Apollo Team announces developing of Vampire standalone version to run as AMIGA and ATARI ST

Moderators: Mug UK, Zorro 2, Greenious, spiny, Moderator Team

ericgus
Captain Atari
Captain Atari
Posts: 429
Joined: Fri Apr 08, 2016 4:53 am
Location: Boston MA - USA

Re: Apollo Team announces developing of Vampire standalone version to run as AMIGA and ATARI ST

Postby ericgus » Thu Oct 27, 2016 7:49 pm

Correct me if I am wrong but all these "new" opcode extensions to the cpu core are great and all but doesn't the code have to be compiled to be aware of these things other wise they wont get used? I know some of the new designs significantly optimized the old original M68K opcodes but anything new opcodes outside those original ones (i.e. modern "imported" opcodes) are pretty much just going to be unused and pointless say to 30 year old binaries. Only new compiled binaries would be able to take advantage of it. Seems very limited audience for this. Dont misunderstand me, its conceptually fantastic, but in practical actual day to day use I don't see these opcode extensions being of much use (aside from hacker and demosceners who could potentially use them.. even then to execute these new binaries you'd also need to have the apollo cpu core on the target machine too)

User avatar
calimero
Fuji Shaped Bastard
Fuji Shaped Bastard
Posts: 2276
Joined: Thu Sep 15, 2005 10:01 am
Location: STara Pazova, Serbia
Contact:

Re: Apollo Team announces developing of Vampire standalone version to run as AMIGA and ATARI ST

Postby calimero » Thu Oct 27, 2016 8:22 pm

ericgus wrote:Correct me if I am wrong but all these "new" opcode extensions to the cpu core are great and all but doesn't the code have to be compiled to be aware of these things other wise they wont get used?

not at all.

To get a sense how CPU business industry work today, you should read Darek Mihocha (author of Gemulator) text about Pentium 4 and NetBurst: http://www.emulators.com/docs/pentium_1.htm

In 80 and 90s CPU manufactures like Intel, Motorola... added new instruction to give programers (and compilers) "tools" to do job quicker. Programers was hungry and used new instruction in new CPUs.
At the end of 90s, things changed: priority for CPU manufactures become to run EXISTING code faster, adding new instruction become secondary. My explanation is that there was too much exiting software, libraries, frameworks at the end of 90s so there was no chance to recompile everything to run faster but CPU simple need to find ways to run exiting code faster.

Mircocode made this possible - CPU start to translate old instruction to microcode that could execute faster, to take advantage of parallelism, new way of predictions... So CPU become somehow "hardware compiler" from old instruction to new microcode. PentiumPro was first intel chips to practice this, MC 68K never got chip with this technology. Apollo is first one that enable this for M68K.


ericgus wrote:I know some of the new designs significantly optimized the old original M68K opcodes but anything new opcodes outside those original ones (i.e. modern "imported" opcodes) are pretty much just going to be unused and pointless say to 30 year old binaries. Only new compiled binaries would be able to take advantage of it.

Apollo team is betting on this that there will be new software that will use new instructions and that Atari and Amiga will get new software that can compete with Windows applications. (I hope too, but odds are really not in favor)
I also wonder what compilers and other tools, libraries... they have for 68080 .... ??? e.g. how complicated would be to make MPEG2 video player application for AmigaOS (or MiNT/TOS) :)

ericgus wrote:even then to execute these new binaries you'd also need to have the apollo cpu core on the target machine too)

because of this they want to keep price down and to have huge, as possible, userbase... to have even Atari users on board.
using Atari since 1986.http://wet.atari.orghttp://milan.kovac.cc/atari/software/ ・ Atari Falcon030/CT63/SV ・ Atari STe ・ Atari Mega4/MegaFile30/SM124 ・ Amiga 1200/PPC ・ Amiga 500 ・ C64 ・ ZX Spectrum ・ RPi ・ MagiC! ・ MiNT 1.18 ・ OS X

User avatar
lp
Fuji Shaped Bastard
Fuji Shaped Bastard
Posts: 2417
Joined: Wed Nov 12, 2003 11:09 pm
Location: GFA Headquarters
Contact:

Re: Apollo Team announces developing of Vampire standalone version to run as AMIGA and ATARI ST

Postby lp » Thu Oct 27, 2016 8:30 pm

.
Last edited by lp on Wed Nov 16, 2016 4:28 pm, edited 2 times in total.

