 128 fast memory.

Would this be an option for the people?

yes. especially that i understand that the hardware is set in stone for starters as soon as the interface part has been finalized. as soon as all the hardware in question is present and only the vhdl to support it and drivers are still missing it would be fine.

2) Would people be interested in something like blog which we create.
We could add an update every other day or post some pictures.
Are people interested in this?

i would prefer hard facts rather than a blog. also blog costs time to maintain. as i said, for me the biggest motivation is to observe the repository, even if i dont understand all the details. i know it is out of question for non open sources, but something of sorts would be most welcome.

[UberFreak]

I have only one question about the accelerator & its CPU core.

In its current state, will I be able to run scene demos written specifically for 68060 ?

[Adrian Browne]

Yup, I would take one earlier and wait for upgrades in good time.

I have an a1200 and an a500 needing repairs.

[Mrs Beanbag]

personally i am only really interested in the CPU side of things

[Jack Burton]

I'm interested by the CPU side of this card too. I hope that if/when released we will be able to see program that really push the Amiga chipset. I think AGA has still a lot to show. And personnally I dream of a usable FM soft synth (6 or 8 voices) using PAULA + the fast CPU (as FM is ultra rare in Amiga land)...

[JimDrew]

Originally Posted by Gunnar

As a matter of fact right now no softcore does guarante this.

That's not true. The FPGA Arcade Replay's CPU core offers full 68EC020 and 68020 modes. All 020 instructions are supported.

[Gunnar]

Originally Posted by JimDrew

That's not true. The FPGA Arcade Replay's CPU core offers full 68EC020 and 68020 modes. All 020 instructions are supported.

Can you please tell us which core this is?
The TG68 is only partially compatible, it used to miss 68010 and 68020 instructions.

[Lord Aga]

Yup. Not to bash FPGA Arcade, I was actually looking forward to it for years. But it is still elusive, not clear, and not on open sale, as far as I know. Granted, some guys have the dev boards, and they report nice things but it is usually in the form of "this particular thing works, it's cool". Mike was on a roll for a while, but I haven't heard anything from him recently.

On the other hand, I can understand the skepticism about the Apollo project. But this is not something that was cooked up yesterday. It has been in steady development for years. And those guys are not going to stop now

[Gunnar]

Originally Posted by wawa

as i said, for me the biggest motivation is to observe the repository, even if i dont understand all the details. i know it is out of question for non open sources, but something of sorts would be most welcome.

If you want to "feel" the development and see how the devs are discussing internal topics then maybe you will like our IRC channel. There are constantly people there and there are discussions of ideas / changes etc.

[Don_Adan]

Originally Posted by Gunnar

To prevent any misunderstanding.

Phoenix -> http://www.apollo-core.com/index.htm?page=releases

Phoenix is 100% 68000 compatible.
This means all code from the 68000 will run on it.

Then movep commands are supported for Phoenix ?
Next commands also works OK for Phoenix?
movem.l ,-(SP)
movem.l (SP)+,
movem.l $10(A5),A3-A6

[Gunnar]

Originally Posted by Don_Adan

Then movep commands are supported for Phoenix ?
Next commands also works OK for Phoenix?
movem.l ,-(SP)
movem.l (SP)+,
movem.l $10(A5),A3-A6

Yes
and
Yes,
Phoenix supports some MOVEM variant which the old 68k did not.

[wawa]

Originally Posted by Gunnar

If you want to "feel" the development and see how the devs are discussing internal topics then maybe you will like our IRC channel. There are constantly people there and there are discussions of ideas / changes etc.

have logged in one or two times out of curiosity, but nothing was going on and i dont like to bother the developers. a public mailing list might be more suitable, still public relation is secondary issue that shouldnt get in the way of development.