Getting closer...


EvilDragon

Administrator
Staff member
Joined
Mar 4, 2003
Messages
29,986
Age
47
Location
Ingolstadt
Update time!

So, I guess I got mostly good news for you this time.
It took a while until I could post updates, but that was mostly thanks to the christmas time. Most companies didn't start resuming work before January 8th - and FormAction didn't even start until January 15th.

Oh well, at least now we're ALMOST back to normal (except for Chinese New Year coming up).

Speaking of FormAction, let's start with the case!

According to them, all changes in the molds have been done and they already produced new samples, including the sealing for the speakers.
Well, that's what they claimed last week.
I'll check back with my contact if he could already check them within the next few days.
I will not trust them until I receive them... but basically, they should have time until February, and you can be sure I'll push them!


Let's continue with the CPU Boards...

If you followed the other thread on the boards, you probably know that the new CPU boards have been ordered.
However... and that's the bad news: Chinese New Year is approaching, and as the bare boards are being produced in China, they won't show up until after that.

Well, not much we can do here.
Good thing is that the simulation shows the PCBs should now work fine with the 4GB RAM, so we only need to wait for them to appear now :)


... on to the mainboards...

The mainboard with the fixes has been tested and so far everything seems to work as it should!
Also, Wifi is now A LOT better. In our testing environment, we had six different Wifi networks with different distances and various strengthes.
Whereas the old mainboard could only find two of them, the new mainboard instantly did find all six at once, and the signal quality was a lot better as well.

We still have to do more tests regarding download speed and reliability, but it's really looking like it's fixed now as well.

That's that.

The only thing left now is getting rid of the sirring noise.
That's mostly trial and error... try as many different capacitors as possible until we find a combination that works.
Sometimes it's even enough simply using a capacitor from a different manufacturer... some might remember the whirring noise some of the 1st Edition Pandoras made. That probably was the same :)

This trying-out doesn't need a layout change and therefore can happen while the new bare PCBs are being produced.

So yes.
All known bugs and issues have been dealt with. We're not aware of any issue left right now!

The mainboards should now ALL be 100% correct. It's time to wait for them to arrive... so we can build one more round of prototypes (which will finally go to the prototype orders!)... and then we can setup the mass production.


... tweaking the cooling...

In the meantime, we're working to improve the cooling even more. The current situation lowers the CPU temperature by 20°C (according to the simulation), but we're trying to improve that even more while we're waiting for the new boards.
Thanks to the simulation software Juliano uses, we can easily try out different scenarios (connecting the Al plate to the OMAP as well, using one or two heatsinks, etc.) and find the best one possible.
It's both interesting and fun! Those simulations are simply awesome. They take in account the different conductivities of the materials as well as the thickness and the OMAP temperature!


... a peek into the future...

Well, next step is definitely the planned prototype production, hopefully now with fully working 4GB RAM Pyras with all the bugs squashed.
The next step beyond is the mass production. Global Components is ready (and have bought a lot more new machines that help improving the production quality and testing), and the parts are ready as well.

Similar to the first months of Pandoras life, the software needs a lot of tweaking as well... but the more people receive a Pyra, the faster any bugs or annyoances are found and improved.

And then?
Well... there's a possible SoC being released which could become the first CPU Board Upgrade:
The i.MX8 M QuadLite, which is the mobile version of the i.MX8.

You can find more information about it here:
https://www.nxp.com/products/proces...cortex-a53-cortex-m4-audio-voice-video:i.MX8M

It's being released 2018, it's freely available to anyone and while the raw processing power is not THAT much higher compared to the OMAP5, the power consumption is significantly lower, it's 64bit, and the 3D chip (GC7000L) has OpenSource drivers available.
It's most probably the SoC the Librem 5 Smartphone will use as well, so there would be two devices with the same SoC running Linux, which would be great :)

Anyways, that's still the future. The SoC is not yet out and we'll concentrate finishing and delivering the Pyra!
But rest assured: We won't stop when delivery has started... we will continue our journey so the Pyra will receive nice updates in the future.