User avatar
calimero
Fuji Shaped Bastard
Fuji Shaped Bastard
Posts: 2276
Joined: Thu Sep 15, 2005 10:01 am
Location: STara Pazova, Serbia
Contact:

Re: Apollo Team announces developing of Vampire standalone version to run as AMIGA and ATARI ST

Postby calimero » Thu Oct 27, 2016 8:33 pm

@rpineau I did not know that Wolfgang start work on Falcon chips! :)

regarding Darek Mihocha, I just check his web site and here is Quick Index running on Gemulator 2000 on old 2.6GHz Pentium 4 Prescott CPU (just to compare):

Image

http://www.emulators.com/docs/nx08_stlb.htm << there is also explanation how these benchmarks works... great text from Darek!


btw
how faster is MiST in compare to 8MH ST?
using Atari since 1986.http://wet.atari.orghttp://milan.kovac.cc/atari/software/ ・ Atari Falcon030/CT63/SV ・ Atari STe ・ Atari Mega4/MegaFile30/SM124 ・ Amiga 1200/PPC ・ Amiga 500 ・ C64 ・ ZX Spectrum ・ RPi ・ MagiC! ・ MiNT 1.18 ・ OS X

ryan
Obsessive compulsive Atari behavior
Obsessive compulsive Atari behavior
Posts: 115
Joined: Fri Nov 07, 2014 12:17 am

Re: Apollo Team announces developing of Vampire standalone version to run as AMIGA and ATARI ST

Postby ryan » Thu Oct 27, 2016 10:28 pm

Well if the Apollo core is truly what they say it is, performance-wise, then they deserve every penny they get and I can understand why they won't release it open source. And it truly would be amazing if they could get an ASIC manufactured, though I really don't know who the market for such a chip would be, and whether they would run the risk of getting their pants sued off by Freescale/NXP.

In any case, I'm willing to throw money at getting Apollo core in some format that can run TOS, but I think an expansion board is a dead end. I really think instead that someone should put together a motherboard (Mini-ITX or something that can fit in the new A1200 cases) that can be configured with either Amiga or Atari peripheral cores on FPGA, and sell the thing with the Apollo core in it for extra $$, or if you want full open source ship it with TG68 or whatever. Maybe there's no market for that either, but I'd buy it, and I bet it could be made cheaper than the Firebee (which I'm also buying) given the component count would be way lower.

ericgus
Captain Atari
Captain Atari
Posts: 429
Joined: Fri Apr 08, 2016 4:53 am
Location: Boston MA - USA

Re: Apollo Team announces developing of Vampire standalone version to run as AMIGA and ATARI ST

Postby ericgus » Fri Oct 28, 2016 5:33 am

calimero wrote:how faster is MiST in compare to 8MH ST?


The MiST implementation of the AtariST is about on par with a stock unit..

I am not too familiar with the ST but if you can provide a link to the benchmark program i'll give it a try with my MiST

AtariZoll
Fuji Shaped Bastard
Fuji Shaped Bastard
Posts: 2978
Joined: Mon Feb 20, 2012 4:42 pm
Contact:

Re: Apollo Team announces developing of Vampire standalone version to run as AMIGA and ATARI ST

Postby AtariZoll » Fri Oct 28, 2016 5:37 am

Is it CISC, RISC is not relevant today. RISC was interesting in times when count of transistors on one chip was much lower, so they needed to simplify some things in favor of speed. Today CISC is fast as RISC, and can perform even multiple instructions at once, in low cycle count. Not to mention that they can put 8-16 cores on one chip. Sadly, even some informatics people still talk about it, as it is 2001 now.
Famous Schrodinger's cat hypothetical experiment says that cat is dead or alive until we open box and see condition of poor animal, which deserved better logic. Cat is always in some certain state - regardless from is observer able or not to see what the state is.

User avatar
calimero
Fuji Shaped Bastard
Fuji Shaped Bastard
Posts: 2276
Joined: Thu Sep 15, 2005 10:01 am
Location: STara Pazova, Serbia
Contact:

Re: Apollo Team announces developing of Vampire standalone version to run as AMIGA and ATARI ST

Postby calimero » Fri Oct 28, 2016 5:47 am

ericgus wrote:
calimero wrote:how faster is MiST in compare to 8MH ST?


The MiST implementation of the AtariST is about on par with a stock unit..

I am not too familiar with the ST but if you can provide a link to the benchmark program i'll give it a try with my MiST


I just found this on youtube: MIST Board based Atari ST on STEroids https://www.youtube.com/watch?v=Yxgwg7qeJyE

Screen Shot 2016-10-28 at 07.46.01.png


Latest GemBench you will find in this thread: viewtopic.php?f=28&t=29168&start=50

Original GemBench can be found here: http://milan.kovac.cc/atari/software/?folder=/SYSTEM
You do not have the required permissions to view the files attached to this post.
using Atari since 1986.http://wet.atari.orghttp://milan.kovac.cc/atari/software/ ・ Atari Falcon030/CT63/SV ・ Atari STe ・ Atari Mega4/MegaFile30/SM124 ・ Amiga 1200/PPC ・ Amiga 500 ・ C64 ・ ZX Spectrum ・ RPi ・ MagiC! ・ MiNT 1.18 ・ OS X

vido
Atari Super Hero
Atari Super Hero
Posts: 671
Joined: Mon Jan 31, 2011 7:39 pm

Re: Apollo Team announces developing of Vampire standalone version to run as AMIGA and ATARI ST

Postby vido » Fri Oct 28, 2016 7:44 am

ryan wrote:Maybe there's no market for that either, but I'd buy it, and I bet it could be made cheaper than the Firebee (which I'm also buying) given the component count would be way lower.

