 |
 |
|
Haiku nVidia TNT/GF driver |
|
 |
|
|
 |
 |
| and more overlay. |
 |
 |
By rudolfc - Posted on November 6, 2003 - 20:45:49 (#9524)
Current version when comment was posted: 0.07 |
 |
 |
vegac,
So did you or did you not recompile the driver?
BTW: setting another colordepth should have nothing to do with the vesa file in our case I think.
Just leave it as it is and use the Screenprefs panel to chance colordepth without changing resolution or refreshrate. Does that work? Or does this give those repeating or only partly visible screen you wrote about?
(just double checking ;-)
Thanks!
Rudolf.
|
|
| and overlay... |
 |
 |
By rudolfc - Posted on November 6, 2003 - 20:42:00 (#9523)
Current version when comment was posted: 0.07 |
 |
 |
Hi Rene!
B_YCbCr422 is the one and only space supported in the nVidia driver ATM. On FourCC (http://www.fourcc.org if I'm not mistaken) it's listed as YUY2, which was registered by nVidia...
Of course, it could be possible some nVidia cards have trouble with it. You never know ;-)
Though all my cards (four TNT series, four GeForce series) are working without a single glitch :)
Rudolf.
|
|
| overlay... |
 |
 |
By rudolfc - Posted on November 6, 2003 - 20:38:31 (#9522)
Current version when comment was posted: 0.07 |
 |
 |
Hi,
s_d, It doesn't matter, I was only posting this technical stuff because I thought it might help one of you out.
Also, you see how important precise feedback is if I myself were to actually use it for driver enhancements. I took the wrong turn in believing you were posting about a laptop as it turns out to be a TNT type card.
The technical stuff (the mods for the driver) only apply for GeForce series cards, TNT series cards have another unit that's programmed in the lines before the stuff I mentioned. Though the setup is about the same ;-)
Thanks anyway: your combined feedback is a hint that there is still room for improvement on the TNT overlay 'front'.
I'm afraid though these kind of 'items' not working on _some_ cards outthere is and will remain typical for a driver which needs to support a wide variety of cards without any support from the manufacturor. :-/
|
|
| Re: Overlay |
 |
 |
By vegac - Posted on November 6, 2003 - 17:16:29 (#9521)
Current version when comment was posted: 0.07 |
 |
 |
I tested Overlay in 16bit color mode. I can't get any other color modes to work. If I set them in the vesa file, then the laptop display won't init, or if I set them in the screen preferences app the screen comes up but looks messed up (I see a quarter of the screen, repeating itself, odd coloring, etc.)
|
|
| Re: overlay fun |
 |
 |
By Rene Gollent - Posted on November 6, 2003 - 12:48:23 (#9520)
Current version when comment was posted: 0.07 |
 |
 |
Are the apps with trouble using a different overlay mode perchance? The most common one is yCbCr422 if I remember the name correctly, but there might be others in use that the TNT's BES doesn't handle correctly.
|
|
| overlay fun |
 |
 |
By s_d - Posted on November 6, 2003 - 12:09:41 (#9519)
Current version when comment was posted: 0.07 |
 |
 |
Hi Rudolf
That old TNT card by ASUS is instaqlled at my work computer in 500 MB PE in file, so it's almost no use:)
Maybe i'll find time for tweaking in order to help development, but chances aren't so big - i just shared my experience for those who wish to watch movies immediately:) BTW, i have feeling that different apps have different success in starting overlay.
|
|
| overlay hint/2 |
 |
 |
By rudolfc - Posted on November 6, 2003 - 11:01:16 (#9518)
Current version when comment was posted: 0.07 |
 |
 |
BTW,
I am assuming both of you are using a laptop internal panel. Please attach an external monitor also to see if overlay (and cursor) are OK on that (without modifying the driver)!
I need to know this as well if possible please..
Rudolf.
|
|
| Overlay fun/more info and questions |
 |
 |
By rudolfc - Posted on November 6, 2003 - 10:53:10 (#9517)
Current version when comment was posted: 0.07 |
 |
 |
Hi there,
vegac, You should tell us which colordepths you tested for overlay also please (I need all of them ;-)..
s_d, can you also test 8-bit colordepth for me?
