thank you very much to the anonymous person who pointed me at a *FULL* Rockchip RK3399 Reference Design including Technical Reference Manuals, full schematic file and full PCB CAD design files.
using that design it will be possible to create an amazing CPU Card... unfortunately running MALI.
http://rockchip.wikidot.com/rk3399
this shows that it has USB3 however it does not have RGB/TTL, so it will be necessary to do a conversion from MIPI to RGB/TTL with an SSD2828 or similar.
the Reference Design has twin 32-bit-wide LPDDR3 ICs so there is the possibility of getting not just decent performance but also low power at the same time.
l.
2018-01-17 11:28 GMT+01:00 Luke Kenneth Casson Leighton lkcl@lkcl.net:
this shows that it has USB3 however it does not have RGB/TTL, so it will be necessary to do a conversion from MIPI to RGB/TTL with an SSD2828 or similar.
Could you design an EOMA68+ with a different pinout?
For example, the EOMA68 for less powerful devices, and an EOMA68+ for more powerful devices. For example, the EOMA68+ would have support for MIPI (instead of RGB), mandatory use of USB 3.0, etc.
For example, I think that an Allwiner A20 or an Ingenic SoC does not make sense on a FullHD screen, and an Rk3399 does not make much sense on a 320x240 screen.
--- crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68
On Wed, Jan 17, 2018 at 12:08 PM, Miguel Garcia gacuest@gmail.com wrote:
2018-01-17 11:28 GMT+01:00 Luke Kenneth Casson Leighton lkcl@lkcl.net:
this shows that it has USB3 however it does not have RGB/TTL, so it will be necessary to do a conversion from MIPI to RGB/TTL with an SSD2828 or similar.
Could you design an EOMA68+ with a different pinout?
the absolute absolute top priority is that there be absolutely NO chance - whatsoever - of confusion in the eyes of a hundred MILLION and above totally non-technical end-users.
if there is any chance that two Cards with different pinouts could be plugged into the same socket
if there is any chance that the owner of two Housings does not know if a Card is safe to plug in
then the answer to the question you ask is NO.
if however there is a way that the exact same connector could be used... *WITHOUT* there being a SHADOW OF DOUBT... then yes.
what would you suggest, that could fit within that absolutely critical constraint?
For example, the EOMA68 for less powerful devices, and an EOMA68+ for more powerful devices. For example, the EOMA68+ would have support for MIPI (instead of RGB), mandatory use of USB 3.0, etc.
For example, I think that an Allwiner A20 or an Ingenic SoC does not make sense on a FullHD screen, and an Rk3399 does not make much sense on a 320x240 screen.
that's not.... within the "right" of EOMA68 to say to people "You Cannot Use An RK3399 Card for purpose X". should it be dictated that users MUST not plug Cards in just because we *happen* to think that there *might* not be *some* perfectly legitimate use-case which *right now* we cannot envisage?
that would result in confusion in the standard, wouldn't it?
l.
On 01/17/2018 05:28 AM, Luke Kenneth Casson Leighton wrote:
thank you very much to the anonymous person who pointed me at a *FULL* Rockchip RK3399 Reference Design including Technical Reference Manuals, full schematic file and full PCB CAD design files.
using that design it will be possible to create an amazing CPU Card... unfortunately running MALI.
Is it possible to remove the mali from the configuration or to use the rk3399 in a libre fashion?
I am just curious. And if not, can you make it work?
On Wed, Jan 17, 2018 at 7:49 PM, zap calmstorm@posteo.de wrote:
Is it possible to remove the mali from the configuration
of course, it's just a co-processor on the same internal shared memory bus - don't enable it, don't talk to it, don't compile the kernel config, don't run the proprietary userspace library
or to use the rk3399 in a libre fashion?
probably not, it's one of the high-end T-blahblah ones
I am just curious. And if not, can you make it work?
with probably about 18 months of full-time reverse-engineering, probably. but i've f*****g had it with ARM, the blatantly unethical f***s, trying to bribe people to shut down projects, and blackmailing companies to not pay software libre developers.
so f*** 'em.
l.
with probably about 18 months of full-time reverse-engineering, probably. but i've f*****g had it with ARM, the blatantly unethical f***s, trying to bribe people to shut down projects, and blackmailing companies to not pay software libre developers.
so f*** 'em.
I don't disagree, I just wondered if it was possible in the near future. Since its not, yeah...
I guess we wait for something better. To arrive. such as Shakti processors right? :)
l.
arm-netbook mailing list arm-netbook@lists.phcomp.co.uk http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook Send large attachments to arm-netbook@files.phcomp.co.uk
--- crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68
On Wed, Jan 17, 2018 at 10:25 PM, zap calmstorm@posteo.de wrote:
with probably about 18 months of full-time reverse-engineering, probably. but i've f*****g had it with ARM, the blatantly unethical f***s, trying to bribe people to shut down projects, and blackmailing companies to not pay software libre developers.
so f*** 'em.
I don't disagree, I just wondered if it was possible in the near future. Since its not, yeah...
if i start now it'll be about... 3 months and it'll cost about $2000 per set of 10 boards to test.
I guess we wait for something better. To arrive. such as Shakti processors right? :)
18 months on that... but yeah.
talking to jeff who did nyuzi, to see what it would take to at least get 25% the performance of MALI 400. nyuzi is a software-driven general-purpose compute engine that "happens to be reasonably good at 3D". based on some research work by intel.
l.
arm-netbook@lists.phcomp.co.uk
-
Luke Kenneth Casson Leighton -
Miguel Garcia -
zap