I believe FireBee could be made for 200-300€. Just take design and produce it in China. Anyone can do that. FireBee is produced in Switzerland which is one of the most expensive country in the world, but you get best quality and service.
Anyway ... I wish Apollo would have much better support of community as the FireBee. FireBee is really great computer and it could be made a lot more compatible with the old software as it is and still faster as it is if somebody would develop for it. But comunity is ignoring it. There is almost no FPGA development at all. Well, new and ported software is compiled for it ... but still not enough altough there is great support of development tools. One of the supportes is also Lonny with great support with GFA :)

I hope in Apollo case it will be different and not just a lot of talk. As I see totaly open sourced design depending on community is failure as most of the people expects somebody else will develop instead of them for free. People just complain and theorize, make promisses to do something and then do nothing :)
There is always great difference of what is possible to do and what is done at the end!
And as I see most of the people would like to run old software mostly optimized for slow computers faster. And that is mostly games. That is nonsense for me. If you have new computer you need new or ported software to take advantages of it. Old software and games could be run under emulation on it as you can run it on PC, Mac, ...

MiST is different story. There is a lot of development. But it just recreates old cores without much speed increase. This provers people want to run old software (mortly games) and arent that much interested in new one as they use PCs for every day work.

I just want to tell you that dont expect to much! Insted of talk of what can be done, start talking who will do what without making empty promisses :)
And counting on that SuperVidel will be integrated into Apollo standalone or accelerator ... dont do that. Nature guys dont want to hear SuperVidel would be used in FireBee. Maybe if you pay enough. I dont see why in Apollo case it would be different. But that would make Apollo more expensive. Doesnt it? Also ... Why contribute to Apollo project for free as this is pure commercial project of Apollo team? ;)

User avatar
troed
Atari God
Atari God
Posts: 1438
Joined: Mon Apr 30, 2012 6:20 pm
Location: Sweden

Re: Apollo Team announces developing of Vampire standalone version to run as AMIGA and ATARI ST

Postby troed » Fri Oct 28, 2016 7:57 am

I applaud all retro projects. People do with their spare time what they want to do - nothing more and nothing less.

However. I don't see a project like this ever becoming very compatible with Atari software (which includes demos and games to a large part). If you want a machine that can run everything - and sometimes really really fast when you want the extra speed - just put Hatari onto your random platform of choice. At this point the emulators are far far superior to what can be put into an FPGA.

joska
Hardware Guru
Hardware Guru
Posts: 4343
Joined: Tue Oct 30, 2007 2:55 pm
Location: Florø, Norway
Contact:

Re: Apollo Team announces developing of Vampire standalone version to run as AMIGA and ATARI ST

Postby joska » Fri Oct 28, 2016 9:14 am

troed wrote:However. I don't see a project like this ever becoming very compatible with Atari software (which includes demos and games to a large part).


Hmm... Can you run ST demos on the Falcon? Falcon demos on the TT? Even the MSTE I had back in the day wasn't very compatible with the old stuff, TOS 2 was enough to break a lot of things.

I'm really looking forward to the Vampire. I couldn't care less if it can't run demos or some games. I know it can't. Neither can my Afterburnered Falcon or my Milan060, but I really love those machines anyway despite the fact that they are a lot less compatible with the ST than the Vampire will be. There are - believe it or not - other uses than games and demos :) I have stock (well, sort of...) ST's and STE's for the oldschool stuff.

troed wrote:At this point the emulators are far far superior to what can be put into an FPGA.


At this point modern computers in general is far far superiour to our old machines. But that's not what it's about.
Jo Even

VanillaMiNT - Firebee - Falcon060 - Milan060 - Falcon040 - MIST - Mega ST - STM - STE - Amiga 600 - Sharp MZ700 - MSX - Amstrad CPC - C64

joska
Hardware Guru
Hardware Guru
Posts: 4343
Joined: Tue Oct 30, 2007 2:55 pm
Location: Florø, Norway
Contact:

Re: Apollo Team announces developing of Vampire standalone version to run as AMIGA and ATARI ST

Postby joska » Fri Oct 28, 2016 9:38 am

vido wrote:FireBee is really great computer and it could be made a lot more compatible with the old software as it is and still faster as it is if somebody would develop for it. But comunity is ignoring it.


