| FORUM

FEDEVEL
Platform forum

Fedevel server project

EVW13 , 05-18-2017, 06:15 AM
Hi Robert,

I've just seen your new video called “What next” in which you mentioned new server design and lack of your own time and forces- can I offer my free help as a PCB Designer and CAD librarian? I have lots of experience and capabilities in High-Speed and HDI stuff.

P.S. Don't even try to do server using Kicad, it's only possible in Allegro and maybe in Altium(with pain, and probably not possible at all)

Best Regards
EVW13
mairomaster , 05-18-2017, 06:42 AM
Hi EVW13,

I am just curious - what difficulties do you expect with Altium, doing server design? I've worked on big designs and I've had various problems as well.

Thanks
EVW13 , 05-18-2017, 07:16 AM
Hi mairomaster,

First of all I need to say that among all proprietary PCB software Altium has worst high-speed functionality set in this class. MG PADS, Cadence OrCAD, Zuken CADSTAR and even Pulsonix have much better high-speed stuff: in basic functions, in advanced possibilities and constraints of course. This reason is enough already, as mentioned design is all around tough high-speed blocks- but actually there are more limitations, one of the most important is lack of performance and overall stability: we all know what it takes to repour polygons and run DRCs on big and complex boards

robertferanec , 05-18-2017, 08:26 AM
Thank you @EVW13

- KiCAD would be only used if for example Google decided to make everything free. Of course, in that case Google (or another big company) would need to invest also into KiCAD development.

Altium could be used for the server design, but I would prefer Cadence:
- I would like to cover also this other software. Covering Cadence could help me to get on this market and it could help to grow FEDEVEL Academy ... at least, that is the theory .
- I think, Cadence is more expected software to be used for this kind of design
- I was expecting Altium actively help with some of our projects, but even we talk to each other, I would not say that "we are really good friends supporting each other" So, there is no a real reason to stick with them.
EVW13 , 05-18-2017, 09:41 AM
Yes, Allegro is very common software for sever-like level of design complexity, and what about Kicad- it doesn't matter how much money Google or any other company can invest into open source, because any proprietary EDA is already far beyond with greater functionality and lot of patents

Meaning Altium- yes, one can try to apply tons of scripts, magic and more, but still it will take few orders more time than in case of suitable software. Of course if that one will be lucky enough and make at least 50% of design, as there will be much more “surprises”.

In the end, what is you opinion about my offer?
robertferanec , 05-18-2017, 07:05 PM
@EVW13 I really appreciate your offer Thank you very much.

Are you interested specially about the sever project? Maybe I should make a mailing list for the server project where everyone interested could sign up and if the project will become alive, maybe people could help to make it happen. What do you think?
EVW13 , 05-18-2017, 09:25 PM
@robertferanec generally, I'm interested in participating in some open-source(fully or partially open) project, but conducted at high-level(x86, GPU, tough GPGA at least etc.), and looking at preliminary description you provided in your video, I consider this server project as best variant. My goal is fun- PCB design is nothing more but a paid hobby for me, and as a contractor/freelancer I have time and capabilities for this

What about mailing list of interested candidates- it's up to you, if you're ready to receive yet another bunch of emails on daily basis Generally, I think it's worth to consider at least 3 variants, from light to tough:

1)x86/GPU based motherboard(embedded applications)
2)x86 based mother board(PC-like)
3)x86 based motherboard(server-like)

But of course in case of server-like board, it will change the game: nowhere in the Internet there is anything which can even touch such level- 99% of current open source stuff is built like really ugly toys.
robertferanec , 05-19-2017, 11:46 AM
Oki, let's see how it goes. For now, this big server project needs to wait until I find a way how it could be financed. When something changes I will definitely post about it. I will then also create mailing list so people interested about the project can stay in touch
EVW13 , 08-26-2017, 11:36 PM
Hi Robert,

Just wanted to ask you about current status of server project- how it's going and is it still on the track to be done? I was looking a lot into server designs from OCP during last couple weeks and the idea seems more and more feasible, but with some notes:

- Allegro only Never ever Altium can handle it.
- Intel stuff only: their packages have extremely nice pinout so even 2 Xeons you can route nicely on 8 layers. It's easy to route and easy to make educational package from such design.
- Two CPU maximum should be nice limit
- current or -1 generation of CPU

As more light alternative there may be regular PCB motherboard design based on either Intel Core or AMD Ryzen, maybe even with embedded graphics

Best Regards
EVW13
robertferanec , 08-27-2017, 07:21 AM
EVW13,

- Allegro: I am working on this. Currently I am preparing a new Learn the essentials of OrCad and Allegro - the first step for the server project. I talk to Cadence and I have support from them ... they are helping me with course preparation (to make it really good) and they also helped me with their software license.

- Intel: This is more tricky point. Intel is ideal, they have the best CPUs. However, I will need to select the manufacturer which recognizes the value of the project and will be willing to support it (and talk to me).

Once I finish the Allegro Essentials course, then I will be re-considering the server project again. Let see if some of the CPU manufacturers will be interested to support it. The biggest problem is, that these people are not interested to talk. But I do not blame them. They probably get a lot of offers for projects / cooperation every day + they have a lot of their own projects and work. So, if there is another 'I have this project in mind', they are just not interested to listen. The key is to catch their attention and make them to want to support the project .... How? I have no idea
EVW13 , 08-27-2017, 08:07 AM
Well, I believe you should think about crowd funding as well- the probability of making entire PCBA with no charge( as open source support) is close to 0, but with some money that should be more feasible What about interest from CPU manufacturer- at least such design, as widely available thing, should attract a lot of attention all around and thus be good for marketing and commercial intents of manufacturer.
robertferanec , 08-27-2017, 08:53 AM
What about interest from CPU manufacturer- at least such design, as widely available thing, should attract a lot of attention all around and thus be good for marketing and commercial intents of manufacturer.
I thought exactly the same when we designed number of iMX6 open source designs. I thought that Freescale / NXP / Qualcomm will find these open source designs interesting and they will help to support the projects, because these designs helps them to get new customers ... nope, doesnt work like that. There is only one person in whole NXP company who likes the projects and talk to me about them.

I may make a video about this topic
EVW13 , 08-27-2017, 09:05 AM
Holy f, this is so retarded- in my opinion they should donate you and promote And of course it will be great if you cover things in video, I personally would like to watch.
Use our interactive Discord forum to reply or ask new questions.
Discord invite
Discord forum link (after invitation)

Didn't find what you were looking for?