We've started to order things!


Has ED ever indicated how much a replacement CPU/RAM/emmc board would cost say given the current chipset as an example?

I know it would depend on future chipsets chosen, but I guess an indication using the cost of the current SOC , RAM and emmc would be good. My guess would be around $150 - $200 given the over $50 cost just for the Omap 5 SOC. OF course then there is the RAM, the emmc , the board coast and the population , contingency etc. Plus then add wholesale and retail markups.
 
I think I recall ED predicting future SoC builds could cost less than 100EUR to the end user.

Regarding ED's cooling solution, I'm a little confused by 'the head pad that connects the SoC to the PCB'. I'd have thought the SoC would be soldered to the CPU board PCB, and a sticky heat pad would be required on top of it to connect to the aluminum plate that would connect to the SD card cages and make them part of the heat sink, and potentially also radiators to the outside of the case.

Edit: I think this is the source of ED's estimate: https://pyra-handheld.com/boards/threads/snapdragon-soc-410e-and-600e.78238/#post-1394120
 
Last edited:
I think I recall ED predicting future SoC builds could cost less than 100EUR to the end user.

Regarding ED's cooling solution, I'm a little confused by 'the head pad that connects the SoC to the PCB'. I'd have thought the SoC would be soldered to the CPU board PCB, and a sticky heat pad would be required on top of it to connect to the aluminum plate that would connect to the SD card cages and make them part of the heat sink, and potentially also radiators to the outside of the case.

Edit: I think this is the source of ED's estimate: https://pyra-handheld.com/boards/threads/snapdragon-soc-410e-and-600e.78238/#post-1394120

To be fair, that was for a cheaper lower power SOC, and even then it was a big if - ie if it the SOC was cheap
 
Regarding ED's cooling solution, I'm a little confused by 'the head pad that connects the SoC to the PCB'. I'd have thought the SoC would be soldered to the CPU board PCB, and a sticky heat pad would be required on top of it to connect to the aluminum plate that would connect to the SD card cages and make them part of the heat sink, and potentially also radiators to the outside of the case.

The SoC gets sandwiched between the CPU board PCB and the Main board PCB - the heat pad in question would transfer heat to the exposed copper on the Main board PCB, effectively turning the Main board itself into a heat-sink.

I'm having trouble picturing where the aluminium plate fits in though.

-Neelix
 
The SoC gets sandwiched between the CPU board PCB and the Main board PCB - the heat pad in question would transfer heat to the exposed copper on the Main board PCB, effectively turning the Main board itself into a heat-sink.

I'm having trouble picturing where the aluminium plate fits in though.

-Neelix

I believe that the aluminum plate would go on the back side of the SoC between the SoC and the case plastics. So, from the desk up:
Battery door, battery, lower case plastics (including stylus holder & stylus), aluminum plate, SoC, thermal pad, main/peripheral board, key mat, keyboard surround, screen bezel, screen, screen back light board, plastic lid illumination piece, screen outer lid, glue, aluminum logo plate.
 
Note that the second 4GB of memory technically resides at 0x200000000, although using 0x300000000 like igep does works too since the excess row bit is ignored by the memory hence the two regions are aliases.

Hey cool good to hear you figured it out! The aliases should be just fine, I think the only advantage for not using the aliases
would be with IO-coherent C-A15 hardware like mach-keystone. But that requires extra hoops for the addressing like
keystone does, and 5432 never was wired to be IO-coherent AFAIK.

I've also uncovered the mystery of the stolen memory at 0xFF000000

OK good to hear we could reclaim that, it's after all 16MB which is quite a bit of memory. The cpu1 startup address pointing
there worries me a bit though if there's some save/restore code relocated there.. I'll try to verify that. So maybe let's reclaim it a
bit later after things really are verified to be working. We don't have much PM yet happening for 5432 in mainline kernel.
 
The aliases should be just fine, I think the only advantage for not using the aliases would be with IO-coherent C-A15 hardware like mach-keystone. But that requires extra hoops for the addressing like keystone does, and 5432 never was wired to be IO-coherent AFAIK.
Thank god it isn't, the cortex-A15 was ARM's first processor supporting a coherent interconnect and it has several "category A" bugs related to it. (e.g. #781819: "If the above conditions occur, the A15 will deadlock." .. "There is no workaround in a coherent ACE system.")

The cpu1 startup address pointing there worries me a bit though if there's some save/restore code relocated there..
That just can't be right. With LISA setup like this, any access to (phys) 0x00FF000000-0x00FFFFFFFF would result in a bus error.
 
It's been 2 weeks since the last news post...

66511846.jpg

:D
 
I believe that the aluminum plate would go on the back side of the SoC between the SoC and the case plastics. So, from the desk up:
Battery door, battery, lower case plastics (including stylus holder & stylus), aluminum plate, SoC, thermal pad, main/peripheral board, key mat, keyboard surround, screen bezel, screen, screen back light board, plastic lid illumination piece, screen outer lid, glue, aluminum logo plate.
What about the air?
 
if you've read the posts lately you would have seen they may have figured out the 4gb ram issue. to me that's a new news post
This is true, but things like that can be easily if one isn't regularly scouring the threads.

A picture tells a thousand words though ;-)
 
Jep, some news would be nice. Even there are not so many news to tell. The expectations of many people including me, are ubiquitously. I thought: November!
Ok, i dont want to be mean, I knew it would be taking a bit longer for sure. Some more updates are just making the time for waiting a bit more pleasant. :)
 
Last edited:
Back
Top