Reply
Thread Tools
GeneralAntilles's Avatar
Posts: 5,478 | Thanked: 5,222 times | Joined on Jan 2006 @ St. Petersburg, FL
#51
Originally Posted by fms View Post
Drivers for this chip exist (dunno how stable though).
About as unstable as they can get and still reasonably be called drivers. They're also for an older kernel.

See the justification page on the wiki for the details you're after.
 
Texrat's Avatar
Posts: 11,700 | Thanked: 10,045 times | Joined on Jun 2006 @ North Texas, USA
#52
Originally Posted by fms View Post
Tablets aside, Nokia's unwillingness to fix firmware problems in certain "unpopular" products (like E70) isn't particularly endearing.
I agree with that sentiment, I just found the hyperbole (as usual) in the other post too funny to resist.
__________________
Nokia Developer Champion
Different <> Wrong | Listen - Judgment = Progress | People + Trust = Success
My personal site: http://texrat.net
 
Posts: 2,102 | Thanked: 1,309 times | Joined on Sep 2006
#53
They're also for an older kernel.
This should not be an issue, the older kernel is still 2.6.x and I presume that the kernel interface has remained the same, should just need a recompile.
 
Posts: 398 | Thanked: 301 times | Joined on Sep 2007 @ Texas
#54
So you got me curious. I went and looked at the OMAP2420 Technical Reference Manual. I can't find the 640x480 limitation for the display or the accelerator hardware. The display hardware does have a limit but it's larger than 640x480.

I understand the internal SRAM is the limit for the resolution you are quoting. Although I don't understand why RFBI (Remote Frame Buffer Interface) mode wasn't used with the internal display subsystem. But the answer to that really doesn't matter.

What matters is how, physically, is the external LCD controller interfaced to the OMAP2420? It seems that if the CPU can access the external frame buffer, the 3d accelerator could as well.

If the bandwidth between the external frame buffer and 3d accelerator is the limiting factor, then this discussion is over. For the purposes of user interface, the 8x0 really doesn't have a 3d accelerator.

I've reviewed the Driver Justification wiki but don't see the answer to the above question.

Frank

Originally Posted by qole View Post
This has been discussed at great lengths elsewhere. It isn't as simple as someone "holding the keys to the drivers needed". There was never a stable, finished driver for the N8x0 graphics accelerator. The graphics hardware isn't just a simple OMAP2 SoC, because the OMAP2 and its PowerVR can only drive a 640x480 display. You may have noticed that the N8x0 screen has 76,800 more pixels than that. They had to hack together a custom solution to drive the big screen, and that meant leaving the PowerVR stuff to gather dust.

I want hardware acceleration as much as anyone. But at this point, I'm hoping for someone to hack together some elegant workarounds.

Last edited by Frank Banul; 2008-12-01 at 15:53.
 

The Following User Says Thank You to Frank Banul For This Useful Post:
GeneralAntilles's Avatar
Posts: 5,478 | Thanked: 5,222 times | Joined on Jan 2006 @ St. Petersburg, FL
#55
Originally Posted by lardman View Post
This should not be an issue, the older kernel is still 2.6.x and I presume that the kernel interface has remained the same, should just need a recompile.
No on both counts, it's for a 2.4.x and the interface has changed within 2.6.x.
 
Posts: 2,102 | Thanked: 1,309 times | Joined on Sep 2006
#56
No on both counts, it's for a 2.4.x and the interface has changed within 2.6.x.
No you are wrong. I have a 2.4.x driver, but Nokia have a 2.6.x driver.

I don't know how large the change inside the 2.6.x series was, but I'm surprised, stuff like that is usually held for a series change isn't it? In any case I'd still reckon (not having looked to see how much has changed) that it would be a simple(ish) re-compile job.

Last edited by lardman; 2008-12-01 at 17:28.
 
Posts: 2,102 | Thanked: 1,309 times | Joined on Sep 2006
#57
So you got me curious. I went and looked at the OMAP2420 Technical Reference Manual. I can't find the 640x480 limitation for the display or the accelerator hardware. The display hardware does have a limit but it's larger than 640x480.
There is no limitation (specific to the 3D driver being able to drive our display that is) afaik, this is mis-information produced by people assuming things.

What matters is how, physically, is the external LCD controller interfaced to the OMAP2420? It seems that if the CPU can access the external frame buffer, the 3d accelerator could as well.
Indeed.

If the bandwidth between the external frame buffer and 3d accelerator is the limiting factor, then this discussion is over. For the purposes of user interface, the 8x0 really doesn't have a 3d accelerator.
The same bandwidth limitation as exists for arbitrary video output will hold (3D hardware writes directly to the framebuffer afaik, framebuffer memory is allocated in SDRAM from where it is copied to the Epson LCD controller chip to perform updates). This in no way (afaict) precludes using the 3D chipset.
 

The Following 3 Users Say Thank You to lardman For This Useful Post:
Posts: 398 | Thanked: 301 times | Joined on Sep 2007 @ Texas
#58
Originally Posted by lardman View Post
The same bandwidth limitation as exists for arbitrary video output will hold (3D hardware writes directly to the framebuffer afaik, framebuffer memory is allocated in SDRAM from where it is copied to the Epson LCD controller chip to perform updates). This in no way (afaict) precludes using the 3D chipset.
If the framebuffer is in SDRAM, the 3d accelerator can access that. Then it's clear to me that the 3d accelerator is usable from a technical standpoint. The rest is a matter of software (not saying it's simple or possible from a licensing perspective).

It seems that the only possible technical limitation would be the update rate possible when copying from SDRAM to external frame buffer. Non technical issues aside, the Fremantle bling would not update at 30fps on the n8x0s. But this shouldn't slow down the overall interface or experience. The CPU does not have to do the 3d rendering. There would be frames dropped is all because the 3d accelerator could update the SDRAM frame buffer faster than it could be copied to the external frame buffer. On video this may suck, but does it matter on menus and other UI widgets? Hard to say without seeing the interface but I would guess not.

Frank
 
lcuk's Avatar
Posts: 1,635 | Thanked: 1,816 times | Joined on Apr 2008 @ Manchester, England
#59
lardman,

unless something drastic has changed the powervr will still render frames which tear when sent over the bus.
Of course if we can use a lower resolution tearfree rgb then we are in business.
__________________
liqbase sketching the future.
like what i say? hit the Thanks, thanks!
twitter.com/lcuk
 

The Following User Says Thank You to lcuk For This Useful Post:
pixelseventy2's Avatar
Posts: 357 | Thanked: 115 times | Joined on Sep 2007 @ Sunny England :)
#60
Originally Posted by deadmalc View Post
You'll get the same problem with windows mobile, have you ever been able to upgrade any mobile from one version of windows mobile to the next? Even if you could for one, can you do it for all?
I have performed winmo upgrades. Many times. Both official (from hardware vendors) and unofficial. There is a thriving community around porting new (admittedly of dubious legality) windows mobile versions onto old hardware. This is why I have devices which came with "pocket pc 2003" running windows mobile 5. Although I don't use them, because my n800 is much better. No, they're not available for all, they tend to be directed towards HTC devices. But given that HTC still sells the most winmo devices, that's not too much of an issue.

And it you want to get really picky, you can always just put linux on many of them
__________________
pixel - pushing buttons that shouldn't be pushed, and fiddling with things that shouldn't be fiddled with
 
Reply


 
Forum Jump


All times are GMT. The time now is 14:01.