A314 - The RPi co-processor Official thread

[CrowStudio]

Originally Posted by utri007

I have watched this thread and in post #157 first time somebody says that there option to buy one from ??? Kikem????

Actually if you take a look at post #131 it tells about the RetroWiki page:

Originally Posted by Eriond

I'm very pleased to hear that Edu from Arananet has plans for a batch of A314 boards, as mentioned on the Spanish "Retrowiki": https://translate.google.com/transla...6t%3D200035270
If you don't mind having them shipped from Spain, I suggest you head over to that poll, and sign up for a board (or more )

Edu = Kikems

[JohnnySlick]

I'm in the same boat-- didn't follow the link to see that there was an interest thread elsewhere. I am already on the verge of being culled from here since I don't post that often, so I hope its understandable why I am reluctant to sign up elsewhere as well. That being said, if the interest poll is added elsewhere (here or amibay) I would be happy to sign up.

[Twiggy]

Definitely interested in this

[André]

Likewise, I would also be interested.

[Magic]

I don't own an A600 but I would be happy to chip in on an A500 or A2000 version. What an Awesome project! I almost passed it up because of the thread title. I too clicked on it for curiosity. Keep up the great work, you are all greatly appreciated!

[arnljot]

Is the A314 dependent on serving the chip ram too the Amiga? What if the machine has a megachip and full 2MB chip ram, will the A314 still be able to do it's thing?

[scu98rkr]

Interested in the a600 version have 2 a600's

[hooverphonique]

Originally Posted by arnljot

Is the A314 dependent on serving the chip ram too the Amiga? What if the machine has a megachip and full 2MB chip ram, will the A314 still be able to do it's thing?

The A314 works by 'replacing'/extending physical ram in the machine with its own internal memory. Thus you will get the most out of it by disabling motherboard ram and let the A314 handle all of it.

[scu98rkr]

" No, there's no room for that... but maybe, just maybe, there will be a solution on par with the ECS, but as a separate "product". No promises."

Think this would be quite important.

Notice Lotharek seems to have a way of supporting it.
https://lotharek.pl/productdetail.php?id=281

I guess as you say it need to fit in A600 case.

[arnljot]

Originally Posted by hooverphonique

The A314 works by 'replacing'/extending physical ram in the machine with its own internal memory. Thus you will get most out of it by disabling motherboard ram and let the A314 handle all of it.

So for an a500 rev5 with 1MB motherboard chip, it's best to use a 2MB 8375 Agnus directly on the motherboard then with no megachip?

[hooverphonique]

Originally Posted by arnljot

So for an a500 rev5 with 1MB motherboard chip, it's best to use a 2MB 8375 Agnus directly on the motherboard then with no megachip?

Probably not - I think the pinout is wrong. I am not sure if you can get 2MB chipmem (total) with the A314 without using an A500+ mobo..

[aligebes]

I got a built A314 today, but I think I'm getting hung up in configuring things on the Pi side, as I don't have a lot of experience to work out what I've done wrong..

Does some kind soul have a configured Pi image with everything all setup that they would be willing to share please?

I'm using it with my A500 with 3.1 kickstart, if that helps at all ? and a Pi 3B+

Thank you in advance

[Eriond]

As I wrote in post #131, the only hardware manufacturer that has notified us that they are willing to make boards, is Arananet from Spain:

Originally Posted by Eriond

...Edu from Arananet has plans for a batch of A314 boards, as mentioned on the Spanish "Retrowiki": https://translate.google.com/transla...6t%3D200035270
If you don't mind having them shipped from Spain, I suggest you head over to that poll, and sign up for a board (or more )

Currently there are 17 calls for making a board, while Edu wants at least 30 before going into production. So... if you want a board (and I know from experience that Edu makes high-quality stuff), head over to that forum to sign up! I don't think Edu takes notice if you are pledging your interest here at EAB.

[arnljot]

Any talk about redesign is second hand info afaik. The current board revision seem to be 1.0, and it's working to specifications.

It seems to me a bit much to "demand" deadlines from people who work for free on their spare time. I think the normal open source/amiga axiom applies to this project too: It's ready when it's ready. And the current 1.1 version is even ready to be built. It's up to the community now to adopt I think.

For my own part I've offered to help with documentation.

