Apollo Maggie 3D Chip - Preview Demo

[pandy71]

Originally Posted by Promilus

No. You're stuck with porting whatever linux community achieved to amiga standards. As I wrote - more available resources are there (both documentations and source code) - easier to use chip features NATIVELY under AmigaOS.

? What is Amiga standard? Mesa? Warp? Isn't more sane is just to run Linux on Amiga?

Originally Posted by Promilus

Maggie has absolutely NO Vulkan capabilities and it would be hard to get even TinyGL level of features implemented (and relevant driver developed). And as I mentioned I'm not really interested with bringing up features like OpenCL, OpenCV, fairly decent GLES or even H.264 codecs to Amiga. I don't care. But USB mass storage with level at least similar to 2.0 standard and mass storage, HID device classes on ANY usb port - that's clearly something which should be achievable and improves functionality for end user. Same with fast SD or with WiFi solutions readily available.
Having hard USB host in ARM SoC doesn't mean it magically supports any USB device. It does only what hardware does, the rest depends on software (so drivers in OS). ARM SoCs aren't magical cure for every problem amiga user might have but are affordable solutions with vast support (yes, mostly in Linux community). And - just like FPGA - if it CAN be used it SHOULD be used.

Perhaps Maggie has no HW Vulkan capabilities but it can form hybrid approach especially if something like Vallium/lavapipe exist already (or https://github.com/kazan-3d/kazan or https://github.com/google/swiftshader) - some limitations may be addressed at HW level...
USB - yes, for sure it will be beneficial but WiFi? how if many modern WLAN chipsets require DSP/security support from main CPU?

[Promilus]

@pandy - I believe you don't have that much knowledge about specific chips.
https://ww1.microchip.com/downloads/...S70005304E.pdf
that's how you can do things and it will work all right on even 8 bit microcontrollers. Because there's already fairly powerful MCU offloading specific tasks from main processor.
As for software graphic pipeline I think you don't realize how much computational power does it take. It's absolutely impractical even on modern CPUs, it's a suicide on Apollo Core. Remember - lack of dedicated units means basically everything except texture mapping and alpha blending has to be done by CPU which is rather slow. Q3 used "shaders" which were calculated by CPU. Those were rather simple ones and even that is beyond capabilities of AC68080. There's no "what if" - it can't do those things fast enough and with decent resolution. It surely can't do VK things which requires access and processing of even more massive amount of data along the pipeline and VK is implicitly designed to make use of multicore architecture (so performance scales better with number of CPU cores, even fairly slow ones, as long as there are many where in amiga there is only 1).

What you can expect from Maggie is something you would see on Saturn or PS1. And that's mighty impressive in it's own way. But as mentioned you can't expect someone to write new 3D games using bare metal approach for V4 and newer - there might be single digit number of small games or more like a demos but that's that. To reach for more you'll have to follow with standards and that's why Warp3D was developed - to expose 3d capabilities of P96-compatible 3d cards and use MiniGL so minimalistic OpenGL implementation capable of running games. And Warp3D is amiga standard (mainly - because there's no other).

There aren't many 3D games for Amiga. Most are ports. There aren't many new ports because 3d accel atm works only with Permedia on BlizzardPPC and VooDoo3 in Mediator/G-Rex/Prometheus (which makes it rather unique and rare combo nowadays). Also - in 68k world 060 isn't backing GPU good enough (and PCI interfaces have their own problems to deal with). That's not the problem with Apollo cards. There's faster CPU, no such silly limitations of data transfers between CPU and GPU. The same applies to e.g. PiStorm. So we might see a flood of new 3D titles as long as RPI 3d capabilities are exposed to AOS and as long as Maggie follows established amiga standards for 3d interface.

[Bruce Abbott]

Originally Posted by Promilus

What you can expect from Maggie is something you would see on Saturn or PS1. And that's mighty impressive in it's own way.

Yep. But not good enough for Amiga fans.

But as mentioned you can't expect someone to write new 3D games using bare metal approach for V4 and newer - there might be single digit number of small games or more like a demos but that's that.

That's right. Amiga fans aren't interested in coding the Amiga 'bare metal', or to be honest any kind of Amiga coding at all.

To reach for more you'll have to follow with standards and that's why Warp3D was developed - to expose 3d capabilities of P96-compatible 3d cards and use MiniGL so minimalistic OpenGL implementation capable of running games. And Warp3D is amiga standard (mainly - because there's no other).

If only Amiga coders had OpenGL back in the day - we would have hundreds of 3D games on the Amiga by now! (not).

There aren't many 3D games for Amiga. Most are ports.

I dispute that. But even if true the reason for that isn't that coding 'bare metal' is too hard. The 3D engine is the easy part. Creating a complete game worth playing is the hard part.

