BeBits - Ati Radeon Graphics Driver - Talkback

BeBits

Ati Radeon Graphics Driver

Talkback

Go back to the Ati Radeon Graphics Driver page

Post a new Talkback comment!

log

By kvdman - Posted on January 7, 2006 - 17:54:29 (#18796)
Current version when comment was posted: 5.1

right, here's my log:

http://s89161082.onlinehome.us/syslog.zip

By goAtsy - Posted on January 7, 2006 - 17:35:21 (#18795)
Current version when comment was posted: 5.1

Here is the syslog:

http://pastebin.com/495025

and thanks for your help.

By Euan - Posted on January 7, 2006 - 17:01:31 (#18794)
Current version when comment was posted: 5.1

Lots of info to go on here.

Adam, thanks for that, I was thinking that, but was too tired to start a search, hopefully there is something in there that is specific to the X300/600/700 series that fixes the problem.

emwe, the 108000 value is in Khz, the 216000000 is in Hz, apparantly the first is pixel clock, and the other is PLL frequency, which should be 2x the pixel clock, so it looks ok after further investigation. I can't access your syslog, can you email me it?

goatsy, thanks for that can you send me a syslog?

...

By goAtsy - Posted on January 7, 2006 - 15:59:32 (#18792)
Current version when comment was posted: 5.1

I get weird results with this driver.

If I just install the two files from the package nothing happens and no "Radeon" info is shown in syslog.

When I drag a link from the driver to the "/boot/home/config/add-ons/kernel/drivers/dev/graphics/" and reboot i get this:

http://picshosted.com/v/7164/screen.png

When I click on the deskbar button the menu takes about 20 seconds to draw.
Nothing that is drawed to the screen is removed, only overdrawed by other text/items.
But in this mode i get all the radeon selections in the screen preferences.

backlight not enabled

By emwe - Posted on January 7, 2006 - 15:12:00 (#18790)
Current version when comment was posted: 5.1

Hi Euan, thanks for your efforts on the driver!

Here is my log:
http://beos.m-phasis.de/zeta_x600_m24_0x3150_pcie_513c.txt

The following line is interesting:
Radeon - Radeon_SetupDefaultMonitorRouting: display_devices=4, whished_num_heads=1, use_laptop_panel=0

It doesn't notice that it is a laptop panel here, and thus I only get black. Additionally, when attaching a LCD to the VGA-out connector, the colors, are totally weird, and the UI builds up hell hell slow.

You mentioned the PLL-freq differences: Once it says 108-thousand, then somewhere else, 216-million. Is this frequency somehow related to how "fast" the 2D-chip is performing it's drawing ops? Then this would be a 2000x differnce.

Michael

Microcode and stuff.

By adamk - Posted on January 7, 2006 - 14:13:39 (#18789)
Current version when comment was posted: 5.1

I'm pretty sure, based on the feedback I get when I launch X on my X700 (vs a 9600), that it's the same microcode. In both cases, it says it's loading the R300 microcode.

However, if you want to look, I believe the microcode is part of the DRM kernel module. CVSROOT = :pserver:anonymous@dri.freedesktop.org:/cvs/dri and the module is 'drm'. It is also part of the linux kernel, if you have that somewhere.

The results with 5.1.3c are the same as with 5.1.3b:

http://68.44.156.246/beos-radeon-x700-513c-mtrr.txt
http://68.44.156.246/beos-radeon-x700-513c-womtrr.txt

The first is with the mtrr drivers back in place and the second is without. I get a hard lockup with 5.1.3c without the mtrr driver.. The mouse is moveable at first, as soon as the appserver launches. I can telnet in, but within a few seconds, the machine locks up as before.

I also grabbed a picture of the screen with my digital camera, in case that gives you some more insight into what's going on:

http://68.44.156.246/beos-radeon-x700-513c-image.jpg

Adam

By Euan - Posted on January 7, 2006 - 07:28:16 (#18786)
Current version when comment was posted: 5.1

Version 3 for today:

Changes to Monitor routing, to fix DVI routing, (should fix one of the X700 problems).