The problem with the FireBee is the total lack of structured documentation and the silence from the two main developers Fredi and Didier. When I had to flash the PIC in my FireBee I had to piece together lots of fragmented information just to perform this rather simple task.

Look at the MIST. Very well documented hardware and software, we can see the results in the form of lots of cores.
Jo Even

VanillaMiNT - Firebee - Falcon060 - Milan060 - Falcon040 - MIST - Mega ST - STM - STE - Amiga 600 - Sharp MZ700 - MSX - Amstrad CPC - C64

User avatar
calimero
Fuji Shaped Bastard
Fuji Shaped Bastard
Posts: 2276
Joined: Thu Sep 15, 2005 10:01 am
Location: STara Pazova, Serbia
Contact:

Re: Apollo Team announces developing of Vampire standalone version to run as AMIGA and ATARI ST

Postby calimero » Fri Oct 28, 2016 9:56 am

joska wrote:I'm really looking forward to the Vampire. I couldn't care less if it can't run demos or some games. I know it can't.

but on Amiga it RUN all demos and games!

otherwise you have point: TT, Falcon already break compatibility.

btw
now I thinking - is it really necessary to re-create ST MMU or GLUE in Vampire FPGA for compatibility with old games and demos? Can Vampire use existing ST MMU and Glue???
using Atari since 1986.http://wet.atari.orghttp://milan.kovac.cc/atari/software/ ・ Atari Falcon030/CT63/SV ・ Atari STe ・ Atari Mega4/MegaFile30/SM124 ・ Amiga 1200/PPC ・ Amiga 500 ・ C64 ・ ZX Spectrum ・ RPi ・ MagiC! ・ MiNT 1.18 ・ OS X

joska
Hardware Guru
Hardware Guru
Posts: 4343
Joined: Tue Oct 30, 2007 2:55 pm
Location: Florø, Norway
Contact:

Re: Apollo Team announces developing of Vampire standalone version to run as AMIGA and ATARI ST

Postby joska » Fri Oct 28, 2016 10:19 am

calimero wrote:but on Amiga it RUN all demos and games!


Demos on the Amiga is very different from demos on the ST. They use the features of the chipset - blitter, copper, sound - and not exploiting bugs and hidden features like you do on the ST. Just the fact that the Vampire does not run at 8MHz breaks all ST demos made since 1988. If you want a faster CPU then you just have to accept that you can't run any demos. Of course, it would be possible to keep the original CPU for a "compatibility mode", but I don't think that's the plan with the Vampire.

Games is a different thing. I would expect a lot of ST games to work with the Vampire, but many of them might run too fast. That is also the case on the Amiga.

calimero wrote:now I thinking - is it really necessary to re-create ST MMU or GLUE in Vampire FPGA for compatibility with old games and demos? Can Vampire use existing ST MMU and Glue???


If you mean the Vampire 2 then I don't see why you would recreate the MMU and GLUE in FPGA, since these chips are already present on the motherboard... In this case the Vampire replaces the CPU, and offers additional features like fast-RAM, IDE, SD-slot and a graphics card. Nothing else than the CPU is replaced.
Jo Even

VanillaMiNT - Firebee - Falcon060 - Milan060 - Falcon040 - MIST - Mega ST - STM - STE - Amiga 600 - Sharp MZ700 - MSX - Amstrad CPC - C64

User avatar
jayminer
Atariator
Atariator
Posts: 21
Joined: Sat May 17, 2014 5:15 pm
Location: Umeå, Sweden

Re: Apollo Team announces developing of Vampire standalone version to run as AMIGA and ATARI ST

Postby jayminer » Fri Oct 28, 2016 10:46 am

calimero wrote:
joska wrote:I'm really looking forward to the Vampire. I couldn't care less if it can't run demos or some games. I know it can't.

but on Amiga it RUN all demos and games!

otherwise you have point: TT, Falcon already break compatibility.

btw
now I thinking - is it really necessary to re-create ST MMU or GLUE in Vampire FPGA for compatibility with old games and demos? Can Vampire use existing ST MMU and Glue???


I'm pretty sure the Vampire won't run ALL Demos and Games on the Amiga, there has been lots of incompatabilities between different Kickstarts and Chipsets and processors over the years. However on the Amiga we have software like WHDLoad that makes most incompatible games and demos work on "newer" hardware. I also read that there is a "compatability mode" in the latest Apollo Core for the Amiga which runs the CPU at the same speed as a 68000-based machine, so something similiar could probably be in the Atari-version aswell, hopefully solving most of the compatability issues.