The hardcursor has no influence on overlay use BTW. s_d has a colorkeying problem in 16-bit colordepth, so maybe vegac has that also.
s_d, if you have the developertools installed, you can start tweaking the driver yourself to find out if you can get keying up and running in 16-bit color.
Open up nv/nv_bes.c in BeIDE, and look at line 623, which looks like this:
BESW(NV10_0SRCPTCH, (((ob->width * 2) & 0x0000ffff) | (1 << 16) | (1 << 20) | (0 << 24)));
Please modify it (as a first test) to be:
BESW(NV10_0SRCPTCH, (((ob->width * 2) & 0x0000ffff) | (1 << 16) | (0 << 20) | (0 << 24)));
So bit 20 is zero instead of 1. This disables overlay keying and should give you overlay in 16bit depth.
After a 'make install' and a reboot, you will notice dropdown menus are nolonger visible (For instance) if they are displayed on top of the video.
If you now have overlay, set bit 20 back to 1 (Re-enable keying).
Now look at line 658:
case B_RGB16_LITTLE:
BESW(NV10_COLKEY, (
((ow->blue.value & ow->blue.mask) << 0) |
((ow->green.value & ow->green.mask) << 5) |
((ow->red.value & ow->red.mask) << 11)
/* this space has no alpha bits */
));
This is the key for 16-bit color. Apparantly the placement is off for your card, so it would be nice if you could tweak a bit with that to see if you can get it to work. Check if dropdown menus work if you get overlay so you _know_ keying is operational.
The register programmed is 32bit wide, so you can imagine there are other setups possible in theory. Though I find it very strange if the bitpositions need to be changed for you!
Try removing or changing the masks for instance to see what happens..
vegac, you can also try to disable keying to see if that gives you overlay. But first, please try all depths without modifying the driver...
Thanks!
Rudolf.
PS Please report back here!
|
|
| Overlay fun |
 |
 |
By s_d - Posted on November 6, 2003 - 04:15:25 (#9514)
Current version when comment was posted: 0.07 |
 |
 |
i got similar situation with color depth = 16 bit - grey screen.
After i changed screen settings to 32 or 15 bits - Overlay started ok in VLC
|
|
| No Overlay for me... |
 |
 |
By vegac - Posted on November 5, 2003 - 20:57:55 (#9510)
Current version when comment was posted: 0.07 |
 |
 |
Well, I loaded up VLC 0.6.2, popped in a DVD, and tried to watch it. It played in Bitmap mode fine. Went to the config, enabled overlay support, restarted VLC, and all I get is a grey box.
Is this somehow related to BDirectWindow which requires hardware cursor to be enabled for it to work? Unfortunately, enabling that gives me no mouse...
|
|
| TVout support! |
 |
 |
By rudolfc - Posted on November 5, 2003 - 19:44:31 (#9508)
Current version when comment was posted: 0.07 |
 |
 |
Hi,
TVout support already exists for a limited number of cards. Just download BeTVOut from BeBits and try it.
I wrote BeTVOut as my first project for BeOS specifically for the purpose of enabling TVout on any nVidia card being TNT1 or later.. And _if_ it works for you, and you like watching DVD, you are going to _love_ it ;-)
BeTVOut will be updated once or twice more in the future to add Philips encoder TVout support, and to eliminate remaining issues with later GeForce cards (see talkback comment from me on BeTVOut page). After that its functionality will be included in this nVidia driver so you will nolonger need BeTVOut or it's driver.
Rudolf.
PS Everyone thanks for the reports!
|
|
| BDirectWindow support hint... |
 |
 |
By rudolfc - Posted on November 5, 2003 - 19:39:13 (#9507)
Current version when comment was posted: 0.07 |
 |
 |
Hi there,
Here's a quick tip, which I just now found out due to some very sharp bug-report from someone (thanks!):
If you want to use windowed BDirectWindow modes (Chart demo app for instance), you have to make sure you have the hardcursor enabled.
BDirectWindow needs Parallel buffer access plus a hardcursor for this to work!!
Of course, if the hardcursor does not work on your card (yet), you are out of luch for BDirectWindow windowed mode support.
I'll include this hint in the nv.settings documentation for the next release.
Rudolf.
|
|
| laptop internal panel tests/overlay |
 |
 |
By rudolfc - Posted on November 5, 2003 - 19:34:59 (#9506)
Current version when comment was posted: 0.07 |
 |
 |
Hi vegac,
One thing left to test: Hardware overlay. Run VLC 0.62 preferably with a high res movie to see if overlay is working. The Video output window should have 'overlay' in it's titel.