I need to find where Xfree86 has hidden their microcode, and associated code for the DMA command engine. If anyone knows where it hides let me know. Maybe that section has moved to the kernel??? I'm thinking that maybe X700 needs a different microcode, and doesn't work with the R300-R420 series microcode?

http://euank.users.btopenworld.com/beos/BeOS_Radeon_V5.1.3c_Bin.zip
http://euank.users.btopenworld.com/beos/BeOS_Radeon_V5.1.3c_SRC.zip

more work to do

By Euan - Posted on January 7, 2006 - 00:53:11 (#18783)
Current version when comment was posted: 5.1

I can't figure out why the X700's aren't working they should be identical to the X800s, makes no sense. :( Additionally I need to port some more code related to monitor routing, this should fix your monitors not being setup right. At the moment it's semi-hardcoded. Your X700 is the opposite of the X800, so hardcoding isn't going to work based on chipset.

hmm

By Euan - Posted on January 7, 2006 - 00:23:12 (#18782)
Current version when comment was posted: 5.1

thats odd then, as the driver is sure you've got something connected, and it then goes and limits your resolution to 1024x768...

By adamk - Posted on January 7, 2006 - 00:13:50 (#18781)
Current version when comment was posted: 5.1

I don't have it hooked up. The card comes with three output ports: tv-out, DVI, and CRT. I have two CRT monitors hooked up, the second uses an DVI-CRT adaptor that came with the card.

The X700 was first a PCIe card.. However, ATI has released AGP versions of the card.

hmm

By Euan - Posted on January 7, 2006 - 00:08:55 (#18780)
Current version when comment was posted: 5.1

Can you try again with tv-out disconnected.

It looks like the 2d engine simply isn't working. Very odd... Also the card id is for a PCIe card, but that shouldn't really make any difference.

By adamk - Posted on January 6, 2006 - 23:49:31 (#18779)
Current version when comment was posted: 5.1

"maybe this is because the device is PCIe???"

This is definitely an AGP card though :-)

"Have you installed the Haiku AGP driver, or via_cpu driver by any chance?"

Nope. Though I have the Haiku driver on my machine (I have a full checkout from svn), it's not installed. Nothing is showing up under ~/config/add-ons or /system/add-ons that has "agp" in it. And no via_cpu driver anywhere on my system.

"Could you do a search for mtrr, and move any kernel drivers with that name to a temporary folder elsewhere?"

I removed the kernel mtrr directory.. I still get a lockup, but no kernel crash to the debugger. The appserver clearly tries to launch but the screen is a mess and the system doesn't respond to pings, so I know the crash is pretty solid.

http://68.44.156.246/beos-radeon-x700-take2.txt.gz

Adam

By Euan - Posted on January 6, 2006 - 22:39:22 (#18778)
Current version when comment was posted: 5.1

That's interesting it crashes when trying to set up mtrr. maybe this is because the device is PCIe??? Have you installed the Haiku AGP driver, or via_cpu driver by any chance? Could you do a search for mtrr, and move any kernel drivers with that name to a temporary folder elsewhere?

By adamk - Posted on January 6, 2006 - 22:32:37 (#18777)
Current version when comment was posted: 5.1

Should your new driver, where you added better support for X600 cards, work better for X700 cards, too? I'm not sure I'm competely following the threads in this talkback, but it looks like you're tracking down display problems for laptop panels, correct, so I'm assuming that there shoudl be no signs of improvement on my dual CRT AGP X700... But, in case you were expecting some improvement, here's the syslog from the most recent (5.1.3b) driver.

http://68.44.156.246/beos-radeon-x700.txt.gz

Interesting...

By Euan - Posted on January 6, 2006 - 21:26:34 (#18775)
Current version when comment was posted: 5.1

Hi,

Thanks for that, it should be very useful. Two things of interest:

Radeon - addFPMode: H: 1400 1448 1560 1640 (v=1400)
Radeon - addFPMode: V: 1050 1051 1054 1065 (h=1050)

This is your video mode.

