Which is the fastest software C2P 1x1 routine

[matthey]

Originally Posted by mc6809e

Actually bitplanes are more efficient for every non power of 2 colors display.

A linear increase in memory usage allows for an exponential increase in color. For example, going from 16 to 32 colors takes 25% more memory but doubles the number of colors. And an 8 color image wastes memory access slots if the display hardware requires 4 bits per pixel. The bitplanes scheme can also help area fill operations since a single plane can be area filled and this single plane can then quickly be written (sometimes inverted) into display planes to produce a poly of a particular color.

Bitplanes are efficient if only considering display bandwidth and memory used. They also allow an accurate palette of colors with fewer colors. But then there is the cost of updating and drawing visuals which is commonly more expensive than chunky even with hardware help like a blitter. The extra cost increases with each additional bit plane. The processing power and resources necessary to work with planar has to be considered in determining efficiency.

Originally Posted by mc6809e

At the time the Amiga was designed, most displays were actually chunky in the sense that they might use 2 adjacent bits or 4 adjacent bits per pixel for color selection and memory was still very expensive per bit. The initial design of the Amiga actually called for something like 128k of chipram.

Had the Amiga not been delayed by corporate politics, the machine would have been released earlier and the bitplane idea would have been seen as a great advantage.

The only way to have more colors back then was to reduce the display bandwidth and memory needs. Amiga tricks like bit planes and HAM mode did that. These tricks required more processing but the Amiga had custom chips to accelerate them. The display bandwidth improved and memory prices dropped some. The Amiga with it's custom chips was still competitive against CLUT8 and 8 bit chunky which were generally faster but didn't look as good. Improving the Amiga's custom chip and memory speed could have kept them viable for longer but display bandwidth continued to improve, memory prices continued to drop and texture mapping was discovered which worked better with chunky. The new 16 bit chunky modes had speed and good enough color accuracy (especially with dithering). Texture mapping games pushed the use of chunky forward with economies of scale. By the time 24 and 32 bit chunky arrived, The Amiga was left with slower and inferior quality gfx. It now had to do c2p for many chunky games making it even slower. Somehow in all those years, Amiga only gave us 1 minor and 1 major custom chip update, without chunky modes, and the bit plane "engine" and chip memory speed didn't get any faster. A 68030 is faster than the blitter on every Amiga. We can have faster Amiga compatible custom chips in an fpga now. A much faster blitter and chip memory are not too bad for planar, which does have some advantages. Saving some display bandwidth in an fpga does also have a small advantage. However, we need to add and support 16 and 32 bit chunky ASAP.

[Mrs Beanbag]

there are all sorts of neat tricks you can do with bitplanes that you can't do with chunky mode, because you can write to and scroll bitplanes independently you can do transparency, shadows &c with very little overhead.

[matthey]

No one has mentioned multiple playfields and color cycling as a planar advantage yet .

Chunky went 3D and eventually developed Z-buffers, stencils, alpha blending, transparency, fog, multi-texturing effects, etc. The resource costs are much higher but some quite amazing effects and demos are possible (and not just 3D). If fpga prices continue to fall, 3D could follow right after RTG for the Amiga. It should even be possible to overlay (like genlocking) an Amiga traditional planar display over a chunky display for the ultimate in old school planar with modern chunky. That's what the Natami was planning. Maybe it's a little bit of dreaming but the Amiga (and cooperating developers) could make it possible.

[Mrs Beanbag]

there's no reason why colour cycling couldn't apply to paletted chunky modes though

[Thorham]

Originally Posted by matthey

If fpga prices continue to fall, 3D could follow right after RTG for the Amiga.

Sounds like Amiga becoming more and more like the peecee, and I already have a peecee Part of the charm of Amiga hardware is exactly that it doesn't have those features. This way, the Amiga hardware just becomes glue. This is the problem I have with adding PPC+RTG+Sound card. Where's the Amiga in that?

I'd rather see brand new Amigas made with ASICs that are based on the reversed engineered Amiga chip set+choice of 68030 or 68060 (again ASIC) with modernized connectivity (SATA/Ethernet/Scandoubler/Flickerfixer, etc) and loads of RAM. That's right, no chip set upgrades and real 68K. I'd even prefer a reverse engineered Paula to have the same non-linear DACs. That way it's really an Amiga, and not a peecee wannabe. If only I was filthy rich

[matthey]

Originally Posted by Mrs Beanbag

there's no reason why colour cycling couldn't apply to paletted chunky modes though

Right. It really has more to do with a CLUT than planar gfx but many Amiga users would associate it with planar gfx. The good old CLUT8 would allow it too.