User avatar
alexh
Fuji Shaped Bastard
Fuji Shaped Bastard
Posts: 2723
Joined: Wed Oct 20, 2004 1:52 pm
Location: UK - Oxford
Contact:

Re: Apollo Team announces developing of Vampire standalone version to run as AMIGA and ATARI ST

Postby alexh » Fri Oct 28, 2016 10:51 am

I do worry for the Apollo team should they attempt to go down the ASIC / IP route. At the moment as a Retro FPGA project they are off people's radar. But should they attempt to try and sell into the same market as Intel/ARM/Freescale/IBM they will get slammed. It is almost impossible to design a modern microprocessor without tripping over someone else's patents. Trust me, that is what I do for a living and have done for 17 years. Even with a small amount of protection that having your own patents gives you (so you can arrange patent exchange) they are crossing a potential minefield.

They do have some truly astounding benchmarks though. Not owning a Vampire I do hope they correspond to real-world user experience.

jayminer wrote:I also read that there is a "compatibility mode" in the latest Apollo Core for the Amiga which runs the CPU at the same speed as a 68000-based machine, so something similiar could probably be in the Atari-version as well, hopefully solving most of the compatibility issues.

The Atari requires cycle accurate compatibility due to the programming techniques the programmers employed. That accuracy goes against the whole point of Apollo. It will need a fall-back to real Physical 680x0 for compatibility. Like some classic Amiga accelerators have. Easy to do, just needs a bit of PCB space.

User avatar
1st1
Atari Super Hero
Atari Super Hero
Posts: 824
Joined: Mon May 07, 2012 11:48 am

Re: Apollo Team announces developing of Vampire standalone version to run as AMIGA and ATARI ST

Postby 1st1 » Fri Oct 28, 2016 12:03 pm

calimero wrote:but on Amiga it RUN all demos and games!


Good Demos on Amiga and ST are different.

Good Amiga demos are using Amiga hardware good to make good effects. They use the Amiga hardware in a way it was made for.

Good ST demos are using ST hardware to make things the hardware was not designed for. Overscan, more colors, digi sound, fine scrolling, ... This is very tricky and needs clock cycle accuracy. On an accelerated machine, or on a modifyd 68000 (has same speed) with different execution times for the same instructions, but same 8 MHz clock, such demos will not look nice, sound will not listen nice. You can plug 68020/30 in ST and these demos will not look and sound nice anymore.

Same for some good games.

PS: ST MMU and Glue are just thumb asics which contain more or less stupid TTL logics. There is nothing configurable by software there, etc., except for memory banking. Everything intelligent is in the other chips. Glue and MMU are only there to save a lot of space for 74LSxxx TTL chips.
Power without the Price. It's not a bug. It's a feature. _/|\_ATARI

1040STFM in PC-Tower (PAK68/2, OvrScn, 4 MB, 1GB SCSI, CD-ROM...) * 2x Falcon 030 32GB/14MB+ScrnBlstrIII * 2x TT030 73GB/20MB+Nova * 520/1040STFM * 520/1040STE * 260/520ST/+ * some Mega ST * 2x Mega STE 500MB/4MB+M.CoCo * Stacy * STBook * SLM605 * SLM804 * SLM605 * SMM804 * SH 204/205 * Megafile 30/44/60 * SF314 * SF354 * 5x Pofo * PC3

User avatar
1st1
Atari Super Hero
Atari Super Hero
Posts: 824
Joined: Mon May 07, 2012 11:48 am

Re: Apollo Team announces developing of Vampire standalone version to run as AMIGA and ATARI ST

Postby 1st1 » Fri Oct 28, 2016 12:10 pm

alexh wrote: ... But should they attempt to try and sell into the same market as Intel/ARM/Freescale/IBM they will get slammed. It is almost impossible to design a modern microprocessor without tripping over someone else's patents.

...

jayminer wrote:I also read that there is a "compatibility mode" in the latest Apollo Core for the Amiga which runs the CPU at the same speed as a 68000-based machine, so something similiar could probably be in the Atari-version as well, hopefully solving most of the compatibility issues.

The Atari requires cycle accurate compatibility due to the programming techniques the programmers employed. That accuracy goes against the whole point of Apollo. It will need a fall-back to real Physical 680x0 for compatibility. Like some classic Amiga accelerators have. Easy to do, just needs a bit of PCB space.


True, they don't want to compete against AMD, Intel, ARM and ... Microsoft. I don't know what they exactly plan with a commercial version, but it will not be a thing to compete PC.