Pixel clock is max horizontal * max vertical * refresh rate.

So your pixel clock should be 1640*1065*60 (assuming 60Hz refresh) = 104796, not 108000 stated by the bios.

What is more interesting is that the video card ignores this and tries to set the resolution to 1024*768. I have to look at the source to figure that one out however. It could be that there are still vesa settings set somewhere, or that the flatpanel custom modeline is being ignored, and the set display mode code is reverting to the nearest similar mode.

Also the lines of code added in the last update were:

Radeon - Radeon_GetBIOSDFPInfo: BIOS provided dividers found (but not yet used): ref 6, post 1, feedback 48

According to the specs, these settings should be used to calculate the pixel clock directly. EG:

27000 * 48 / 6 * 1 but this equals 216000 (oddly enough exactly 2x the pixel clock given by the bios (108000). Not to mention feedback should be a value in the range 128 to 255...

syslog.. once again :)

By fanskapet - Posted on January 6, 2006 - 20:30:44 (#18773)
Current version when comment was posted: 5.1

hmm Zeta wants it to be named radeon.driver so i just renmed it :)

here's the result!

http://pastebin.com/493808

By Euan - Posted on January 6, 2006 - 20:29:29 (#18772)
Current version when comment was posted: 5.1

The radeon driver wasn't started at all in that last log, can you try installing them again?

new syslog!

By fanskapet - Posted on January 6, 2006 - 20:24:47 (#18771)
Current version when comment was posted: 5.1

http://pastebin.com/493800

By Euan - Posted on January 6, 2006 - 19:28:25 (#18770)
Current version when comment was posted: 5.1

What happens if you manually select a resolution, without selecting "use failsafe resolution".

Can you also post/email a syslog.

tested mobility 7500 m7

By kvdman - Posted on January 6, 2006 - 19:14:01 (#18769)
Current version when comment was posted: 5.1

before I would just get a white haze on my screen, with this driver update i'm able to boot into 800x600 resolution in 8 bits, but cant change that.

next verson: 5.1.3b

By Euan - Posted on January 6, 2006 - 19:11:25 (#18768)
Current version when comment was posted: 5.1

This one will print out some values read from the bios which were previously ignored. For the curious they are the reference, post, and feedback drivers. In case your wondering, these 3 values, and the PLL reference clock are used to calculate the pixel clock. If your bios's give is sensible numbers, I'll use these numbers instead of the algorithm that looks for the "nearest fit".

http://euank.users.btopenworld.com/beos/BeOS_Radeon_V5.1.3b_Bin.zip

By Euan - Posted on January 6, 2006 - 18:30:23 (#18767)
Current version when comment was posted: 5.1

I'll take a look at this one. It would be great if you do the same for the one I'll post in 30mins.

By fanskapet - Posted on January 6, 2006 - 17:45:04 (#18766)
Current version when comment was posted: 5.1

well il just post the one i got now anyway if you have any use of it :)

http://pastebin.com/493594

the driver still works with an external monitor, works like a charm when i plug in my LCD TV in the VGA outlet on the laptop.. but when i boot without an external monitor i just get a pitch black screen.. some gray light though so maybe some strange thing relating the backlight has occurd? (the laptop panel gets pitch black when using an external monitor too.. but the external one works.)

By Euan - Posted on January 6, 2006 - 16:25:50 (#18765)
Current version when comment was posted: 5.1

I'll post a new driver for you to use, it should print out some extra info that will be useful.

Screen is visible for me (but noisy)

By Philippe Houdoin - Posted on January 6, 2006 - 15:44:26 (#18764)
Current version when comment was posted: 5.1

Oh, I can see the screen, it's just noisy. The framebuffer is valid, as a screenshot displayed under another OS/system show. It's just some noise due to DAC small clock timing shift.
Even if it's more tiring looking at such screen than under VESA, it's far more speeder that way!
;-)

Anyway, I'll post my syslog when I'll be home tonigh (in 5 hours).