If this is not the case, check the VLC preferences, video configuration for enabled overlay output.
A good way of testing overlay is to run high-res video in a 100% output window (DivX or DVD), and hit pause.
Now move the window partly offscreen to the left side of your screen. Does output remain in the correct 'spot'?. Now move the window to be partly beyond the top of your screen. Still OK?
If all these tests are positive, overlay is 100% OK ;-)
Rudolf.
|
|
| Nice work Rudolph |
 |
 |
By verranm - Posted on November 5, 2003 - 13:37:39 (#9501)
Current version when comment was posted: 0.07 |
 |
 |
Works beautifully, I just yanked an old Radeon out of my box and replaced it with a 5200 as it was cheaper than the radeon9600 i wanted.
I've got to say your driver is just as stable and fast as Thomas's radeon drivers - keep up the excellent work. I look forward to TVout!
|
|
| Excellent! |
 |
 |
By dpzektor - Posted on November 5, 2003 - 08:27:04 (#9497)
Current version when comment was posted: 0.07 |
 |
 |
Thank you greatly for the driver! I have not had the ability to use overlay since I moved to the Riva TNT2 card a long while back, but with this driver I do now! It is very fast and stable to boot (I have been running it for a few days now). Once again, thank you. Any chance of allowing the driver to utilize the TV-Out on my card in the future? That would just make my day :)
|
|
| Oddities in Laptop displays... |
 |
 |
By vegac - Posted on November 5, 2003 - 07:24:51 (#9496)
Current version when comment was posted: 0.07 |
 |
 |
Ok, so I upgraded to the latest driver (0.7)
My screenmode that I have set with VESA comes up. That tells me it's using Vesa. However, if hardmouse is set to true, my mouse disappears. Setting it to false and I have a mouse again. That tells me it's using the nVidia driver.
Window dragging, especially Firebird, and scrolling, both seem to be much MUCH faster. This tells me I think it's running accelerated. Window drags are almost perfectly smooth (vs. before I would watch it redraw).
So, what I'm saying is, setting the mode via vesa, and using the driver (albeit w/ hardmouse off) seems to be working for me right now with the internal display on my GeForce2 GO based laptop.
Anything else I should test?
|
|
| Refresh rate (more) precise settings |
 |
 |
By s_d - Posted on November 4, 2003 - 18:14:00 (#9483)
Current version when comment was posted: 0.07 |
 |
 |
Addition to VESA/Flatpanel tricks.
If someone don't know yet - if you choose "Other" for refresh rate, you can use Ctl+arrows ot shift+arrows to move refresh rate slider in 0.1 Hz steps
|
|
| laptop panel support /2 |
 |
 |
By rudolfc - Posted on November 3, 2003 - 20:49:09 (#9474)
Current version when comment was posted: 0.07 |
 |
 |
Oh yes,
I am very aware laptop support would be _very_ welcome. I also know having to setup some nv.settings presets to get it working on a specific laptop is a minor detail. So, don't worry: I'll do my best to get it up and running, and I already have _some_ experience with this stuff from the (open)BeOS NeoMagic driver that already contains some support.
I'll keep a very sharp eye out for any laptop containing a Vidia graphics card, I am _even_ considering just bying one so I can actually test for at least _one_ card. It just seems a big 'waste' of money, considering I already _have_ a laptop.. ;-)
Tell you what: I am moving this item up one 'point': first dualhead support, then laptop panel support. This is the logical order of things, as it allows me to rule-out confusing the panel registers with the dual(multi) head registers.
Besides: Dualhead support is testable over here :-)
Rudolf.
|
|
| possible laptop panel 'support' tricks |
 |
 |
By rudolfc - Posted on November 3, 2003 - 20:29:31 (#9473)
Current version when comment was posted: 0.07 |
 |
 |
Hi,
vegac, thanks for your hint ;-)
Everyone: here's the full story:
(Assuming you mean to tell us you have a working internal flatpanel by setting up a vesa mode..)
There are users who did _not_ setup the vesa file that still have working internal panels, but I guess your chances of success increase somewhat if you use a vesa mode, setup for your panel's native resolution. This is because a realmode call to your VGA BIOS will be done by BeOS then before switching to protected mode. The BIOS VESA extension call will program your card to do the mode selected.