Yes, there seems to be a thing to slow down Appollo, but not cycle exact. For that you can buy an original Amiga or ST for a few bucks and stand it beside. Take care that it does not get jealous...
Power without the Price. It's not a bug. It's a feature. _/|\_ATARI

1040STFM in PC-Tower (PAK68/2, OvrScn, 4 MB, 1GB SCSI, CD-ROM...) * 2x Falcon 030 32GB/14MB+ScrnBlstrIII * 2x TT030 73GB/20MB+Nova * 520/1040STFM * 520/1040STE * 260/520ST/+ * some Mega ST * 2x Mega STE 500MB/4MB+M.CoCo * Stacy * STBook * SLM605 * SLM804 * SLM605 * SMM804 * SH 204/205 * Megafile 30/44/60 * SF314 * SF354 * 5x Pofo * PC3

ryan
Obsessive compulsive Atari behavior
Obsessive compulsive Atari behavior
Posts: 115
Joined: Fri Nov 07, 2014 12:17 am

Re: Apollo Team announces developing of Vampire standalone version to run as AMIGA and ATARI ST

Postby ryan » Fri Oct 28, 2016 3:39 pm

vido wrote:FireBee is really great computer and it could be made a lot more compatible with the old software as it is and still faster as it is if somebody would develop for it. But comunity is ignoring it. There is almost no FPGA development at all.

MiST is different story. There is a lot of development. But it just recreates old cores without much speed increase. This provers people want to run old software (mortly games) and arent that much interested in new one as they use PCs for every day work.


Well apart from MiST and Firebee being two totally different things, there are two key differences: price, and breadth.

I have a Firebee on order from the current batch being made. It is ridiculously expensive. I'm fine with that because I am a well-paid software engineer who can afford such a hobby toy. Most people can't. MiST is (relatively) cheap. It's possible to buy it 'for fun', it's like 1/4 or 1/5 the price of Firebee..

And it supports a wide breadth of machines, and it seems like a big chunk of the publicity for it is from the Amiga community, not the Atari.

But the two are totally different. Firebee is something new, not a 'retro' device. It's nice that they've put as much work into making it compatible, but I personally don't care about that. For compatibility I can run an emulator. What I want from Firebee is a novel new device that shows us what a 68k GEM/TOS based machine would look like on more recent hardware.

Unfortunately at this point Coldfire is a dead-end. They won't be making any new models of them. And old 68k platform retro enthusiasts were never Freescale's target market for it anyways. I have no doubt that Coldfire can outperform Apollo core on a lot of benchmarks. But Apollo seems to have a development community actively working on it.

There will come a point when projects like Apollo are the only place to get a 'modern' well performing 68kish CPU.

vido
Atari Super Hero
Atari Super Hero
Posts: 671
Joined: Mon Jan 31, 2011 7:39 pm

Re: Apollo Team announces developing of Vampire standalone version to run as AMIGA and ATARI ST

Postby vido » Fri Oct 28, 2016 6:38 pm

ryan wrote:Well apart from MiST and Firebee being two totally different things, there are two key differences: price, and breadth.

I agree on that. But as I wrote, I believe the price of the FireBee could be much lower. Just somebody should produce it in china or by any other cheaper producer as it is now. But quality would suffer and probably you would not have 2 years warrantty.

And it supports a wide breadth of machines, and it seems like a big chunk of the publicity for it is from the Amiga community, not the Atari.

You can put the whole MiST into the FireBee FPGA to run faster and parallel to the FireBee. Somebody should "just" do it. MiST screen could be on the FireBee desktop in GEM window. Well ... I am not 100% sure this is possible but with my (poor) knowledge I dont see why this wouldnt be impossible?

But the two are totally different. Firebee is something new, not a 'retro' device. It's nice that they've put as much work into making it compatible, but I personally don't care about that. For compatibility I can run an emulator. What I want from Firebee is a novel new device that shows us what a 68k GEM/TOS based machine would look like on more recent hardware.

100% my toughts! :)

Unfortunately at this point Coldfire is a dead-end. They won't be making any new models of them. And old 68k platform retro enthusiasts were never Freescale's target market for it anyways. I have no doubt that Coldfire can outperform Apollo core on a lot of benchmarks. But Apollo seems to have a development community actively working on it.

Not exactly. You can buy Coldfire licence and put Coldfire core into FPGA. Coldfire core in FPGA is allready faster as Coldfire used in the FireBee. And if you buy such licence you can do ASIC version of the Coldfire that would be much faster. ASICs versions of Coldfires are being produced by companies for their internal use. Coldfire licence holder likes FireBee project. Maybe it would be possible to get some kind of sublicence to use ASIC Coldfire core of some company to be used only in FireBee computer. But that nobody asked them yet. Other way licences for Coldfire core are to high to be used in FireBee computer.

There will come a point when projects like Apollo are the only place to get a 'modern' well performing 68kish CPU.

I really like Apollo project. But for now it is just not mature enough. It didnt reach the final ISA stage by now. Still no FPU and MMU.