- Philippe

fanskapet, and phillipe

By Euan - Posted on January 6, 2006 - 15:18:20 (#18763)
Current version when comment was posted: 5.1

Can you both email me syslogs, as you can't see the screen just wait a while for disk activity to stop, then do ctrl-alt-del a couple of times to restart.

not working, pitch black

By fanskapet - Posted on January 6, 2006 - 13:38:58 (#18762)
Current version when comment was posted: 5.1

hmm sadly my LCD panel just get's pitch black with this updated driver aswell.. im able to run in vesa mode though :/ but maybe this driver is unsupported by ZETA?

agh i wish my Mobility 9700 would work so i could get back in Be once again :) VESA is not any fun at all :P

Re: New binaries for testing

By Philippe Houdoin - Posted on January 6, 2006 - 10:50:04 (#18761)
Current version when comment was posted: 5.1

Hi Euan,

I've tested your latest 5.1.3 binaries on my Asus W1N laptop's ATI Radeon Mobility 9600 (M10). It's recognized by your driver but unfortunatly the screen still presents noisy lines artifacts here and there due to wrong pixel clocks.

According to syslog, the BIOS report flat panel info with a wrong pixel clock. But on PROPOSE/SET_DISPLAY_MODE, the clock value proposed sounds more better. Except that after Radeon_CalcPLL(), the clock pixel is set to rounded value, hence the wrong DAC timing visual issue.

As you didn't include the source/patch over the 5.1 source, I can't investigate more deeper.

Anyway, thanks for your efforts to unify all these fixes/hacks over 5.1 into one single version. Maybe you should consider release soon this unified version? And what about updating the Haiku's radeon driver directly in the SVN repository next?

Regards,

New binaries for testing

By Euan - Posted on January 5, 2006 - 01:09:55 (#18754)
Current version when comment was posted: 5.1

http://euank.users.btopenworld.com/beos/BeOS_Radeon_V5.1.3_Bin.zip

By Euan - Posted on January 4, 2006 - 23:39:00 (#18751)
Current version when comment was posted: 5.1

some big changes for people with mobility, IGP, and rv3x0 chips (yes thats X600's). Hopefully they do some good.

Just going down into beos to do the compile, fix typo, compile, fix typo routine. Should have something in about an hour assuming I haven't messed something up.

The version will also turn on Dynamic clocks for mobility chips, so that needs some special attention when testing...

Idem on M50-142 (X600se)

By ljr - Posted on January 4, 2006 - 20:34:02 (#18749)
Current version when comment was posted: 5.1

>All 16bit resolutions:
<Very Dark. Darkgreen background, can hardly see icons. A >little better visibility on the lower part of the screen.

>All 8 bit resolutions:
>Brightly distorted colors. Lightgrey background. Menus >and windows contains orange, blue, etc.

same here in vesa.
I modded the source to add the pci ids and boot screen is nice but zeta hangs when desktop is loaded... I try different profile (R300, 350 360 etc... but same result). The X600se seems to be a r380.

:-(

Philippe...

By goAtsy - Posted on January 4, 2006 - 18:06:22 (#18746)
Current version when comment was posted: 5.1

Results are the same for all vesa resolutions in respective bitdepth.

All 16bit resolutions:
Very Dark. Darkgreen background, can hardly see icons. A little better visibility on the lower part of the screen.

All 8 bit resolutions:
Brightly distorted colors. Lightgrey background. Menus and windows contains orange, blue, etc.

The background selected is the standard blue beOS background.

Re: More info

By Philippe Houdoin - Posted on January 4, 2006 - 16:19:38 (#18744)
Current version when comment was posted: 5.1

> Looks really wierd, to bad i can't show you on
> a screenshot as the shot looks perfect when
> watching it on a working system.

So it's a pixel clock timing issue. What's the VESA resolution you selected?

By goAtsy - Posted on January 4, 2006 - 12:12:05 (#18743)
Current version when comment was posted: 5.1

I checked som more and the driver isn't even loaded. My X300 isn't supported.