That's it for today :)

So, mostly good news - except for the probable delay of the bare PCBs because of Chinese New Year.
Everything else should be done and we're already thinking about the future.

Exciting times are up ahead!
 
Be careful about hyping up the second SoC too much; I'm already considering holding out for it.

It really depends on how long it takes me to save up the money versus how long it takes you to release the second CPU board. I know we both like to take our time in those respective departments.

Edit: I just realized I may have encouraged you to take even more time in delivering the second CPU board. Oops. >_>
 
Last edited:
I'm always eager to hear Pyra news. Thanks for the update!

I tried to collect everything into one remaining "todo" list. I'm not sure if this list is entirely correct; please point out if I missed anything or if something below is wrong:

Cases
- wait for samples of the new cases to arrive
- test cases to make sure they are final
- when final, setup production at the new company and order a production batch of cases

CPU boards
- wait for the new CPU board prototypes to arrive
- test CPU boards to ensure they are correct
- when final, order a production batch of CPU boards

Mainboards
- finish testing main board wifi
- try to improve the cooling via simulation
- try to fix the stirring sound via trial and error
- send the mainboards out to the prototype orders for further testing
- give testers time to tinker and wait for any new bug reports (?)
- when final, order a production batch of mainboards

Final production
- get the hardware certified
- wait for final cases, mainboards, CPU boards to arrive
- write the manual
- print the manual
- design the packaging
- print the packaging
- assemble everything
- ship em!

Did I miss anything?
 
Last edited:
I hope you find out if this new CPU is susceptible to the spectre bug before you put in a big order or commit too much time to ordering it. I'm not convinced the linux kernel can software patch to avoid this bug in its entirity, and the CPU caches will need to have some more smarts factored into them. Or if it can patch them, it'll be horribly slow to access RAM. Best off having the flaw eliminated at the hardware level.

I have an iMX6 SBC and I found the heat the CPU produced running simple python code to be rather excessive until I stuck a little heatsink on top of it. That's three generations behind the chip you're looking at, so hopefully that won't be a problem.

Fingers crossed the voltage simulations are good enough to predict successful prototype boards at 4GB, and that the 'sirring' can be eliminated or at least significantly reduced. What does the capacitors do in this instance; if they're driving the display backlight, the fact their noise is in audible frequencies might suggest the display will also be broadcasting audible radio noise that can be picked up by people with hearing aids using the telecoil setting (as we just found out about in another thread recently). Not a reason to revise the boards I suspect, but it's worth knowing nevertheless.

But good news all told. I've no doubt there will be more problems exposed, but that's what a prototype round is for.
 
Unfortunately, while the A53 on the i.mx8 will be useful with supporting armv8 and aarch64, and a "good" open source GPU driver a massive plus, the end performance of an A15 will likely be quite a bit higher at the same clocks. Same with the GPU, which is nice and fancy and supports GLES3.1, the performance at this point is relatively unknown (and previous vivante cores have often not been powerhouses).
 
I have an iMX6 SBC and I found the heat the CPU produced running simple python code to be rather excessive until I stuck a little heatsink on top of it. That's three generations behind the chip you're looking at, so hopefully that won't be a problem.

That's the non-mobile version which means it needs a lot more power and produces a lot more heat.
The normal iMX8 even needs fan :)

Fingers crossed the voltage simulations are good enough to predict successful prototype boards at 4GB,

Well, it's a common practice these days when developing such systems, so it should be fine :)

and that the 'sirring' can be eliminated or at least significantly reduced. What does the capacitors do in this instance; if they're driving the display backlight, the fact their noise is in audible frequencies might suggest the display will also be broadcasting audible radio noise that can be picked up by people with hearing aids using the telecoil setting (as we just found out about in another thread recently). Not a reason to revise the boards I suspect, but it's worth knowing nevertheless.

They're used for powering the USB port and the sirring noise results from the capacitors vibrating at high frequencies. The PCB starts to vibrate as well and this is actually the frequency you can hear.
There are no radio broadcasts.