[Promilus]

@Bruce

There aren't many 3D games for Amiga

I'm sorry, I should've write "there aren't many 3D games for Amiga with hardware 3D acceleration" - that was the whole point and it should be pretty obvious from the context (of hardware 3d graphics on amiga).

Amiga fans aren't interested in coding the Amiga 'bare metal', or to be honest any kind of Amiga coding at all.

It is funny you say that (with obvious sarcasm) but that's something which was actually true back in 90s and is true nowadays. Most amiga users aren't software developer and hardly any of those is willing to learn a thing or two about coding in rather ... unwelcoming IDEs (by today's standard)

The 3D engine is the easy part. Creating a complete game worth playing is the hard part

Game isn't rendering engine itself. It's also game mechanics, storyline, artwork. And creating 3d characters and animate them isn't the same effort as creating and animating 2d sprites and requires different tools. Old 3d games like Doom were more 2,5D as plenty of things were still 2D models in reality. Same goes to AvP for Jaguar.
So I ask you now - do you honestly believe there will be any significant number of native and exclusive to amiga 3d titles using maggie? Not reworks, remasters, ports. Brand new, either bare metal or whatever API they apply.

[tygre]

@pandy71

I don't disagree but, if history is any guide, the Vulkan API will be abstracted... it's just that, right now, either people haven't found the right abstraction(s) or people are just too busy using the low-level API.

[pandy71]

Originally Posted by Promilus

@pandy - I believe you don't have that much knowledge about specific chips.
https://ww1.microchip.com/downloads/...S70005304E.pdf
that's how you can do things and it will work all right on even 8 bit microcontrollers. Because there's already fairly powerful MCU offloading specific tasks from main processor.

Well... i prefer to not involve believes in technical discussion - proposed IC is not substantially different form ESP8266 or ESP32.
All of them are 1x1, 2.4GHz, 20MHz BW. No 5GHz, no AX (and this mean that at some point you will be forced to provide dedicated infrastructure only for Amiga).

Originally Posted by Promilus

As for software graphic pipeline I think you don't realize how much computational power does it take. It's absolutely impractical even on modern CPUs, it's a suicide on Apollo Core. Remember - lack of dedicated units means basically everything except texture mapping and alpha blending has to be done by CPU which is rather slow. Q3 used "shaders" which were calculated by CPU. Those were rather simple ones and even that is beyond capabilities of AC68080. There's no "what if" - it can't do those things fast enough and with decent resolution. It surely can't do VK things which requires access and processing of even more massive amount of data along the pipeline and VK is implicitly designed to make use of multicore architecture (so performance scales better with number of CPU cores, even fairly slow ones, as long as there are many where in amiga there is only 1).

So paradoxically Maggie is even better from average Amiga perspective than any modern SoC as all overhead and complexity can be replaced by native hardware access and no API in the middle...

Originally Posted by Promilus

What you can expect from Maggie is something you would see on Saturn or PS1. And that's mighty impressive in it's own way. But as mentioned you can't expect someone to write new 3D games using bare metal approach for V4 and newer - there might be single digit number of small games or more like a demos but that's that. To reach for more you'll have to follow with standards and that's why Warp3D was developed - to expose 3d capabilities of P96-compatible 3d cards and use MiniGL so minimalistic OpenGL implementation capable of running games. And Warp3D is amiga standard (mainly - because there's no other).

There aren't many 3D games for Amiga. Most are ports. There aren't many new ports because 3d accel atm works only with Permedia on BlizzardPPC and VooDoo3 in Mediator/G-Rex/Prometheus (which makes it rather unique and rare combo nowadays). Also - in 68k world 060 isn't backing GPU good enough (and PCI interfaces have their own problems to deal with). That's not the problem with Apollo cards. There's faster CPU, no such silly limitations of data transfers between CPU and GPU. The same applies to e.g. PiStorm. So we might see a flood of new 3D titles as long as RPI 3d capabilities are exposed to AOS and as long as Maggie follows established amiga standards for 3d interface.

To be honest i rather doubt if 3D will be mass experienced in Amiga world - 3D train departed without Amiga on board decades ago - sad but true outcome of the Commodore fail. And as we don't see too many exclusive 3d titles for Macintosh then i don't expect anything for Amiga.
At some point RPi or Maggie can be state of the art in Amiga 3D world.

[pandy71]

Originally Posted by tygre

@pandy71

I don't disagree but, if history is any guide, the Vulkan API will be abstracted... it's just that, right now, either people haven't found the right abstraction(s) or people are just too busy using the low-level API.

For sure - there is no technology given for lifetime - even voltage in power line was raised relatively recently (after almost 100 years).
But for today seem Vulkan is quite appreciated by customers and developers.

[Promilus]

@pandy

All of them are 1x1, 2.4GHz, 20MHz BW. No 5GHz, no AX (and this mean that at some point you will be forced to provide dedicated infrastructure only for Amiga).