Originally Posted by Thorham

Sounds like Amiga becoming more and more like the peecee, and I already have a peecee Part of the charm of Amiga hardware is exactly that it doesn't have those features. This way, the Amiga hardware just becomes glue. This is the problem I have with adding PPC+RTG+Sound card. Where's the Amiga in that?

The Amiga used to be a PC (Personal Computer) before the PC stole the exclusive use of the name. I want a modernized 68k which I believe can be competitive enough in performance to be fun and usable for modern computing. I would also prefer an integrated custom chip enhancement (SoC) for gfx (chunky), 3D and sound. There should be efficient RTG gfx and sound with enough performance that programmers could use them. I am still for opening up most of the hardware specs and letting programmers bang the hardware if they want (more for driver creation, learning, demos).

Originally Posted by Thorham

I'd rather see brand new Amigas made with ASICs that are based on the reversed engineered Amiga chip set+choice of 68030 or 68060 (again ASIC) with modernized connectivity (SATA/Ethernet/Scandoubler/Flickerfixer, etc) and loads of RAM. That's right, no chip set upgrades and real 68K. I'd even prefer a reverse engineered Paula to have the same non-linear DACs. That way it's really an Amiga, and not a peecee wannabe. If only I was filthy rich

ASICs would cost a lot of money and everything would need to be developed and debugged in an fpga first. It would have to be pretty awesome to be worthwhile also. That probably means enhancements that would appeal to others, probably some "PC" features you may not consider enhancements. Everything would have to be integrated well and polished. I thought the Natami route was a good step toward a possible future ASIC. Maybe it was too early and too expensive. The fpgaArcade is now carrying the custom chip torch forward if they plan on releasing the custom chip updates as promised (necessary under the MiniMig license). It will be interesting to see what enhancements happen from there. The fpgaArcade concept of a programmable fpga for hardware emulation isn't exactly aligned with becoming a super Amiga ASIC. However, MikeJ has put a lot of effort into providing high quality emulation and it shows. Sound output is one area that can be immediately discerned by a good ear and sets the atmosphere. The Amiga Paula did have amazing quality for it's time and still sounds good today. I wonder how good and accurate the Paula (and C64 SID) sounds on the fpgaArcade?

[Thorham]

Originally Posted by matthey

The Amiga used to be a PC (Personal Computer) before the PC stole the exclusive use of the name.

You know what I mean

Originally Posted by matthey

I want a modernized 68k which I believe can be competitive enough in performance to be fun and usable for modern computing.

Then why not drop Amiga hardware compatibility entirely and make something completely new? We already have Amigas, and a new 68K machine would certainly be interesting. Much more interesting than a souped-up Amiga clone.

Originally Posted by matthey

ASICs would cost a lot of money and everything would need to be developed and debugged in an fpga first.

Yeah, that's why I wish I was filthy rich, that way I could just do it.

Originally Posted by matthey

The Amiga Paula did have amazing quality for it's time and still sounds good today.

Sure does. Plenty of music that sounds really good in 44100 hertz 16 bit stereo. Things like classical piano pieces sound noisy, though.

Originally Posted by matthey

I wonder how good and accurate the Paula (and C64 SID) sounds on the fpgaArcade?

Depends on what you're playing on the Paula. If it's a WAV ripped from CD played in above mentioned mode, you won't hear much difference, except for the noise caused by the non-linear DACs not being present. For many other things that typical Amiga sound is probably just caused by the lowfi samples.

[Mrs Beanbag]

Originally Posted by Thorham

Then why not drop Amiga hardware compatibility entirely and make something completely new? We already have Amigas, and a new 68K machine would certainly be interesting. Much more interesting than a souped-up Amiga clone.

Did you think that AGA was "not an Amiga" when it came out?

I wouldn't mind a "souped-up" Amiga but i'd be quite picky about exactly how it was done, i wouldn't just want a standard PC graphics card, it would have to be something special in the Amiga way of thinking.

[TCD]

Originally Posted by Mrs Beanbag

I wouldn't mind a "souped-up" Amiga but i'd be quite picky about exactly how it was done, i wouldn't just want a standard PC graphics card, it would have to be something special in the Amiga way of thinking.

And yet the souped-up Amigas use standard PC graphics cards. I think that Thorham draws the line when Commodore went bankrupt and the 'lineage' wasn't continued, but I'm not going to argue about it here. Maybe it's better to just get back on topic...