Because the driver will not overwrite any laptop specific registers that might get programmed by the BIOS, you have a chance for it to work.
You may _not_ change the mode's refreshrate or resolution (you need to have set the same resolution for the driver for it to cooperate with VESA mode).
You _may_ choose another colordepth than you choose for the vesa file.
OK, _if_ you actually have a proper image on your panel with driver 0.06 or 0.07 (with or without a vesa file), You already _know_ 2D acceleration is working. If you see a cursor as usual, you _know_ the hardcursor is working. This is because building the screen alone already invokes all these funtions in the driver!
(blue background == rectangle fill, any Window shown is a rectangle fill with numerous screen to screen blits!)
Well, now you only have to test overlay to see if that also works :-)
Of course, people getting the driver to work with internal panels this way, are encouraged to let me know.. It's important to get a decent general picture of where we stand ATM ;-)
------
OK, the part below is for the people liking to take a chance on destroying some monitor (if they are not carefull)...
Most of you will have a problem setting the BeOS ScreenPrefs panel to the same settings as you'd have setup in the vesa file while running BeOS without a driver in B/W mode or VESA (only) mode.
There's a trick to that: Open the file ~/config/settings/app_server_settings with StyledEdit.
This file contains all modes for all workspaces the app_server will set on startup. If you overwrite this file it will take effect on the next reboot _if_ you don't touch the ScreenPrefs panel, as that causes this file to be overwritten also.
Here's a snippet from my system:
Workspace 0 {
timing 48486 800 856 976 1040 600 637 643 666 0x60000000
colorspace 0x00000008
virtual 800 600
flags 0x8000001f
color 466698
The 'timing' line is the refreshrate in kHz, followed by horizontal total visible, hor. start of sync, hor. end of sync, hor. total, vert. total visible, ver. start of sync, vert. end of sync, vert. total, timing_flags.
Calculating the refresrate is simple:
refresh = (pixelclock * 1000)/(hor. total * vert. total)
In this example you get (48486 * 1000) / (1040 * 666) = 70.0Hz. I am looking at my screen working on 70Hz refresh ATM :-)
The timing_flags can be left alone. Of course you can find the meaning of the used bits in some Be header file. (I am using both positive Hsync and Vsync. Only 640x480 modes use negative sync by default AFAIK).
The next line:
colorspace 0x00000008
Tells you I am running B_RGB32.
The line:
virtual 800 600
tells you the virtual screen size (hor. vert.). Normally this should be set to the same size as hor. total visible and vert. total visible.
flags 0x8000001f are the mode flags.
Best left alone: this reports your cards capabilities with the driver used. Bit31 tells you (if set) overlay is supported.
color 466698 probably is the background color used for the Desktop. Don't actually know: never touched it.
--------
What you need to set is in fact a 'modeline' Some of you might have such a line written down already because you know it from Linux. You can almost just type it in here :)
If you don't know any modeline you wish to set, just look at the driver's sourcecode. In the file ProposeDisplayMode.c, at the very top, is a list of them used by the driver.
Just choose the one matching 60Hz refresh for your panel and 'copy' it in the app_server_settings file.
An example:
{ { 40000, 800, 840, 968, 1056, 600, 601, 605, 628, T_POSITIVE_SYNC}, B_CMAP8, 800, 600, 0, 0, MODE_FLAGS}, /* Vesa_Monitor_@60Hz_(800X600X8.Z1) */
You can use the numbers before T_POSITIVE_SYNC to fill out the
'timing' line in the same order.
Make sure you set the 'virtual' line accordingly (800 600 here).