AtariZoll
Fuji Shaped Bastard
Fuji Shaped Bastard
Posts: 2978
Joined: Mon Feb 20, 2012 4:42 pm
Contact:

Re: Apollo Team announces developing of Vampire standalone version to run as AMIGA and ATARI ST

Postby AtariZoll » Sat Oct 29, 2016 8:17 am

Maybe it is worth to start complete new thread, but I will put it here for now:
Team working on Vampire has pretty high ambitions with this, and targets some high speeds (GHz CPU, Majsta mentioned even mobile devices ..)
So, basically they really want that it will be competitive with modern CPUs. What people here mostly don't see. That's why they need large user base, contributors from second biggest 68K CPU platform (in sales, Atari is still biggest overall :D ) - hmm, I forgot Apple, but they are traitors, so don't count. All this want to be way over what Firebee targeted.
And then we need to ask main question: is forcing 68K CPU architecture good idea at all now ? Basic design is now old over 35 years. I like it's register model, mnemonics, asm coding, but that is not what makes progress. + I don't think that it was what made Atari ST to what it was. 68K was good choice around 1983-92, even IBM had it as choice for first PC. But development was stopped by Motorola self later.
What we want here actually ? Amiga people want to continue Amiga line, Amiga OS. There is even Power PC based Amiga follower - I don't know how successful is it, and how much SW is made for - they tried what Apple did, but small team.
Even Vampire team does partial abandoning of 68K compatibility - making some new PMMU, new modes, adding new instructions. And that's actually the only way to make it faster. Here I must say few words about their benchmarks and claims that it is 200x faster than 8 MHz 68000. Since Vampire is clocked at some 100 MHz, later seems really exaggerated - 12x bigger clock and 200x bigger speed ? Only explanation is that they count in test AMMX.
But that's very unfair and we should have speed test based on old SW, what runs on 68000. So, how much is Vampire faster with some old Atari ST, Amiga 500 SW ? I think that then we will see much lower number, max some 30-40. And we know now that Coldfire is still faster, but "less compatible" - that makes me laughing - because Vampire is at moment 0% Atari ST compatible - and will be until we see some SW working on - at least TOS.

All this talk above is to make my point : forcing old architecture, but better is to say register model and instruction set of 68000 may be just bad idea.
Some will say that Intel did exactly that, and still uses old 8086 instruction set, of course expanded. But here may be the key: 68K instruction set is very well structured, and there are often operands embedded in instructions self (moveq etc.). Adding new instructions can be problem, and was problem.
Why Motorola abandoned whole 68K line, and made on new ? Maybe just because it was not good for expanding instruction set.

So, for those who want to keep TOS, GEM, maybe Atari ST basic architecture alive (what is what defined Atari ST line most) - can it be redone for some different CPU ? Afterall, you programm mostly in C, so CPU type matters not much.
Famous Schrodinger's cat hypothetical experiment says that cat is dead or alive until we open box and see condition of poor animal, which deserved better logic. Cat is always in some certain state - regardless from is observer able or not to see what the state is.

User avatar
calimero
Fuji Shaped Bastard
Fuji Shaped Bastard
Posts: 2276
Joined: Thu Sep 15, 2005 10:01 am
Location: STara Pazova, Serbia
Contact:

Re: Apollo Team announces developing of Vampire standalone version to run as AMIGA and ATARI ST

Postby calimero » Sat Oct 29, 2016 10:23 am

AtariZoll wrote:And then we need to ask main question: is forcing 68K CPU architecture good idea at all now ? Basic design is now old over 35 years.

you could ask same question for x86, but look what intel and AMD manage to make from 35 years old design (as you mentioned later in your post).

AtariZoll wrote: because Vampire is at moment 0% Atari ST compatible - and will be until we see some SW working on - at least TOS.

EmuTOS working on Vampire.

AtariZoll wrote:All this talk above is to make my point : forcing old architecture, but better is to say register model and instruction set of 68000 may be just bad idea.

Apollo team already spend 8 years developing Apollo core so today status is: take it or leave it :D
using Atari since 1986.http://wet.atari.orghttp://milan.kovac.cc/atari/software/ ・ Atari Falcon030/CT63/SV ・ Atari STe ・ Atari Mega4/MegaFile30/SM124 ・ Amiga 1200/PPC ・ Amiga 500 ・ C64 ・ ZX Spectrum ・ RPi ・ MagiC! ・ MiNT 1.18 ・ OS X

AtariZoll
Fuji Shaped Bastard
Fuji Shaped Bastard
Posts: 2978
Joined: Mon Feb 20, 2012 4:42 pm
Contact:

Re: Apollo Team announces developing of Vampire standalone version to run as AMIGA and ATARI ST

Postby AtariZoll » Sat Oct 29, 2016 11:29 am

calimero wrote:
AtariZoll wrote: because Vampire is at moment 0% Atari ST compatible - and will be until we see some SW working on - at least TOS.