[matthey]

Originally Posted by TCD

And yet the souped-up Amigas use standard PC graphics cards. I think that Thorham draws the line when Commodore went bankrupt and the 'lineage' wasn't continued, but I'm not going to argue about it here. Maybe it's better to just get back on topic...

With Thorham's retro Amiga thinking, adding fast c2p may or may not be necessary at all.

1) c2p not important because all programs should directly use Amiga planar gfx
2) c2p important because chunky screen modes are non-Amiga
3) c2p not important because chunky modes are not too "PC" for the Amiga

I use a 3000T with CSMK3 68060@75MHz+UltraSCSI 30MB/s HD, Mediator+Voodoo4+ethernet+usb and I can say it doesn't feel like an old Amiga anymore. It's fast and I can do a lot more with it than a 68000 Amiga 1000 (an awesome machine for 1985). I do wish the gfx were integrated better (one monitor output) and had AGA compatibility. Chunky is a great fit for the Amiga. C= should have added it to the Amiga and would have added it with AAA. The fastest c2p is avoiding c2p and that should be the goal of fpga Amiga projects, IMO of course .

[Samurai_Crow]

Originally Posted by matthey

The fastest c2p is avoiding c2p and that should be the goal of fpga Amiga projects, IMO of course .

I'm afraid I have to side with matthey. Of course it is much easier to put one screen pointer in a Copper-list than 8 bitplane pointers. As such, it would be faster in most circumstances.

Chunky would take better advantage of modern sequential-access burst-fetches that modern memory controllers use. No amount of money is going to change that or it would have by now on the other platforms.

I'd even go so far as to say that the round-robin DMA scheduler in AGA shouldn't be a requirement for future chipset cores because it, once again, uses very random memory accesses while a new one could be optimized to be more sequential.

[Vot]

Originally Posted by Thorham

You know what I mean


Then why not drop Amiga hardware compatibility entirely and make something completely new? We already have Amigas, and a new 68K machine would certainly be interesting. Much more interesting than a souped-up Amiga clone.

.

I agree completely. Like in os 4.1 the old systems should live on via emulation. Hardware compatibility is pointless. While its fine to recreate the old system, these shouldn't be the basis of future ones. Even commodore was going to ditch legacy support. They were aiming for api compatibility through the os not hardware register compatability.

What the amiga community needs is a 100% open system with 100% open hardware so its cheap enough for everyone who is an amiga fan to adopt. It needs to be redesigned from the ground up.

[Gunnar]

My opinion to this topic is:

1) Building a compatible AGA system is possible

2) Extending the AGA system with some features like being faster or having more planes, or more colors is not complicated and can be done in a compatible way.

3) Creating a RTG hicolor /truecolor display is quite easy

4) Adding the RTG into the old system in an "amiga-like" way - which means the copper can change the RTG registers or that AMIGA sprites can be displayed on the RTG - as not difficult

5) FPGA resources to create an RTG display are minimal

6) FPGA resources to create the compatible and exnhance AGA system are not big.

This means from a technical point of view - one can for less than $10 FPGA cost create an AMIGA compatible system which can display OCS/AGA modes which can display some extra features like 9 Plane EHB, which supports AMIGA sprites and which can display an extra chunk 8/16bit or 24bit playfield in addition to the planar playfields.

So why not?

[alexh]

Because it is a waste of resources and almost no one would have or use the new features?

1) Yes - The work is already done / in progress for MiniMig AGA. Competition Welcome. This is where your time spent will benefit all Amiga users. In the past I considered AGA more important than CPU development but Apollo/Pheonix may prove me wrong if it can make a new generation of classic accelerators. I reserve judgement.

2) The only users would be FPGA owners correct? IMO the number of applications would dry up very fast. You yourselves have struggled to find anyone to code applications for Apollo/Phoenix enhancements.

3) Cool but would just an RTG screen be enough? For it to be an enhancement over say a classic AGA Amiga + PCI graphics card you'd need to invest a lot of time and effort? no? There is hardware offload of 2D and 3D functions in today's Amiga RTG graphics cards isn't there? You'd have to do a lot of work?

4) See my opinion for (2)

[Gunnar]

Originally Posted by alexh

1) Yes - The work is already done

Yes it was already done in the NATAMI project.
Also the topic 3D acceleration was planned for in the NATAMI.

What I say is it can be done.

And I'm sure it will be done.

There are several people involved in the APOLLO project which will continue to drive this at some point.

[matthey]

Originally Posted by alexh

2) The only users would be FPGA owners correct? IMO the number of applications would dry up very fast. You yourselves have struggled to find anyone to code applications for Apollo/Phoenix enhancements.

