Power, Memory and Schematics


I think the 4GB is more important than 1.5GHz, or maybe both are not that important for end user. Rock stable is the most important thing.

Sent from my SM-G900F using Tapatalk
As someone using a 2GB OMAP5 devboard for a couple years now, I can say I barely use the amount of system memory that has, Being able to reach 1.5GHz is more important.

It's most likely a power stability issue, something that can be worked out.
 
What we need is a license that prohibits charging for the schematics, but allows charging for a product made using them. Surely there is a solution?

I should think that is exactly what we have here. Schematics are NC, under copyright. AFAIK hardware design falls under patent law. You can use parts (i think), but if you lift the whole thing, it would be AFAIK a case of EDs wording, and he specified the use. That is if you wanted to be righteous about it, but we all know, and saw what the intended use was, to ask first.

I condemn restrictions on commercial use of the schematics. I want to implore you to also forbid modifications. Non-commercial restrictions are infectious to at least the same extent as copyleft.

If you have, say, 10 people modifying an NC-licensed work, that means you need to go and get permission from all 10 of those people for any commercial use, which depending on the circumstances may be impossible.

This property is only acceptable if you consider all commercialism to be something you want to oppose, which I'm sure you don't. In this sense, a fully proprietary license is better than a license that permits modifications but not commercial use.

This is wrong. Copyleft only has a "infectious" effect on turnarounds, as it gets progressively harder to hollow out the freedoms granted. NC is comparatively dead in the water, as you cannot get a ball rolling if you need to ask everyone.

And it is then similar in that it is hard to hollow out any freedoms with it. Yes, they may not be granted, but we live in a best of breed world. There is no need to do so yourself, out of fear that it would suddenly work…

CC-BY-NC does not dictate that all modifications made to it be licensed NC too, you could pile non-NC atop it, and even sign away CLAs.

Goldelico and ED can each launch products using the design. Reading the original statement, it states an intent of sharing back to learn, which is as copyleft as it gets.

It is impossible to make any inroads on 'bad-things-can-happen' with NC, because you cannot challenge scale on your own, or at home, and for all intents then the proprietor is ED/hns.

I think it constitutes open hardware, where hardware is not free hardware, because it isn't software, but you would only be employing the same ideas. Copyleft AFAIK, do not exist functionally for hardware, because it is so cumbersome to get patents.

Funnily enough, the argument would make sense if you made it against the BY clause, rather than the NC, because that can be a pyramid-scheme, similar to original BSD, which requires attribution for each and every modification along the way.
 
I should think that is exactly what we have here. Schematics are NC, under copyright. AFAIK hardware design falls under patent law.

That may be true; I am not a lawyer, and I have no idea how the relevant laws differ between jurisdictions. It is clear, however, that the use of "NC" here is causing some FUD, and so should be avoided if without good reason.

Besides, who would try to sell the schematics -- or succeed -- if they are well archived on some gratis service (perhaps archive.org, the Internet Archive)? Is the risk of someone trying to sell the schematics really worth considering? Especially as they are likely to be freely available to anyone with internet access (at least in the free/rich world) for the foreseeable future? Ignorant or uninformed people may be duped by someone trying to sell something free, but I assume that someone who could make use of the schematics would be reasonably savvy.



Also, I have no experience in reading schematics, but do all the crosses indicate parts removed from the design, or is there more than one type of cross (and meaning)?
 
The 1,5GHz freeze has not yet been fixed. We had hoped that the one trace we cut was the issue, but it wasn't
[...]
Feel free to look through them, learn from them - or report any potential issues you find to us, so we can check them!
Not sure about the SOCs power domains, but In the cpu-sch.pdf, I see mostly bulk capacitors (10uF, 2.2uF) on non-RAM supply lines. Because of capacity vs. size, the dielectric of bulk capacitors is often "slow" (1.5GHz clock means that you will see those 1.5GHz on the bypass capacitors, which means they have to be "fast" enough (ESR)).