[Eriond]

Thank you for mentioning that information about redesign is "in the wild". Let me explain the current situation:

Yes, I have had a brief discussion with Edu regarding the possibility to make a CR (cost reduced) version of the A314-500. That did spring from our experience while building the 600 prototype.
It has passed some time since the initial design of the 500 model was done back in early 2018. Price levels of the IC's we use have been adjusted by the market, and an interesting effect is that a 2MB SRAM now costs about as much as a 1MB did back then. Another detail is that our current design only occupies about 14% of the available logic elements in the FPGA. This means we could switch to a smaller, and thereby cheaper variant of that chip.
The reason for the current FPGA size is twofold: In the beginning we had just a vague idea how big our design was going to be, so we chose a sufficiently large FPGA to be sure it would fit. There is also this other idea/notion: since this is an open source project, we had some hopes that people would like to have free space in the FPGA to add their own acceleration / co-processing routines to super-charge specific operations/calculations. For instance, the FPGA has 48 9-bit hardware multipliers, of which we don't use a single one; 2D to 3D matrix multiplications in a single cpu cycle, anyone? There are of course endless possibilities, but someone has to code them and I'm the first to admit that HDL coding isn't for everyone
Finally we have also found a way to, in certain circumstances, simplify the power circuit for the Raspberry Pi. That is, given that the Amigas PSU is up to standards. Which, unfortunately seldom is the case for most Amiga owner out there.

So in summary we have the following cost reducing options:
1) Go for the 10M02SCE144C8G? FPGA variant that retails for apx 6.5€/each, compared to 15.5€ for the one we're using now (having a 65% resource utilization, and leaving only 35% for user enhancements)
2) Find a cheaper 8Mb (512k x 16) SRAM no slower than 12ns. They exist, and cost 6.85€ compared to the 13.7€ we pay for our current memory chip.
3) Completely bypass the 5.15V boost regulator circuit, and feed the Raspberry directly off the Amigas 5V supply rail. While this might sound like an obvious thing to do, we noticed early on that the Raspberry is rather picky about it's power. If it falls below 4.85V it will immediately throttle back it's cores (affects Amiga communication), and if it should fall even lower, like 4.65V, there is imminent risk of data corruption on the SD card.

We have found a way to software-configuration-wise persuade the Raspberry to maintain it's core speed, thereby safeguarding a proper Amiga-RPi comms link, but... we won't be able to avoid other complications such as data loss/corruption and [your favourite deity here] knows what. Besides, the 5.15V circuitry is "only" apx. 5€, but in my opinion 5 well spent €.
This in total means a CR version could shave off between 16 and 21€ in component cost, depending on if you consider your 5V rail to be sufficient (over time )
Now I have to consider if it's worth spending x evenings and nights to do this, or if my time is better spent on designing new and interesting hardware. But maybe you're right... maybe those 16€ are a deal-breaker to some, and it would be a shame if this project fell out of interest due to that.

Anyway, you are the Amiga community; tell me what you think.

[Eriond]

Oh, there's one more thing I'd like to mention. Due to the Chinese new year holiday period, all PCB manufacturing is halted (at least the ones I can afford ). I just missed the window to get boards produced before the holiday, so the A314-600 development is paused until PCBs arrive beginning/mid February.

[Magic]

This is just me personally but for 21€ and all of the extra work it will take I would rather have the extra space in the FPGA as well as more stable power. Of course others will disagree. This is just my vote. Great project and I do hope to own the A500 version soon either way. Thank you to everyone involved for all of your innovations and hard work for very little money.

[CrowStudio]

Originally Posted by Magic

This is just me personally but for 21€ and all of the extra work it will take I would rather have the extra space in the FPGA as well as more stable power. Of course others will disagree. This is just my vote. Great project and I do hope to own the A500 version soon either way. Thank you to everyone involved for all of your innovations and hard work for very little money.

I'd agree with @Magic, is it really worth the extra effort?!

One thing I do not get with some Amigans is their unwillingness to pay for their hardware! If you look at PC gamers they put out alot on their PC builds and new GPU's!


I can agree that some hardware is overpriced. However, with the A314 I do not think that's the case! If you do not have the proper equipment for assembly and programming already, the asking price is really great!



Please keep up the good work producing new amazing Amiga hardware!

[Eriond]

I have checked, and unfortunately both the cheaper FPGA and the SRAM have different footprints/pinouts compared to the ones in v1.1
I have made the decision that it is too much work (for me) compared to the relatively small advantages this would bring, so I'll leave it for the time being.
If someone else feels it's worth the effort, please feel free to make a new spin. It's all on GitHub.

[hyami]

Originally Posted by Eriond

That is one of the mysteries of this project, to me. What is it that makes people so reluctant to get one of these boards?

I'm definitely not reluctant, but I don't have any A500's anymore. I'm awaiting the final version for the A600. I've got twho of those. I just hope it's going to be build by someone who's got experience in smd soldering.


One thing that I couldn't find reading the whole thread again; Niklas started working on the bsdsocket.library last year. What's the status of that? Has it been finished? Does it work well enough to use the A314 as a network card?