EmuTOS working on Vampire.

EmuTOS for Amiga, on Amiga.
Famous Schrodinger's cat hypothetical experiment says that cat is dead or alive until we open box and see condition of poor animal, which deserved better logic. Cat is always in some certain state - regardless from is observer able or not to see what the state is.

User avatar
1st1
Atari Super Hero
Atari Super Hero
Posts: 824
Joined: Mon May 07, 2012 11:48 am

Re: Apollo Team announces developing of Vampire standalone version to run as AMIGA and ATARI ST

Postby 1st1 » Sat Oct 29, 2016 11:49 am

vido wrote:
Unfortunately at this point Coldfire is a dead-end. They won't be making any new models of them. And old 68k platform retro enthusiasts were never Freescale's target market for it anyways. I have no doubt that Coldfire can outperform Apollo core on a lot of benchmarks. But Apollo seems to have a development community actively working on it.

Not exactly. You can buy Coldfire licence and put Coldfire core into FPGA. Coldfire core in FPGA is allready faster as Coldfire used in the FireBee. And if you buy such licence you can do ASIC version of the Coldfire that would be much faster. ASICs versions of Coldfires are being produced by companies for their internal use. Coldfire licence holder likes FireBee project. Maybe it would be possible to get some kind of sublicence to use ASIC Coldfire core of some company to be used only in FireBee computer. But that nobody asked them yet. Other way licences for Coldfire core are to high to be used in FireBee computer.

There will come a point when projects like Apollo are the only place to get a 'modern' well performing 68kish CPU.

I really like Apollo project. But for now it is just not mature enough. It didnt reach the final ISA stage by now. Still no FPU and MMU.


There is 68040 compatible FPU in Apollo. They only removed some FPU instructions which do not make sense. Nobody uses them. All math instructions are already included in Apollo FPU, except for tangens function and a few monsters like that. The reason that they didn't include is because tangens needs more than 400 clocks for execution. You can have tangens function in 68k plus other FPU instructions in less than 300 clocks. And while tangens ins executed by FPU, you have to wait with the main processor on the result. System does nothing for 400 clocks, except of waiting for the tangens result. Replacing code can be interrupted, multitasking, and so on. So nobody use tangens and a few similar operations from FPU. So Apollo team did not include it. Additionally they have AMMX, running 64 bit instructions in 32 bit processor, even AMD, Intel and ARM can't do that. AMMX is very fast, can play MPEG better than PPC 600 MHz, and since 2 days they can execute two AMMX instructions in one clock. They have tested MPEG on Vampire II already in Amiga web browser, Youtube...

PMMU is a topic, and a fter discussing with Gunnar about it, it looks like he understands more why ATARI guys needs it. (MiNT, ROMRAM-patches on TT/Falcon/...).

The other point, for near future: Why using Coldfire core in an FPGA if you also can have Apollo core in FPGA or even ASIC? Apollo seems to be much more compatible, and it has active developement. And look about policy of Freescale... FIrebee shoud rework their design, replacing Coldfire by FPGA/ASIC and having the choice using Apollo.
Power without the Price. It's not a bug. It's a feature. _/|\_ATARI

1040STFM in PC-Tower (PAK68/2, OvrScn, 4 MB, 1GB SCSI, CD-ROM...) * 2x Falcon 030 32GB/14MB+ScrnBlstrIII * 2x TT030 73GB/20MB+Nova * 520/1040STFM * 520/1040STE * 260/520ST/+ * some Mega ST * 2x Mega STE 500MB/4MB+M.CoCo * Stacy * STBook * SLM605 * SLM804 * SLM605 * SMM804 * SH 204/205 * Megafile 30/44/60 * SF314 * SF354 * 5x Pofo * PC3

User avatar
Eero Tamminen
Atari God
Atari God
Posts: 1928
Joined: Sun Jul 31, 2011 1:11 pm

Re: Apollo Team announces developing of Vampire standalone version to run as AMIGA and ATARI ST

Postby Eero Tamminen » Sat Oct 29, 2016 11:09 pm

AtariZoll wrote:EmuTOS for Amiga, on Amiga.


Clean Atari GEM programs work on Amiga EmuTOS. If somebody writes fVDI drivers for Amiga color modes, also in something else than monochrome. What higher graphics modes Vampire supports besides Amiga's standard planar modes?

I.e. for clean Atari GEM programs only CPU compatibility will be an issue. If programs need MiNT, then somebody would need add basic Amiga HW support for MiNT too...


Return to “FPGA Chat”

Who is online

Users browsing this forum: No registered users and 2 guests