Beautiful landscapes are prettier than portraits

What are your feelings about that development?

  • I want the FullHD display, no matter what! It'll look so much better! I don't care about the perform

    Votes: 14 4.0%
  • FullHD would've been nice, but with those issues, HalfHD is better. I still would've preferred FullH

    Votes: 72 20.8%
  • FullHD is overkill anyways in my opinion. HalfHD is way better suited for a screen that size.

    Votes: 139 40.2%
  • CPU performance and easy development are more important than the higher resolution. I'm fine with th

    Votes: 208 60.1%
  • I don't really care about the difference between FullHD or HalfHD - it's more important that the gam

    Votes: 116 33.5%

  • Total voters
    346

The only measurement so far is about 46% CPU time for rotating 1080p in software (don't know if this was at 1.7GHz or what).
That was not rotation, just plain update of 1080x1920 @60Hz @1.5GHz.

The biggest issue with any other solution is going through the headache of getting someone to implement it. But so long as the SSL chip can be bypassed that can still be done at the discretion of the poor coder who demands it. It's not something that the Pyra will have to get right at launch.
If we find a solution for the TILER, that actually bypasses the need for even having the chip.
Not really, TILER needs that padded framebuffer format which is still a burden on the programmer, also incompatible with what pandora uses so backward compatibility would be harder. If it turns out that internal format has to be used, it would be actually quite hard to use it.
 
Last edited by a moderator:
As mipi seems to be a pretty new standard I would think that those new screens are mostly portrait for Tablets or smartphones of that size.


wouldn't it be an option to use lvds withnormal cables and instead of ribbon cables like the adafruit one does?
 
Are you guys sure that the nvidia shield uses a phone screen?

If you check the following picture, it seems to me that the ribbon is on the lower side of the lcd panel (above the fan), doesnt that indicate that the rendering is landscape natively?

shield-up.jpg


Because all phone screens i have seen had the ribbons on the side, instead of the bottom like the pandora and the shield.
 
Last edited by a moderator:
It isnt the screen connection you are seeing there, only the plane that the nubs are on, the screen is behind that on another set of hinges.
 
Last edited by a moderator:
You sure?, i mean the area above the fan, the fan is on the main pcb, and the main pcb seems to be ribboned to the screen.

I dont mean the thin long ribbon, but the wide short one slighty above the fan.

btw, is it possible to find tech spec manual/sheets whatever about the shield, so that i can look up technical specs of the screen in it?

A few quick searches didnt deliver for me, and i would like to look into them to see what decisions they made for the lcd.
 
Last edited by a moderator:
The only measurement so far is about 46% CPU time for rotating 1080p in software (don't know if this was at 1.7GHz or what).
That was not rotation, just plain update of 1080x1920 @60Hz @1.5GHz.
Is my understanding right, that if we assume 32 bits per pixel:

1080x1920x4       : 8,294,400.00 (about 7.9 MB)


46% of CPU @ 60Hz :         7.67 ms


Giving us         :     1,031.76 MB/s (about 1 GB/s)


I guess the source/destination format would dictate how the data is moved; either way I am making the assumption the limitation is memory bandwidth vs any CPU bottleneck. If this is the case is 1 GB/s what we should expect? I am out of my depth here completely, but random unreliable sites from Google indicate DDR1 gives 1,600MB/s -> 3,200MB/s, whilst DDR2 offers 3,200MB/s -> 8,533MB/s, with DDR3 going more like 6,400MB/s -> 12,800MB/s. I do wish to stress I know very little about these things, I am just trying to understand the actual restriction/limit, I am fully prepared to hear my understanding above is completely wrong, it just feels to me that 46% of the CPU is quite high (but maybe in reality it is what it is when you are talking about 1920x1080).
 
Last edited by a moderator:
As mipi seems to be a pretty new standard I would think that those new screens are mostly portrait for Tablets or smartphones of that size.

wouldn't it be an option to use lvds withnormal cables and instead of ribbon cables like the adafruit one does?
 Even with LVDS, there's no landscape panel with 1280x720 and a usable size.

And LVDS would double the amount of needed pins once again - with a normal cable, you can't get that throught the hinge and with a flex cable it'll be the same LCD Cable issue we had with the Pandora.

Without a landscape LCD available, there's no need to switch to LVDS anyways ;)

Because all phone screens i have seen had the ribbons on the side, instead of the bottom like the pandora and the shield.
I've got LCDs with the cable to the side and to the bottom, all of them are portrait.

You can even let that customize by the manufacturer...

That doesn't say anything about a display being landscape or portrait.
 
You sure?, i mean the area above the fan, the fan is on the main pcb, and the main pcb seems to be ribboned to the screen.

I dont mean the thin long ribbon, but the wide short one slighty above the fan.

btw, is it possible to find tech spec manual/sheets whatever about the shield, so that i can look up technical specs of the screen in it?

A few quick searches didnt deliver for me, and i would like to look into them to see what decisions they made for the lcd.
I'm 100% certain

I ported open dingux to the shield, and it's native portrait, even had an engineer at nvidia confirm this.

http://www.openhandhelds.net/forum2/index.php?/topic/358-linux-for-shield/

I'm very intensely watching this thread, if a software solution is found here, I want to use it there as well, if the solution becomes hardware then my project will remain discarded.
 
Last edited by a moderator:
K tnx for the explanation guys :)

Hopefully Snesfans will get anwsers we need in the dingux/Shield topic.

Also, it seems like quite a shame that Nvidia build a device with impressive specs, only to have it bottlenecked by the perf hit on the orientation.

I can hardly imagine Nvidia making such a business decision...
 
Last edited by a moderator:
That was not rotation, just plain update of 1080x1920 @60Hz @1.5GHz.
So long as that meant copying it from some other buffer that won't fit in cache that, the two should be about the same (for a properly written blocking rotation routine)

Hope we're not actually supposed to run Pyra at only up to 1.5GHz..

I'm very intensely watching this thread, if a software solution is found here, I want to use it there as well, if the solution becomes hardware then my project will remain discarded.
Sorry but I don't think anything anyone says here is going to apply to what you're doing..
 
Last edited by a moderator:
From my experience the screen at 720 will be great. Because most YouTube videos are that quality. And if this panel was designed to reduce power consumption its benefits far outweigh the minimal con of no full HD. Also remember the pyra will evolve over time just like the Pandora. Cheaper production runs appeals to me.. because it helps ed.
 
Last edited by a moderator:
Back
Top