Most designs I've seen so far have bulk AND high-speed capacitors for each power domain, for each physical location (e.g. if you have 2 VCore pins apart from each other, you have 2 Bulk* capacitors and 2 High-speed** capacitors). You can then easily try a better dielectric, like NP0 or C0G without decreasing the overall capacity (since you still have the the bulk capacitor).

In case the capacitor values are equal to a working reference design***, there is still the possibility that the capacity (and ESR) "seen" by the chip is different if the voltage rating or dielectric is lower, since ceramic capacitors change the capacity with DC voltage applied, especially for X5R (or even Y5_), which is not too uncommon for bulk capacitors.

* e.g. between 100nF and 10uF, depending on overall power consumption
** e.g. between 1nF and 100nF, depending on peaks
*** e.g. OMAP Eval board
 
Last edited:
I would also be okay whit 1,3 ghz, but as the Omap was meant to have 2 x 1,5 ghz, it should make sure to work out..
also the 1,3 ghz sounds pretty slow, if you see the Pandora 1 ghz, but its 2 x 1,3 ghz, so it should be lot faster than my CC Pandora..
Whit the Cases its sounds pretty good, as i can remember, they were the issue of the pandora..
I hope you find the time to make some new Case Videos, to show how good the hinge works, and other parts of the case, like the dpad and the shoulderbuttons..

i ditnt care much about the shematics, as i m not the hardware guy who understands shematics, but its pretty cool that there are getting released..
 
Also, i have a CC Pandora, which i can overclock to about 900 mhz, and its also quite good for my tasks, so even a dual core pyra at 1 ghz would be much faster, than the 900 mhz CC Pandora..
I hope to get decend Dreamcast and N64 Emulation, and the last one works for some games quite good on my Pandora (per exemple mario 64), but there are more demanding games like the zelda games, which ditnt work that good,
the dreamcast got some cool Truck Games so it will be the ideal handheld for me
 
Hi @matzesu : clock speed is not that relevant IIRC, i.e: you'd be able to do a lot more with a single core Pyra @ 1 GHz than with a Pandora @ 1 GHz.

Cheers, Magic Sam
Yes, certainly.
One core running at 500MHz should already be a bit faster than the Pandora running at 1GHz. Clock rates can only be compared on the same generation of CPUs. :)
 
Thanks for the update and your work. I'm happy to see the progress, and fwiw here're my priorities:

- Sleep. I'm against labour abuse, and being self-employed or enterpreneur is absolutely no excuse for condescending to unhealthy work conditions. Slavery used violence in addition to social constructs, and as it evolved toward serfdom, absolutism, industrialist capitalism and eventually "web 2.0 collaborative economy" it is virtualizing chains more and more until they are build by the slaves themselves in their own brains. You can do what you want to your own health but you don't get the right to sell the result. Delay, ship as is, cancel or do whatever it takes, but get healthy sleep, please. Otherwise the Pyra hanging in a Christmas tree might be a very macabre gift.

- Open hardware. I dislike the NC clause. Of course authors can choose to license as they please, but it's dissappointing. I know most hardware is even worse with secret schematics, but we can still wish for more. Some possibilities might be relicensing freely later on, either at a given date or at a given profit, or maybe just setting a price on the design under a free licenses and letting people contribute to it in some crowdfunding to recover costs. I don't think copyright stops any copycats. They just don't care and operate somewhere you won't easily sue them. But NC stops any group putting enough effort that it organises itself to pay for some or all of the effort. With that perspective, even individuals may feel little motivated to contribute. This is about design modifications. The production of the device itself should not be covered by copyright, although I've seen some surprising news posted in the EOMA68 mailing list (one of the possible copycats that I'm sad to see prevented). Anyway, maybe the license only applies to PDFs because they are not that useful as source material to derive from and it is going to be delivered in more source-like form with a free license sometime.

- Speed. I'd like as much MHz as possible so I welcome your efforts to reiterate and fix it. I don't know enough electronics to guess the costs or probabilities, so I must trust you there. But 1,2 GHz should be enough if it is becoming too expensive to fix.

