Getting closer...

[JonnyH]

That looks like one to maybe evaluate, but judging by the mounting holes for a fan and the metal top to the SoC suggesting good thermal contact for a heatsink, that might run a bit hot. But maybe the Pyra's current cooling solution is better than I think it will be, and we can afford to use a hotter chip, we'll have to wait and see.

Nearly all SoCs can be tuned for a specific power use (and heat output) - they just clock lower. No need to create a new SoC for each power target when you can just wind the same silicon down for the same result.

This same point should be kept in mind when considering any lower performance core due to power use or thermal limits - you really have to tune both for the same power output and compare performance then. While the omap5 is getting rather old now, being at least 2 major process nodes behind, a more modern "high power" A73 will likely beat the pants off an A55 (of a similar silicon node) in anything but the most limited scenario...

That is to say, if the A55 is having to turbo the pants off itself to keep up, it may use more power than a larger core in the middle of it's performance curve.

from here, it seems to run with lots of watts:

A lot of that will likely be i/o - PCIe signalling is /expensive/, if nothing is connected, or can be disconnected for large amounts of time, you should be able to completely disable the phy and save pretty much everything. Likely other busses can be disabled in similar ways (Assuming, of course, that the SoC implementation allows that - I would assume so, as it's just wasted power otherwise, and not really difficult to gate whole units in a coarse fashion like that..)

A usb HDD can draw "up to" 900milliamps @5v (so 4.5w) itself, so most of the difference may not be the usb implementation in the SoC itself.

And the rather high "standby" power use is likely a software issue - that's higher than many x86 laptops idle at - sceen on and everything....

 

[WolfpackN64]

What about the RK3399

https://liliputing.com/2018/01/rock...ockchip-rk3399-coming-soon-for-60-and-up.html
https://forum.pine64.org/showthread.php?tid=5614

That one has a Mali GPU, which isn't ideal.

 

[levi]

That one has a Mali GPU, which isn't ideal.

What's your reasoning? Is it just because it needs a binary blob, as does the current PowerVR core in the OMAP5, or is is suboptimally performant in some way?

 

[Grench]

That one has a Mali GPU, which isn't ideal.

Name a GPU that is.

Then name one that is ideal (has Linux drivers), has good 3D performance and has open source drivers.

 

[WolfpackN64]

What's your reasoning? Is it just because it needs a binary blob, as does the current PowerVR core in the OMAP5, or is is suboptimally performant in some way?

Name a GPU that is.
Then name one that is ideal (has Linux drivers), has good 3D performance and has open source drivers.

Indeed, because Mali needs a binary blob. If you can have a GPU with a FOSS driver, go that route, and there are actually 2 options. Anything that has an Adreno GPU (Snapdragon) or a Vivante GPU (mostly i.MX). Both are well performing options.

 

[Yuji Sakai]

maybe future low-energy ryzen?

 

[JonnyH]

Indeed, because Mali needs a binary blob. If you can have a GPU with a FOSS driver, go that route, and there are actually 2 options. Anything that has an Adreno GPU (Snapdragon) or a Vivante GPU (mostly i.MX). Both are well performing options.

If there's *anything* even close from AMD/Intel/Broadcom to the power envelope possible here I would strongly prefer those.

I really dislike giving companies dollars when they haven't earned them - even if they have a "good" open source driver, I feel they're taking advantage of the community who are putting all the work in to reverse engineer and work on said drivers.

I feel it's a bad precedent, saying "No need to bother helping with OSS, as some volunteers would put in their time anyway", and giving them money due to having a community-supported driver feels like encouraging bad behavior.

 

[kaprikawn]

If there's *anything* even close from AMD/Intel/Broadcom to the power envelope possible here I would strongly prefer those.

I really dislike giving companies dollars when they haven't earned them - even if they have a "good" open source driver, I feel they're taking advantage of the community who are putting all the work in to reverse engineer and work on said drivers.

I feel it's a bad precedent, saying "No need to bother helping with OSS, as some volunteers would put in their time anyway", and giving them money due to having a community-supported driver feels like encouraging bad behavior.

That sounds a lot like you're willing to cut off your nose to spite your face. Options are already limited, beggars can't be choosers. If there's a FOSS driver available, even if it's built off the hard work of the community, I doubt a project of this size is going to sway the thinking of a big corporation.

 

[WolfpackN64]

If there's *anything* even close from AMD/Intel/Broadcom to the power envelope possible here I would strongly prefer those.

I really dislike giving companies dollars when they haven't earned them - even if they have a "good" open source driver, I feel they're taking advantage of the community who are putting all the work in to reverse engineer and work on said drivers.

I feel it's a bad precedent, saying "No need to bother helping with OSS, as some volunteers would put in their time anyway", and giving them money due to having a community-supported driver feels like encouraging bad behavior.

Broadcom would be a good choice, especially once SoC's with VideoCore V become more widespread, but they're not easy to get I've heard.

 

[Djoga'Ro]

beggars can't be choosers

It's gratis SoCs now?

 

[Letalis Sonus]

Broadcom would be a good choice

Are we talking about the same "Let's slap the same crappy ancient peripherals on every chip!" Broadcom? The Pi's chip series is nothing but a cashcow to them and even those ones are everything but available for purchase. Their SoC core business are semi-custom industrial chips, unless you're going to throw a bunch of money into their faces to get your own chip you won't get anywhere.