The number of programmers using specific enhanced custom chip features would probably initially be few unless they were able to be supported by existing software. This would be the case for chunky RTG gfx modes as soon as P96/CGFX/AROS support was added. Reverse engineering and optimizing drivers is something I know about as opposed to writing hardware banging demos which sounds like fun but is a specialized skill. We also need to support and improve high level languages in order to attract programmers. I think I can help here also. I am currently working on more modern and optimized vclib code for vbcc. I would be surprised if Frank Wille did not quickly add support for Apollo in vasm when the time comes. Chris Hoehne created a Phoenix target for GCC. I'm sure Samurai Crow will help to add support in AROS for enhancements. Even meynaf would probably add support in his picture viewer and games despite his frustrations with Gunnar. Natami drew quite some interest by looking at the web site hits. It's amazing how many hits it still gets considering how few posts there have been recently. Faster 68k compatible processors and an enhanced custom chip set could breath new life into the dying Amiga. How fast it could come back would of course depend on performance, prices and features. Those who say it won't happen are ignoring the reality of the fpgaArcade and Vampire accelerators.

[britelite]

Originally Posted by matthey

The number of programmers using specific enhanced custom chip features would probably initially be few unless they were able to be supported by existing software.

I'd go as far as saying there won't be much support for any enhanced features as long as (Win)UAE doesn't support them

[JimDrew]

FPGA Arcade Replay's Amiga core already has full AGA support, and RTG will also be available. I have chatted with Gunnar about the Apollo core, so I have an idea of where they going with that. MikeJ (owner/developer of the FPGA Arcade Replay) is working on MMU/FPU support in his CPU core. MMU/FPU is really what will set apart his CPU core and will open the door to large number of programs that can be used - like my Mac emulation, enforcer, VM programs, and speed up renderings dramatically due to the FPU support, etc. Mike also needs this CPU core for the Atari Falcon emulation.

[Mrs Beanbag]

Originally Posted by britelite

I'd go as far as saying there won't be much support for any enhanced features as long as (Win)UAE doesn't support them

where we're going, we don't need UAE

[matthey]

Originally Posted by britelite

I'd go as far as saying there won't be much support for any enhanced features as long as (Win)UAE doesn't support them

UAE support would create a much larger user base helping to proliferate enhancements. The same applies for CPU enhancements. However, UAE may have trouble emulating some custom chip enhancements at full speed. Also, Toni Wilen seems resistant to adding support for enhancements. Maybe that is so he doesn't get stuck doing a lot of emulation work for vaporware hardware. Creating open standards and documentation for enhancements would help also.

Originally Posted by JimDrew

FPGA Arcade Replay's Amiga core already has full AGA support, and RTG will also be available.

68020+AGA+RTG is a very good base Amiga to support. The fpgaArcade should help turn around the Amiga user base erosion. We shouldn't stop there if we want to attract non-Amiga users. Adding a fast modernized 68k CPU and 3D support could attract the hobbyist computer market (similar to the RaspberryPi) and have more embedded applications in addition to the retro market. Some people think like Thorham and would stop at bringing back a memory. Other people want more.

Originally Posted by JimDrew

I have chatted with Gunnar about the Apollo core, so I have an idea of where they going with that.

I didn't know Gunnar had it figured out where he was going with it yet . It remains to be seen if all the problems and complexity can be managed. It's certainly a lot of work with little in compensation at this point. There is huge potential though. Phoenix looks promising but it is barely scratching the surface of what Apollo should be able to do. Of course the potential applications for it extend outside of the Amiga market as well (primarily embedded).

Originally Posted by JimDrew

MikeJ (owner/developer of the FPGA Arcade Replay) is working on MMU/FPU support in his CPU core. MMU/FPU is really what will set apart his CPU core and will open the door to large number of programs that can be used - like my Mac emulation, enforcer, VM programs, and speed up renderings dramatically due to the FPU support, etc. Mike also needs this CPU core for the Atari Falcon emulation.

That's good news. I hadn't heard that a 68k MMU or FPU were being created for the fpgaArcade. I have heard about the Suska 68030 fpga MMU targeting the Atari. I would wish for a 68040/68060 compatible MMU as the 68030 MMU is rather ancient but there probably is more software using the 68030 MMU. The FPU support is more interesting to me. Are they planning full 6888x support? It would be challenging to implement all the FPU instructions and it would take a huge amount of fpga space. Do you have any links to public forums where support is being discussed?