--------
For the people programming the BeOS here's a snippet from Accelerant.h that tells you exactly what all this stuff actually is:
typedef struct {
uint32 pixel_clock; /* kHz */
uint16 h_display; /* in pixels (not character clocks) */
uint16 h_sync_start;
uint16 h_sync_end;
uint16 h_total;
uint16 v_display; /* in lines */
uint16 v_sync_start;
uint16 v_sync_end;
uint16 v_total;
uint32 flags; /* sync polarity, etc. */
} display_timing;
typedef struct {
display_timing timing; /* CTRC info */
uint32 space; /* pixel configuration */
uint16 virtual_width; /* in pixels */
uint16 virtual_height; /* in lines */
uint16 h_display_start; /* first displayed pixel in line */
uint16 v_display_start; /* first displayed line */
uint32 flags; /* mode flags */
} display_mode;
enum { /* timing flags */
B_BLANK_PEDESTAL = 1 << 27,
B_TIMING_INTERLACED = 1 << 28,
B_POSITIVE_HSYNC = 1 << 29,
B_POSITIVE_VSYNC = 1 << 30,
B_SYNC_ON_GREEN = 1 << 31
};
enum { /* mode flags */
B_SCROLL = 1 << 0,
B_8_BIT_DAC = 1 << 1,
B_HARDWARE_CURSOR = 1 << 2,
B_PARALLEL_ACCESS = 1 << 3,
B_DPMS = 1 << 4,
B_IO_FB_NA = 1 << 5
};
And some GraphicsDefs.h:
B_RGB32 = 0x00000008,
B_RGB16 = 0x00000005,
B_RGB15 = 0x00000010,
B_CMAP8 = 0x00000004,
------------
Ok, for the diehards amongst you (needing a new laptop anyway ;-), have fun trying to 'tweak' your display settings..
Best regards,
Rudolf.
|
|
| Re: Laptop internal panel support |
 |
 |
By vegac - Posted on November 2, 2003 - 23:23:01 (#9451)
Current version when comment was posted: 0.07 |
 |
 |
First, I had emailed you about this before...
I have your driver installed, but set a VESA mode set...it uses that VESA mode but seems to think that your driver is functional because it lets me change the resolution (however it breaks when I do so...)
I havent' yet tested to see if there's any hardware acceleration when running like this...
For items 2 and 3 on your must-do list, even if they were both values set in the config file that would be fine. The important thing is just being able to use it...and having to edit a config file once and forget about it is better than not having support.
I'd be more than happy to test support for you if needed as would some other people I'm sure. And really, if my laptop dies, it gives me an excuse to buy a new one :)
|
|
| Laptop internal panel support |
 |
 |
By rudolfc - Posted on November 2, 2003 - 21:00:51 (#9448)
Current version when comment was posted: 0.07 |
 |
 |
Hi there,
Thanks for the compliment :-)
In order to get flatpanel support going, there are a few things that need to be setup:
1.-flatpanel support registerprogramming
2.-flatpanel presence/absense detection
3.-flatpanel native resolution detection
1.
I think I saw some registerprogramming in the linux Xfree driver that I'll probably try to setup also. It will be hard or next to impossible probably for me to get that going correctly without being able to test on laptops myself though :-/
But I'll try anyway later on and hope I can find someone in my neighbourhood with a laptop a can test on.
OK, here's the really bad news: 2. and 3. cannot be done reliably I fear. This should probably be read from the nVidia VGA BIOS in some info structure (the same one that contains the CORE and RAM speeds probably). Here's what I plan to do:
2.
Already in place, based on the cardID. All 'Go' type of cards are assumed to be in laptops, the other cards are assumed to be 'normal' cards. There are some ID's that could be both a laptop or a normal card maybe though. For now I'll assume these are normal cards, and wait for feedback to proof otherwise. I got a hint from a user though that could be correct: those cards will not be used in laptops. So this would work.
3.
Detecting the native resolution is impossible I think. This means we are going to need a 'nv.settings' file 'preference' for this, that by default will be set to 640x480.
The user is required to find out what the native resolution of his (or her) flatpanel is, and modify the settings file accordingly.
640x480 is a 'safe mode' preset, that should prevent people from distroying their panels.
2. and 3. are required for the driver to know because it will shutdown and/or limit the drivers modes that can be set on the panel to be always 60Hz and always in the native resolution. Higher modes would need to be virtual screens, while lower modes could be 'centered' (not likely due to more register programming stuff needed: no/not enough info available)
Of course the simplest way to add support is to just add 1., and forget about the rest. The downside is that if the user selects a refreshrate above 60Hz, and/or a resolution above the panel's native resolution, this panel might get destroyed.
--------
OK, I hope this clarifies the situation a bit...
I plan to first setup dualhead support, followed by TVout. I want to do this first, because I can actually test this and I think I can pull it off :-)
Once this is implemented, I'll turn to flatpanel support. And I hope that in the meantime I find someone over here with a laptop so I can test with it for flatpanel support.
