The Pyra is crying!


How much RAM are you going with? I don't want to be crying for RAM too. :p
At the very least it looks like U-Boot was configured for 2GB since that's what it said it expected. AFAIK this is what ED said was probably going to be used as of a while ago.

small question, what does the:

Hack! Disabling faulty EMIF1
message mean??
EMIF1 = "External Memory Interface 1" and it's disabling it because it couldn't find memory there.
 
Last edited by a moderator:
nice! so, if I understand correctly the boot-up test was the cpu board plugged in the mainboard, booting from an SD card? (so everything working minus the RAM)
I don't think so.  That's just the debug output from the processor.  It's got as far as starting uboot, which is nice, but it hasn't got to booting anything from an SD card - it hasn't got any RAM to load a kernel into.  Uboot's probably XIP (eXecute In Place) meaning it can print out that debug information just using the program counter and a few other registers.
The MLO (Boot loader) and U-Boot actually load from the SD Card, and it also got all it's power from the mainboard.

So yes, the mainboard and CPU Board are working together here.
 
gratz everyone working so hard to make this come togather and a big thank you for it
 
Uboot's probably XIP (eXecute In Place) meaning it can print out that debug information just using the program counter and a few other registers.
The OMAP has a small amount of RAM in the chip itself, something like 64K I seem to recall (though that may have been the OMAP3, the OMAP5 may be more) into which it loads the uboot. It's gotta have some RAM for any calculations it needs to do about RAM and external storage, notably the location on that storage of the kernel.
 
How much RAM are you going with? I don't want to be crying for RAM too. :p
At the very least it looks like U-Boot was configured for 2GB since that's what it said it expected. AFAIK this is what ED said was probably going to be used as of a while ago.
I totally missed the end of that line.  Thanks for pointing it out.  I guess I was still hopeful for more "just in case".
 
The OMAP has a small amount of RAM in the chip itself, something like 64K I seem to recall (though that may have been the OMAP3, the OMAP5 may be more) into which it loads the uboot. It's gotta have some RAM for any calculations it needs to do about RAM and external storage, notably the location on that storage of the kernel.
And if it lets you lock L2 cache that's a pretty generous extra 2MB. It might even be possible to do so without locking if you're really careful.

You don't really need RAM to do a bootloader, especially when you consider they will often do a complete RAM check by writing and reading every location. But as this is a second stage bootloader (or is it a third stage?) so U-Boot would have needed to be loaded into some kind of RAM in the first place.
 
Last edited by a moderator:
Wow, how did Nikolaus solder those small traces? Did he use special tools, or did he manage all that with an ordinary soldering needle and wire cutter / knife ?
 
One thing no one's mentioned so far is that the fully populated PCB also has eMMC chip P/N H26M31001HPR, which has a 4GB capacity (see here: <a data-ipb="nomediaparse" data-cke-saved-href="https://www.skhynix.com/inc/pdfDownload.jsp?path=/datasheet/Databook/Databook_Q3"href="https://www.skhynix.com/inc/pdfDownload.jsp?path=/datasheet/Databook/Databook_Q3" 2015_nand.pdf)"="">https://www.skhynix.com/inc/pdfDownload.jsp?path=/datasheet/Databook/Databook_Q3'2015_NAND.pdf)

ED, can you comment on if this is the chip that will be used for Pyra at release or if it's just a placeholder for testing? And has anything been done to validate it, or is that impossible w/o the RAM?

Also, I think this is the RAM chip: http://www.intelligentmemory.com/fileadmin/pdf-datasheets/IM4G08-16D3FABG1.2-0821.pdf

512MB DDR3L so yes, 2GB of RAM total.
 
Last edited by a moderator:
One thing no one's mentioned so far is that the fully populated PCB also has eMMC chip P/N H26M31001HPR, which has a 4GB capacity

ED, can you comment on if this is the chip that will be used for Pyra at release or if it's just a placeholder for testing? And has anything been done to validate it, or is that impossible w/o the RAM?
That's just a placeholder. We'll use whatever chip makes most sense (price / capacity) when it's time for the mass production.

At least 8GB, but probably 16GB, I guess.
 
Awesome! Great news! I am checking for a new Pyra post almost every day. Each of them is a small treat for us Pandora/Pyra fanatics. :-D

It seems that the progres is being made at a rapid pace! So I'm happy to hear it.

Forgive me my ignorance, but didn't OMAP itself include some RAM? I was under the impression that the Pandora RAM was decided from the OMAP itself...
Unlike the OMAP3 the OMAP5 doesn't have a PoP memory that resides on top of the chip. The SDRAM and eMMC memory are separate.
Does this mean that one can add more RAM if one wants to by manually soldering bigger RAM chip like Nikolaus did here?  
 
Awesome! Great news! I am checking for a new Pyra post almost every day. Each of them is a small treat for us Pandora/Pyra fanatics. :-D


It seems that the progres is being made at a rapid pace! So I'm happy to hear it.


Forgive me my ignorance, but didn't OMAP itself include some RAM? I was under the impression that the Pandora RAM was decided from the OMAP itself...
Unlike the OMAP3 the OMAP5 doesn't have a PoP memory that resides on top of the chip. The SDRAM and eMMC memory are separate.
Does this mean that one can add more RAM if one wants to by manually soldering bigger RAM chip like Nikolaus did here?
As long this guy who wants to change the ram, have the shematics to know how to wire it up, and the right soldermask to avoid the new ram from floding around..
I think its easear than, to wait for a future CPU Board..


Also i ditnt think more ram is nessecary, because in the Pandora was 512 mb in the best case and the Pyra maybe have 2 gb also 4 times more..
 
Does this mean that one can add more RAM if one wants to by manually soldering bigger RAM chip like Nikolaus did here?
It's BGA so this would be very hard to do without the proper equipment, or at least that stencil mask that ED referred to. Getting the old ones off is probably not trivial either.

There are actually 1GB RAM chips by the same manufacturer that can probably be used as a replacement. The downside is it increases power consumption a lot (and power consumption is already on the high side since it's not a low-power DDR technology)
 
Again, great stuff ED.
I remember the feeling I had when I made software for a hardware prototype, to plug it in and see it come to life: It's magical.
I hope it made you feel the same way.
 
Back
Top