From syslog:
KERN 'app_server'[34]: Radeon - init_hardware: Version: 5.1.1.0
KERN 'app_server'[34]: Radeon - Radeon_CardDetect: no supported devices found

The dimmed screen occours as soon as i set Vesa to 16bit mode.
Looks really wierd, to bad i can't show you on a screenshot as the shot looks perfect when watching it on a working system.

By goAtsy - Posted on January 3, 2006 - 21:49:30 (#18739)
Current version when comment was posted: 5.1

I used the one linked below, assuming i didn't do anything wrong when i installed it.

I overwrote all 'radeon.accelerant' with the one from the package.

I removed all 'radeon.driver' and placed 'radeon' from the package in the same locations.

I removed old symlinks and made new ones named 'radeon' pointing to the drivers.

By Euan - Posted on January 3, 2006 - 17:21:55 (#18735)
Current version when comment was posted: 5.1

Using the driver I linked below, or the one that comes with Zeta 1.1???

By goAtsy - Posted on January 3, 2006 - 13:17:15 (#18732)
Current version when comment was posted: 5.1

I have a Mobility Radeon X300 PCIE-card (venID:1002, devID:5460) sitting in a IBM T43 notebook.

I get the same dimscreen problem described by another user here --> http://yellowtab.com/phorum/phorum.php?read,3,123713,123713#msg-123713

By Garjala - Posted on January 1, 2006 - 19:25:03 (#18720)
Current version when comment was posted: 5.1

HIS Radeon X800 Pro IceQ2 vivo + Viewsonic VP191b in DVI seems to work with DVI fixed driver

New Test driver for people DVI problems

By Euan - Posted on December 31, 2005 - 20:00:36 (#18715)
Current version when comment was posted: 5.1

Hi all,

This driver remaps the edid information on Cards newer than 9800 series to the correct location. I also added full dumps of the EDID DDC monitor information to the syslog. Tested on a 9600, and X800.

No specific changes for X700 folks and people with PCIe who are having problems yet. Soon...

Binaries:
http://euank.users.btopenworld.com/beos/BeOS_Radeon_V5.1.2_XSeries_DVI_Fix_Binaries.zip

Source:
http://euank.users.btopenworld.com/beos/BeOS_Radeon_V5.1.2_XSeries_DVI_Fix_Source.zip

Usual disclaimer, use at your own risk...

Please post your results and monitor setup / type info etc for sucess, syslogs for failure. :)

Happy Hogmany!

By adamk - Posted on December 30, 2005 - 13:10:06 (#18704)
Current version when comment was posted: 5.1

That's all she wrote :-)



	Comment Pages: << prev \| 1 \| 2 \| 3 \| 4 \| 5 \| 6 \| 7 \| 8 \| 9 \| 10 \| 11 \| 12 \| 13 \| 14 \| 15 \| 16 \| 17 \| 18 \| 19 \| 20 \| 21 \| 22 \| 23 \| 24 \| 25 \| 26 \| next >>













			Recent Downloads - # 2 Total Downloads - # 2 Total Views - # 2 User Ratings - # 629




			Video Drivers




			1. BePodder - 9.80 2. QEMU - 9.68 3. ScummVM - 9.57 4. cpu_fix - 9.42 5. Jukebox - 9.40 6. libdl.so - 9.40 7. Haiku AGP busm... - 9.35 8. vim6 - 9.31 9. Beezer - 9.26 10. HandBrake - 9.25




			1. BeOS 5 Perso... - 13,657 2. Realtek RTL8... - 13,142 3. Ati Radeon G... - 12,641 4. Ensoniq Audio... - 7,598 5. ATI Rage 128... - 7,466 6. USB Joystick... - 5,685 7. Broadcom 440x... - 5,432 8. S3 Trio 64 v2... - 4,753 9. USB Serial dr... - 4,721 10. Intel Extreme... - 4,483




			You are not logged in. Login


	Unless otherwise noted, everything is copyright © 1999-2009 Haikuware. All Rights Reserved. For more legal trivia, take a gander at our Legal Stuff page and our Privacy Statement.