I have noted that the driver should work on laptops as long as the external monitor port is used, because this is what I heard from several users through reports for several types. This does _not_ mean that there won't be laptops that still don't work with the driver at all, just like it is with 'normal' cards.
Personally I have to say, I am glad with how the driver works on most cards outthere though. I wouldn't have been surprised if I would have had much more difficulty getting a 'success rate' this high ...
Best regards,
Rudolf.
|
|
| Geforce 2 Go |
 |
 |
By DustBird - Posted on November 2, 2003 - 01:58:15 (#9440)
Current version when comment was posted: 0.07 |
 |
 |
Hello Rudolf, wonderfull work you are doing here!
I have tried your driver and it doesn't work for me with my GeForce 2 Go card, on my laptop. And your instructions mentions that my card only works with an external monitor.
Well, my question is, are you working on some kind of fix? And what can I do to help you fix this? I am willing to test and bug report, just say what I have to do.
Greets
|
|
| driver should work on all versions of BeOS AFAIK |
 |
 |
By rudolfc - Posted on November 1, 2003 - 14:38:46 (#9438)
Current version when comment was posted: 0.07 |
 |
 |
Hi,
the word 'open' in the name of the driver does not mean it will only work on openBeOS. It's known to work on:
R5.0.x
Dano
MAX
Zeta.
Rudolf.
|
|
| Go! Go! Go! Go! |
 |
 |
By C-Fu - Posted on November 1, 2003 - 14:32:21 (#9437)
Current version when comment was posted: 0.07 |
 |
 |
Finally there is light for us laptop users!!! :D Too bad it's just for OpenBeOS, but anyway take your time and keep up with the good work!
|
|
| Driver works fine. Finally, some 32 bit color.... |
 |
 |
By Zutecool - Posted on October 31, 2003 - 20:01:54 (#9432)
Current version when comment was posted: 0.07 |
 |
 |
My review is as such, Yeah!! damn it! Finally 32 bit color, 2D acceleration, and source code. VESA bye-bye.
BeOS 5.0.3 running on a 2 Gig HD, in a DELL Dimension 4550 with -
nVidia GeForce4MX420
Pentium 4 2.5Ghz
Sound Blaster Audigy 2
|
|
| nVidia BIOS flash for core and ram speed... |
 |
 |
By rudolfc - Posted on October 30, 2003 - 20:29:40 (#9424)
Current version when comment was posted: 0.07 |
 |
 |
Hi again Richard,
That's a very nice link you gave there! Just downloaded the new utility (runs in Windows). This is indeed what could do the trick.
Mind you, I would not overlclock! I would try to determine the official RAM and CORE speeds for your card, and then check what this utility says about the actual programmed speeds.
If these actual programmed speeds are much lower than what the official spec says, you could flash it (any higher setting than the current ones will improve the situation).
If you run the util (if only to determine the speeds that are currently programmed) please post your findings here! You can actually help me get on the right track here ;-)
Be carefull though.. If you flash the card wrongly you can probably throw it away. I won't be able to help out!
Best regards,
Rudolf.
|
|
| overlay distortion |
 |
 |
By shields - Posted on October 30, 2003 - 20:07:13 (#9423)
Current version when comment was posted: 0.07 |
 |
 |
Well first off, Rudolf, thanks for replying. The distortions do go away when played in a lower resolution workspace, so based on what you say it probably is a bandwidth problem.
Whilst switching workspaces is a perfect work-around for me, If your hardcore and/or (to my afraid mind) rediculously stupid enough, something like this;
http://www.short-media.com/forum/showthread.php?s=&threadid=120
may work for you. (That is if I actually understand whats causing the distortion)
Richard Shields (quite frankly, typed out so people dont think I've chosen shields as a nickname)
|
|
| Pardon, works OK with my card |
 |
 |
By Pink Chick - Posted on October 30, 2003 - 19:37:55 (#9421)
Current version when comment was posted: 0.07 |
 |
 |
You are right, Rudolph, sorry for that. I made a couple of stupid mistakes while installing.
|
|
| TNT overlay / distortions |
 |
 |
By rudolfc - Posted on October 30, 2003 - 19:04:43 (#9420)
Current version when comment was posted: 0.07 |
 |
 |
Shields,
You could try testing overlay on low res workspaces, with low colordepth. Also you could try lowres video. If you see the problem disappear with 'lower settings', the distortions are coming from a bandwitdh problem.