But good news all told. I've no doubt there will be more problems exposed, but that's what a prototype round is for.

Well, there's not really much left :)
[doublepost=1516680235,1516679975][/doublepost]
Unfortunately, while the A53 on the i.mx8 will be useful with supporting armv8 and aarch64, and a "good" open source GPU driver a massive plus, the end performance of an A15 will likely be quite a bit higher at the same clocks. Same with the GPU, which is nice and fancy and supports GLES3.1, the performance at this point is relatively unknown (and previous vivante cores have often not been powerhouses).

True, though the iMX8 has four cores and the OMAP5 gas two. The most interesting thing though is not the CPU power, in my opinion, but more the reduced power consumption and heat production due to it being of newer technology :)

But let's not focus too much on that - first, we'll release the Pyra. Then we start actively looking for upgrades. And of course we'll test any candidate first with a devboard.
 
So if you do use that for a CPU board upgrade you might as well upgrade to a 4K display. :p
 
  • Like
Reactions: rSl
So does this mean Q2/Q3 2018 is a very real possibility for mass production? Maybe I'll have more money by then :p
 
I'd say Q3 2018 is about the earliest we'll see units in the hands of the general public is the earliest possible estimate at the moment. Maybe I'm being overly pessimistic though.

@rygD Yeah, because what you really want are pixels so small they can create false colours in the same way the pits on a CD or DVD put out a spectrum when reflecting light.
 
The i.MX8 M definitely looks interesting, but I think you're right to suggest "first let's release the Pyra". By the time the Pyra finally comes out, maybe a more attractive SoC might have been announced. I know that I would not be very interested in an "upgrade" board that didn't really net us any new gains in emulation and game performance, and others may feel that way as well. Something to keep in mind, EvilDragon.

-God Ginrai
 
@rygD Yeah, because what you really want are pixels so small they can create false colours in the same way the pits on a CD or DVD put out a spectrum when reflecting light.
"Because I can!". Not serious, btw.

Even once we have an upgrade option I will probably stick with the OMAP until I have something to do with the old CPU board, or if everyone switches to the new one and we no longer get software updates.
 
  • Like
Reactions: rSl
Hmm, even if the Pyra ends up releasing before or during Summer, I'll probably just end up waiting. The GPD Win 2 just took a huge chunk out of my bank account and I got some...things I need to finish paying off. So, earliest I could pay for a Pyra would MAYBE be Q4, but I think I'll just wait for the first CPU board upgrade, even if that ends up being next year or the year after.
 
Sounds pretty good, (unless the chinese new year, but this was to expect as its every year)..
last 2 Months (tm) and then some weeks hopefully..

Mybe i can get my Unit just in time for the Unimog Gathering in June to make a 3 Days Trial to proofen its Unimog Proofness :)
 
  • Like
Reactions: rSl
Out of curiosity I had started looking at what Arm chips were available that were likely be acquirable when ED had last mentioned the cpu board upgrade.
The imx8 was one of the ones to stand out as a possibility. There's enough projects out there that use the imx6, that makes me hopeful that there will be a good community around the imx8.

I'm really excited by the possibility of the Pyra shipping this year.
 
I hope you find out if this new CPU is susceptible to the spectre bug before you put in a big order or commit too much time to ordering it. I'm not convinced the linux kernel can software patch to avoid this bug in its entirity, and the CPU caches will need to have some more smarts factored into them. Or if it can patch them, it'll be horribly slow to access RAM. Best off having the flaw eliminated at the hardware level.

AFAIK Cortex-A8 is susceptible to Spectre but Cortex-A53 is not, so an i.MX8M should actually be better off than the OMAP5.

That said, retpoline is already merged in Linux 4.15-rc9 so with the final 4.15 release Spectre variant 2 should be fixed for everything except Intel Skylake and newer.

Variant 1 is yet to be fixed at the kernel level but GCC patches have been floating around for a while and Mozilla manually patched Firefox against it (to my understanding both the core browser stuff and JavaScript).
 
Any updates from the developers who already have a board/prototype?
What are they doing?
 
Back
Top