Video out of range on GeForce 6100 (KM51G-754)

[richienyhus]

This is the C51G chipset.

When booting Haiku on a Albatron KM51G-754 , and about the time the red rocket is lit up, I get a error message from my LCD screen:

"Out Of Range

8. Frequency 80 kHz

22. Frequency 99.7 Hz"

when booting with video fail safe on, the everything is fine.

Tested on hrev27764

[richienyhus]

haiku in video failsafe mode

[richienyhus]

haiku video in normal mode

[richienyhus]

I forgot to list the motherboard's website ​http://www.albatron.com.tw/English/product/mb/pro_detail.asp?rlink=Specification&no=210

Also since its a GeForce 6100, shouldn't it be using that driver?

[jackburton]

Maybe the id isn't included in the list supported by our driver. Can you use "listdev" to retrieve the id ?

[axeld]

This is obviously a problem of the nVidia driver. I actually have the predecessor of that monitor (ViewSonic VP171b) :-)

[richienyhus]

listdev & listdev -d

[richienyhus]

From what I understand its 0x0242. yet that should of been fixed in hrev22219.

[richienyhus]

still there in hrev28781

[richienyhus]

in hrev28974

[rudolfc]

Hi there,

If the screens you use are 4:3 types, try setting the resolution of Haiku to the native one of your screen in failsafe video mode. Then reboot with the driver active. Problem solved? (would be nice ;-)

Bye!

Rudolf.

[richienyhus]

That workaround works under hrev30884. :)

I take it this will be fixed when EDID is fully implemented?

Would this also be a dup of #1696, #1557 or #1026 then ?

[rudolfc]

Hi,

To loose the problem, you could place the file nvidia.settings in your home folder, and modify it in the line that reads: force_ws true to read: force_ws false

If you now reboot, you should be able to use all normal 4:3 resolutions. Can you confirm this please?

This problem will be gone once DDC is in use. It will probably (at least partially) solve these other bugs too.

Bye!

Rudolf.

[richienyhus]

I downloaded nvidia.settings from here on trac (was it meant to have been in the file system?), and placed it in ~/config/settings/kernel/drivers, then rebooted as said.

The list of resolutions in Screen changed, but only one extra one is usable. So 1152x864 works fine, plus 1400x1050 works too, but it only shows part of the screen (1280x1024 is the highest this screen is meant to go anyway).

I can use 1024x768, 800x600 and 640x480 under other operating systems, but not with haiku.

[rudolfc]

Hi again,

What's the monitor you are using again? Can you name the brand and type? Do you connect via VGA or DVI?

And: can you enable logging (modify nvidia.settings for that) and upload the driver's logfile from your homefolder after a reboot?

Bye!

Rudolf.

[richienyhus]

Gday, its a ViewSonic. Haiku says its a VP171b-2, the model number on the screen just says VP171b. The screen has two VGAs and one DVI, My haiku box is plugged into the first VGA.

[rudolfc]

Thanks!

Your monitor looks like a ordinary type, 1280x1024 can be done at 75Hz for instance, and all normal VESA lower res modes should work.

Maybe you can test connecting your screen with a DVI cable if your card has a DVI connector. If so, I'd like to hear what happens then and I'd like a logfile from that as well.

On a sidenote: The driver is not able to init any I2C bus on your card. EDID will therefore not work from the driver. Strange. Looks like missing hardware specs. If I search the net I find linux users with the same sort of problem: I2C doesn't work on 6100...

However: if this doesn't work that still doesn't mean the driver can't properly use your card and screen. If all is right (apart from I2C).

Bye!

Rudolf.

[rudolfc]

Hi,

I am hoping the I2C buses now work on your card. Could you upload a new logfile from revision 30938 or later?

Thanks!

Rudolf.

[richienyhus]

video card logging 2

[richienyhus]

^{From hrev30941 GCC4}

The motherboard only has a VGA connector on it.

[rudolfc]

Aha, I see!

I hope you don't mind to upload another log, now from revision 30946 or later? There's a third I2C bus that might just work..

Thanks in advance :-)

Rudolf.