In order to fix this, I need to setup the RAM and CORE speed for the cards. Also I might need to setup arbitration in case the optimal RAM and CORE speed do not suffice. If this is the case, chances are the distortions will always remain, though less evident.
For instance Matrox G200 cards suffer from the same problem on high-res video (DVD).
These distortions are known to be visible on a few nVidia cards outthere, because apparantly nVidia clocks the cards slow on startup (VGA BIOS init) on some cards.
Another solution to this problem _might_ be flashing the VGA BIOS, and hoping that the new version clocks the card faster.
Be carefull though, you could end up with a non-functional card if you make a mistake with this!
Setting up CORE and RAM speed are on my wishlist, but it may be some time before I get around to (trying to) implement that. It's in fact part of a card 'coldstart' procedure. Little information is available for this.. :-/
Rudolf.
|
|
| FX5900 not working |
 |
 |
By rudolfc - Posted on October 30, 2003 - 18:52:11 (#9419)
Current version when comment was posted: 0.07 |
 |
 |
Hi,
Ingo, I just checked the driver's card identification stuff, your card should be recognized just fine. Chances are you did not correctly install the driver.
If you have the developertools installed, you can do a 'make install'. If not, you can manually copy the files to the correct locations on your HD. See the included README file for details.
Owners of cards not previously supported on BeOS can check if the driver is loaded by starting a terminal session, and type:
ls /dev/graphics
You should see at least two entries:
->The 'stub' entry, which is a 'fake' driver always there;
->An entry looking like "10de_0281_010000".
10de stands for nVidia's manufacturor ID, 0281 is the cardID (should be 0331 for you), and 01 00 00 stands for bus, card, function inside your system AFAIK. This part may vary a bit.
If you have this entry, the kerneldriver is loaded. If not, you did not install the driver (as the ID's are known by it).
If you have this entry, but the driver isn't working (still VESA mode), the accelerant is not loaded.
If the accelerant is loaded, you will get a file in your home folder called nv.accelerant.log. Depending on how much logging you enabled (minimal by default), the file grows slowly to very fast.
Hope this helps. Keep me posted!
Rudolf.
|
|
| Sorry, won't do on on FX5900 |
 |
 |
By Pink Chick - Posted on October 30, 2003 - 18:28:12 (#9418)
Current version when comment was posted: 0.07 |
 |
 |
A nvidia FX9500 won't be recognized, it's an 'ASUS V9950'.
Device Type : Display Controller (VGA)
Base: 3 Subtype: 0 Interface: 0
Vendor : Nvidia Corporation (0x10de)
Card ID : 331
|
|
| Thank you |
 |
 |
By shields - Posted on October 30, 2003 - 18:20:18 (#9417)
Current version when comment was posted: 0.07 |
 |
 |
Thank you for this, the overlay support on my tnt2 (I think Vanta) is very very welcome. I can actually watch divx movies larger divx movies now.
My machine has been on and acting as reliable as always for around 40 hours now, so no problems with stablity it seems.
There are some small problems (they dont actually bother me, so if Im the only one reporting them dont bother spend hours trying to fix them or anything), theres a little distortion when using overlay, tiny little white lines appearing randomly (only over video) and when colour depth is changed nv.accelerant.log is written to the home directory. This is with minimal logging by the way.
Again thank you I really appreciate it, I can finally watch some of the anime I've downloaded.
|
|
|
|
 |
 |
|
 |
 |
 |
 |
 |
 |
 |
|
|
 |
 |
 |
| |
Recent Downloads - # 62
Total Downloads - # 30
Total Views - # 16
User Ratings - # 12
|
 |
 |
 |
 |
| |
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... - 13,606
2. Realtek RTL8... - 13,115
3. Ati Radeon G... - 12,584
4. Ensoniq Audio... - 7,566
5. ATI Rage 128... - 7,442
6. USB Joystick... - 5,665
7. Broadcom 440x... - 5,411
8. S3 Trio 64 v2... - 4,727
9. USB Serial dr... - 4,708
10. Intel Extreme... - 4,471
|
 |
 |
 |
 |
| You are not logged in.
Login
|
 |
 |
 |
 |
|