- RAM. I'm now sorry to once have suggested to wait to test 4GB before ordering, since in the end you've had to order anyway without final 4GB tests and it might have cost you more. I didn't think in rice fluctuations. I hope you wasn't hit too hard. I always tend to think components get cheaper until they're no longer available :(. I'm curious what changes are needed. I guess RAM configuration is done in u-boot. Then what needs to be changed in the kernel ? Enabling LPAE? regulators or clocks in DTS ? I used to think a 32 bit ARM SOC could not use 4GB without LPAE because part of the address space was needed for MMIO, ROM, etc. But then I saw one boot with boot an LPAE kernel and the same without LPAE configured and top reporting all the 4GB and was confused as how that was possible... I feel I know less each day.

- Videos. I wouldn't care much about videos when there are other more urgent things to do. Even this textual news updates I like so much may wait if you have other work. Don't worry too much.

Well, I'm happy prototypes seem to be getting soonish to developers. I hope that will help a lot in development and hopefully reduce a little load on the few people that is currently working on it. And it will be someone able to shoot videos who is not busy in more urgent things.

Thanks for your efoorts, but please take care.
 
Last edited:
Yes, certainly.
One core running at 500MHz should already be a bit faster than the Pandora running at 1GHz. Clock rates can only be compared on the same generation of CPUs. :)
You got my Unit 2 Times on your Workbench (one Case Change, and one time to fix the Display) and you know my unit, its a old First batch, ditnt handlet it whit care, but i still like it, and im everythime surprised how much you can doo, whit only 600 mhz and 256 mb ram..
So as long the Pyra performs better than 2 of these, i think i will be fine..
Compare to the smartphones these days, even the Pyra is a bit slow but compare to my Pandora, it will be something like a "NASA" Computer..
 
Also, i have a CC Pandora, which i can overclock to about 900 mhz, and its also quite good for my tasks, so even a dual core pyra at 1 ghz would be much faster, than the 900 mhz CC Pandora..
I hope to get decend Dreamcast and N64 Emulation, and the last one works for some games quite good on my Pandora (per exemple mario 64), but there are more demanding games like the zelda games, which ditnt work that good,
the dreamcast got some cool Truck Games so it will be the ideal handheld for me

Try the newest version of Mupen64Plus with a fresh appdata. I get surprisingly comfortable results at 800MHz (the highest my Pandora is stable at without increasing the OPP) out of almost every game in my library, except for the handful that don't get along with the default GPU plugin.
 
Try the newest version of Mupen64Plus with a fresh appdata. I get surprisingly comfortable results at 800MHz (the highest my Pandora is stable at without increasing the OPP) out of almost every game in my library, except for the handful that don't get along with the default GPU plugin.

Thanks, but i need a reason for my shiny new pyra :) Pandora can run N64, but its diverent if you drive whit a 1000 ccm Motorcycle whit 80 kmh over the German Autobahn, or whit my 125 ccm Motorscooter whit 80 kmh over the German Autobahn,
booth works, but the 1000 ccm Motorcycle drives the Speed much easea..
I will use my Pyra even for the less demanding tasks, which my Pandora also runs easy, like Atari 2600 or Gameboy, but due to the new CPU, the Pyra will last much longer for these, per example..

Someone said some months ago, "Underclocking will be for Pyra, same as Overclocking for Pandora"..
 
Yes, certainly.
One core running at 500MHz should already be a bit faster than the Pandora running at 1GHz. Clock rates can only be compared on the same generation of CPUs. :)

Even when they're both based on the same ArmV7A reference design? I guess cache configuration and memory speed would have the next most real-world impact, but something small that sits in the L1 cache and so only stresses the CPU and nothing else should behave pretty similarly I'd have thought.

It may be sufficient just to say that you can't compare clock rates between different SoC+RAM configurations though, to be more accurate.
 
ARMv7A is an instruction set, not an implementation. In the same sense that an Intel Celeron may implement the exact same instruction set as an Intel Xeon.

The effects go further than just cache/bus configuration, even if it helps.
 
Honestly, I'd rather have the issue fixed than rushed out the door with a potential hardware issue, it should run at designed speeds. I know people want it now, just I feel that people will enjoy the device more if it's running like intended.
 
Last edited:
Back
Top