[richienyhus]

Video Card logging 3

[richienyhus]

Video Card logging 3

[richienyhus]

G'day

Sorry for the delay.

I added in nvidia.10de_0242_000500.0.3.log by mistake sorry.

Cheers.

[richienyhus]

I forgot to say this is from hrev30946 gcc2

[rudolfc]

Hi,

Thanks!

Hmm, still none of your I2C buses work. Strange. At the moment I have no further ideas of how to proceed here :-/

If I think of something I'll report back here..

Bye!

Rudolf.

BTW: would you have another screen there to test with to see if that works as it should?

[rudolfc]

Hi,

Since Haiku R31213 the nVidia driver uses DDC/EDID to detect ws screens that are connected using a VGA cable. Also the force_ws setting is reset to being false by default.

This should fix the dvi connected non-widescreen monitors shutting off/displaying blackness in non-native modes I think. Furthermore the driver now logs the pixelPLL programming settings done by the BIOS during startup, which could give me some important info on why some monitors shut off or detect a wrong refreshrate, so maybe it's interesting to upload another logfile from the current nVidia driver.

If the current driver doesn't solve the problem, it might be interesting to test a special version of the driver to see if a shutting-off monitor would remain on and display the desktop if the pixelPLL's aren't programmed at all.

See ​http://dev.haiku-os.org/ticket/1026#comment:27

You'll find a zip archive there with info and the special driver.

If anyone tests some more, please let me know the results.. :-)

Bye!

Rudolf.

[richienyhus]

Gday,

I do have another screen, but I have lent it to a family member for a good while. I will have to convince them that this will not affect their data, so it may take awhile to get back to you on that.

I will try the special version of the driver in the next few days, as for the normal driver I will add the log file after this message.

[richienyhus]

card logging 4

[richienyhus]

special version of the driver

[rudolfc]

Hi there,

You should set force_ws to false in nvidia.settings please, then retest the normal current driver.

with the normal driver from which you take card logging 4, do you have a picture? If you retry with force_ws false, do all resolutions upto 1280x1024 work ok?? If not, which ones don't work?

Bye!

Rudolf.

[rudolfc]

Hi once again,

Is the problem still existing? Otherwise I'd like to close this ticket.. If it is, maybe you can test the current default driver and upload a log from that.

IF that's not OK, you could also test these:

​http://dev.haiku-os.org/ticket/4084#comment:17

specially tweaked versions from the current driver, and upload logs from them as well.

I'm still trying to get those I2C buses to work...!

Bye!

Rudolf.

[rudolfc]

Hi once more richienyhus !

I might have a fix for the I2C detection problem. Could you please test the latest binary I uploaded to the above ticket as well? (first try the latest one please..)

Thanks for your time and effort in advance... :-)

Bye!

Rudolf.

[richienyhus]

Hi, sorry I haven't replied in the last week, I've been busy studying for tests. I tried to test yesterday but I only managed to get haiku booting in fail safe mode before something came up. I will have a free day on Thursday and will test then.

Sorry again ~Richie

[richienyhus]

current default driver

[richienyhus]

nvidia-improved-i2c-timing-0.95

[richienyhus]

0.94-test-i2c-noClkRead-inverse-wiring

[richienyhus]

nv_0.94-test-i2c-noClkRead

[richienyhus]

nv_0.92_pll_test

[richienyhus]

Current default driver = NV_0.94 from hrev32448.

All other drivers were also tested in hrev32448.

I'll test on another screen in a day or two, after my exams anyway.

cheers

~Richie

[rudolfc]

Thanks for the logs, It's now clear none of the testversions make your I2C buses work unfortunately.

Please do test that other monitor though, a log from the current default driver is enough, if you want to upload it.

Good luck with the rest of your exams!!

Rudolf.

[mmlr]

I see that some of the versions were tested with GCC4. The output of these are invalid, as before hrev32593 I2C sending was broken on GCC4.

[rudolfc]

Hi again,

Please upgrade to hrev32946. Does it work correctly now? Please upload a driver log as well.

Thanks!