 

[JonnyH]

Are we talking about the same "Let's slap the same crappy ancient peripherals on every chip!" Broadcom? The Pi's chip series is nothing but a cashcow to them and even those ones are everything but available for purchase. Their SoC core business are semi-custom industrial chips, unless you're going to throw a bunch of money into their faces to get your own chip you won't get anywhere.

From what I remember they actually pretty much shut down all their SoC development teams - I guess the pi is just enough to keep them limping along doing "least effort" refreshes with what IP they already have lying around (or pretty much complete but not yet in product when they did that), or can get off the shelf from someone else. I certainly wouldn't expect anything "new" from them.

That sounds a lot like you're willing to cut off your nose to spite your face. Options are already limited, beggars can't be choosers. If there's a FOSS driver available, even if it's built off the hard work of the community, I doubt a project of this size is going to sway the thinking of a big corporation.

I guess it kinda depends on what you want out of a pyra. You want a more powerful device? They're available elsewhere now. You want a lower cost device? They're available elsewhere now. For me the real differentiator is the "openness" of the platform and community.

 

[Yori]

The Pyra is not even out the door and we are already talking about a new cpu!?!

I'm hoping there will not be a new cpu board soon, I don't want my brand new Pyra becoming obsolete in a few months.

And I'm hoping if we do get a new cpu soon it won't fracture the software & toss out Pyra hardware optimization out the window due to having two different types.
We already are going to have to account that some Pyras will only have 2GB memory insted of 4GB when making software.

 

[Letalis Sonus]

I guess the pi is just enough to keep them limping along doing "least effort" refreshes with what IP they already have lying around (or pretty much complete but not yet in product when they did that), or can get off the shelf from someone else. I certainly wouldn't expect anything "new" from them.

They are still doing new SoCs for automotive ECUs - but not the kind you'll read about in any media.

 

[rygD]

The Pyra is not even out the door and we are already talking about a new cpu!?!

I'm hoping there will not be a new cpu board soon, I don't want my brand new Pyra becoming obsolete in a few months.

And I'm hoping if we do get a new cpu soon it won't fracture the software & toss out Pyra hardware optimization out the window due to having two different types.
We already are going to have to account that some Pyras will only have 2GB memory insted of 4GB when making software.

My guess is ED feels the same way. Additionally, there is a little less rush (and work) to make a new CPU board, so I expect it will take a while. Even once we get an upgrade I like to I imagine there will be a device that I can put the old one in. Maybe we can do a KS if there are enough sales and interest to get something from @hns. If he and his company are willing to do it I will probably be we rolling to buy it.

 

[matzesu]

One Big Advance of the Pyra over other Devices is, that you dont need to buy a whole new Unit if there is a new SOC, just order the New SOC, put your OLD SOC in the Anti Static Bag, where the new one came whit, and put the new one in..
Mybe a Carry Case whit a Spot for CPU Board would be a great Addition to the Dragonbox Shop..

Im also Hoping the Software Devs make the Software that runs on the Omap also for the Omap,

I dont want to Change my SOC every Time i change the Software i wants to run..

Like: The SNES Emulator needs the Omap Board, and the NES Emulator needs a diverent board..

 

[PCXT]

Dependent on definitions of 'very thin', if that were true my Pandora along with a number of other electronic devices I own ought to be little puddles of white hot plastic by now. I suspect over short timescales this is a valid assumption, but it doesn't come close to holding as deltas increase and longer time periods are considered.

I'm talking about making these parts like "pipes": With one "input" transfer condition on one part (usually elements faces) and "output" on another, everything else is not modelled at all. The problem I found is that modelling the other conductivity makes the conditions a bit more ideal than they really are and in many cases not including the flow there nicely compensates presence of heat from other parts of the system.
But this is for metallurgy, not plastics. This works if you have a 1200*C-hot workpiece near your 400*C-hot complex, multi-part die and you really don't feel like writing radiation exchange to FE package .

 

[elvissteinjr]

The Pyra is not even out the door and we are already talking about a new cpu!?!

I hope it's obvious enough how long developing the current CPU board took (actually is still taking) a long time. As much as you don't want your Pyra to be not outdated for as long as possible, the reality is that we're already on a pretty outdated SoC to begin with.
With this thread you can also see that it's worth having SoC candidates discussed within the community. Side-grades are probably not worth developing.

I guess it kinda depends on what you want out of a pyra. You want a more powerful device? They're available elsewhere now. You want a lower cost device? They're available elsewhere now. For me the real differentiator is the "openness" of the platform and community.

"Portability without sacrificing capability" was the slogan of the Pandora. The more powerful devices out there lack other capabilities or half-ass them.
I value the community and the openness of the platform as a user. I don't see a closed-source GPU as a huge burden, personally, as long as it works.
One does not necessarily exclude the other. We can still aim to have a powerful device to the extent possible.

If I only wanted community and open platform, I'd already have that with my Pandora.

 

[EvilDragon]

Yep, we don't even HAVE a deal with any SoC manufacturer right now, so any CPU board is still FAAAAAR away, no worries about that.

That said, RockChip can't really be used (horrible Linux support), AllWinner isn't really much better than the OMAP5 (if it even is).

So yeah... nothing serious at the moment.

Some quick update:
My contact in Greece did have the time to check on FormAction and the moulds yesterday. He inspected them thoroughly and all discussed changes have been implemented.
They have a production run planned next week and he hopes to find the time to pay them a visit on Monday and send me some pictures.

 

[rygD]

Then after you do a quick check once you get them, a clear run, right?