That's a common misconception as routers support older standard (but on different SSID than new one). So with home network you most likely have full control about which standard is used. It isn't the same in office, train, mall - but those places aren't the one you get your Amiga into often, right?

proposed IC is not substantially different form ESP8266 or ESP32

Well it is wifi transciever with supporting logic, what did you expect. I just proved you that you don't need to burden main CPU with security as it's already on chip. Also chips using SPI can achieve bigger data rates that's why ENC28J60 (or was it 80) used on Vampire also uses SPI for data transfer. Sure, getting PCI-E based wifi chipset would be even better and something Apollo Team would've been able to do (as FPGA can implement PCI-E as well) but that would require some serious effort to implement (and also would give hint for some ppl to complain why the hell Vampire doesn't support pcie graphic cards). My point with ATWINC was ... if they could make ENC28J60 work for small vampires it takes the same amount of effort to make ATWINC work as well (so, yes, WiFi).

So paradoxically Maggie is even better from average Amiga perspective than any modern SoC as all overhead and complexity can be replaced by native hardware access and no API in the middle...

As usual - it depends. I have no data about how many coders interested in 3d games have apollo hardware and are willing to use Maggie "as is" and how many - in contrast - would like to use OGL implementation on top of Warp3D. What I do know is that there's rather small number of applications using Warp3D in the first place and it hasn't grown much during those past 2 decades.

[jbenam]

Indeed, I am thinking of swapping the Voodoo 3 in my Amiga with a Radeon because there isn't anything very Amiga-y for me to play. 95% of the games are PC ports and I already have a Win9x PC next to the 4000. These games just feel more natural to be played on their original platform.

Literally the only developed from scratch Amiga 68k Warp3D game I can see is Payback?

[Tigerskunk]

Originally Posted by Promilus

As usual - it depends. I have no data about how many coders interested in 3d games have Apollo hardware and are willing to use Maggie "as is" and how many - in contrast - would like to use OGL implementation on top of Warp3D. What I do know is that there's rather small number of applications using Warp3D in the first place and it hasn't grown much during those past 2 decades.

The biggest problem we have in the Amiga community is a lack of able coders creating new stuff.

And those we have tend to concentrate on the old tried and tested things. Which is only natural.

Unfortunately for most people it doesn't make much sense to spend hundreds of hours of their hobby time to create stuff only for people who happen to have a V4 if you could reach most Amiga users if you code something for vanilla AGA&OCS machines up to 68030 with some fast RAM (I'd guess).

But everybody is free to spend their time on whatever they like, so YMMV off course...

[pandy71]

Originally Posted by Promilus

@pandy
That's a common misconception as routers support older standard (but on different SSID than new one). So with home network you most likely have full control about which standard is used. It isn't the same in office, train, mall - but those places aren't the one you get your Amiga into often, right?

It is not misconception, it is impacting whole infrastructure by lowest common denominator - allocating time for slowest member.
2.4GHz is very busy part of spectrum, it is shared by many participants and as Amiga will be not impacted due overall low requirements then other participant may be affected.

Originally Posted by Promilus

Well it is wifi transciever with supporting logic, what did you expect. I just proved you that you don't need to burden main CPU with security as it's already on chip. Also chips using SPI can achieve bigger data rates that's why ENC28J60 (or was it 80) used on Vampire also uses SPI for data transfer. Sure, getting PCI-E based wifi chipset would be even better and something Apollo Team would've been able to do (as FPGA can implement PCI-E as well) but that would require some serious effort to implement (and also would give hint for some ppl to complain why the hell Vampire doesn't support pcie graphic cards). My point with ATWINC was ... if they could make ENC28J60 work for small vampires it takes the same amount of effort to make ATWINC work as well (so, yes, WiFi).

Once again - from practical sense you need to create separate network for Amiga and bridge with other devices over LAN.
I fully understand that this is not problem for Amiga unless you start using something like WPA3.
My point was to show that if you can accept some WLAN NIC limitations then perhaps Maggie can be accepted also - perhaps it is not modern 3D GPU but still better than nothing. Also i see Maggie as living proof of concept that Commodore could propose something like this way earlier and as such perhaps it could at least try to extend own and Amiga existence.

Originally Posted by Promilus

As usual - it depends. I have no data about how many coders interested in 3d games have apollo hardware and are willing to use Maggie "as is" and how many - in contrast - would like to use OGL implementation on top of Warp3D. What I do know is that there's rather small number of applications using Warp3D in the first place and it hasn't grown much during those past 2 decades.

It should be clear that nobody expect from you professional market analysis about expected Maggie popularity. From my perspective Maggie is not substantially different than VBXE for 8 bit Atari. Particular curiosity but if it can be provided within already existing HW by only SW upgrade then why not.

[d4rk3lf]

From Apollo site:

Maggie renders rhino turnable:
[ Show youtube player ]