Rudolf.

[rudolfc]

Hi there,

Please disregard the previous message and upgrade to hrev32958. Now the fix seems complete. Please test and let me know!

Thanks!

Rudolf.

[richienyhus]

The boot icons are half off the screen now, but it boots now without going out of range, So I'm happy for this ticket to be closed.

Also at 800 X 600 the screen goes all fuzzy at 60hz but not 70hz.

All screen resolutions that this screen hand handle work, but there are still some resolutions that are way too large for the screen in the list. I guess this is due to EDID failing ?

This is on r1a1.

Cheers for all your work! ~Richie

[rudolfc]

Hi there!

Good to hear from you :-)

So: all supported modes work perfectly in all resolutions: except the combination of 800x600 @ 60Hz?

---

If there are (much) more troublesome setups there's one more test that I can think of you should probably try:

Modify just one thing in the nvidia.settings file you were using to generate the latest log you uploaded: modify the line: switchhead false

To become:

switchhead true

Keep logging enabled. Remove the current logfile, and now reboot. Would you have a much more correct working picture now? If not, undo the change in the switchhead line. Let me know the result please!

Oh, also upload the new log created please.

About the list of modes: indeed, since your monitor can't be detected via EDID the max. resolution it can do is unknown. That's why there are too much there.

Thanks in advance!

Rudolf.

[richienyhus]

I tried switchhead true, but my screen turned off while booting, I grabbed the log via video safe mode.

Weirdly after changing back to switchhead false and rebooting the booting icons are centered again.

I just played around a bit more and it is not just 800x600 @60hz that is fuzzy, added an attachment with my findings.

[rudolfc]

Hi again.

Thanks for your tests/report! I expect the half off-screen booticons screen was caused by a modeset command to a resolution above 1280x1024 (that would happen if during the previous Haiku session you selected say 1400x1050: if you do that will be activated during the next boot.

Anyhow: please don't select modes above 1280x1024 as those are indeed not meant for your screen, these could do harm even, if left enabled for too long a time.

---

About the report you wrote: you wrote down refreshrates. Did those rates match the rate reported by your screen (on-screen menu), or did you not verify that?

If you did not verify that, it would be nice if you could still do that for a few modes (a few OK, and a few fuzzy).

Also, can you take a (sharp-enough) snapshot of a fuzzy picture using a cell-phone or so and upload it here? I'd like to literally see the problem myself if possible..

I am now suspecting the VCO in the PLL is programmed out of range for a part. Please check the onscreen menu for me in the modes?

Thanks!

Rudolf.

[rudolfc]

Oh, yes: If you are doing this also upload the driver logfile! It will contain all settings the PLL is using in the modes you test. It would be nice if you could reboot a number of times while different modes are set, since also the PLL progrogramming the BIOS did will be dumped then before being overwritten by the driver...

Bye!

Rudolf.

[rudolfc]

OK, OK, one more thing: please change the nvidia.settings file for 'switchhead false' and '#logmask 0x0027100f'

This will hopefully log just enough info so I can search faster. (setmode, crtc1/2, dac1/2)

Bye!

Rudolf.

[tbe]

Hi Rudolf,

I have the same chipset on an ASRock K8NF4G-SATA2 motherboard, and experience similar problems in that I2C/EDID does not work and the screen goes black when the nvidia driver kicks in. This is with the R1/alpha1 release (hrev33109).

My screen is a Samsung SyncMaster 710N (MJ17ASKS/EDC) with 1280x1024 native resolution. The motherboard only has VGA out and the screen only VGA in, so DVI is not an option.

Originally, the OSD would show "44.5kHz 55Hz NN", with no resolution mentioned, but I know from the driver log that this was 1024x768. I don't have that log at the moment because after selecting 1280x1024 from the boot menu, it keeps using that resolution, for which OSD says "59.3kHz 56Hz NN". I'll attach the syslog and driver log below. This is with "force_ws false" by the way, but that doesn't seem to make any difference.

In VESA mode, things have always been fine, and OSD says "63.8kHz 60Hz PP 1280x1024", which is exactly the same as on the boot screen. I'll attach the syslog for this as well. If I lower the resolution using the Screen preflet, OSD tells me "48.3kHz 60Hz NN 1024x768".

Thanks, Truls

[rudolfc]

note to self: richienyhus has one VGA port only on his mainboard (checked manual PDF)

Rudolf.

[rudolfc]

Hi there tbe and richienyhus,

Would you (or one of you) be at Begeistert the coming weekend? If so, it would be very interesting if you could bring along the 6100 system so I can test it (I'll be there if all is right).

Bye!

Rudolf.

[richienyhus]

No, sorry I'm not. I'm from down under, so it is too far away.

[rudolfc]

nvidia driver 1.07: added fix for gf 6100/6150/6200 black screen

[rudolfc]

Hi again,

Well, that's too bad. It was a nice weekend! Anyhow, I have an update to the driver I'd like you to test.

Could you test the latest driver (hrev33966), also uploaded to this ticket, and see what it does for you?

Thanks!

Rudolf.

PS: A driver logfile is appreciated..

[richienyhus]

fuzzy 800x600 60mhz

[richienyhus]

ok 800x600 72mhz

[richienyhus]

fuzzy 1152x864 60mhz

[richienyhus]

ok 1152x864 85mhz

[richienyhus]

fuzzy screen1

[richienyhus]

screen2 fuzzy

[richienyhus]

screen2 non-fuzzy

[richienyhus]

Hi Rudolf,

Sorry It has taken me 7 months to upload these, I had lent this PC to someone after Haiku stoped working on it. After R1A1 I asked for it back to try, and Haiku seems to be working well on it again.

---

I don't know if this Info, from the monitor, will be of any help:

Fine while at 72Hz Refresh rate:
H. Frequency:	71kHz
V. Frequency:	66.7Hz
Resolution:	1280x1024
Pixel clock:	123 MHz

And:

Fuzzy while at 75Hz Refresh rate:
H. Frequency:	72kHz
V. Frequency:	67.8Hz
Resolution:	1280x1024
Pixel clock:	122 MHz

(Double checked the numbers and found it can sometimes be this:)

Fuzzy while at 75Hz Refresh rate:
H. Frequency:	71kHz
V. Frequency:	67.1Hz
Resolution:	1280x1024
Pixel clock:	120 MHz

[scottmc]

So has this one been fixed? Can you recheck it with a recent build?

[richienyhus]

I was still having the fuzzy problem late last year. I am not able to test this hardware for awhile as it is in storage, and even then I don't know if it has been damaged in one of the earthquakes that has plagued my city in the last year.

The person in #7659 seems to think the first(out of range) problem is still happening.

[rudolfc]

Hi,

The specific chipset in question is an onboard one. Since I don't have a mainboard with this chipset I cannot test. Chances are that it remains a problem therefore. There are a few more reported bugs with this chipset, for them the same goes.

Bye!

Rudolf.

[rudolfc]

Hi,

So from the looks of it we have a PLL related problem (the lock does not aquire/hold). Might be that the other 6100's without picture (for instance) suffer from the same problem, but more intense so to speak..

Card needed to do actual testing on to fix the 6xxx problems I think.

Bye!

Rudolf.

[rudolfc]

We have a general GF 61xx card problem, if I remember correctly,this is a card-type which is integrated on the mainboard of systems. I need such a mainboard to be able to test this card type. Which did not happen yet unfortunately.

The GF61x0 problem is the last remaining item which should be fixed in the existing nVidia Haiku driver.

Related tickets: #2780, #4670, #6292, #12955.

[rudolfc]

Just a heads-up:

I was able to get my hands on a GF6100 mainboard the other day, so I plan to test it first hand myself and hopefully fix the remaining errors for these cards.

[rudolfc]

Hi,

Finally I haveHaiku running on this system. driver works OK apart from the PLL calculations, therefore has wrong refreshrate, 'no picture', or 'wobbling screen'. I'll dive into this, this should be fixable.

[maquak]

I was recently digging into mode settings for #16413. The main code for setting mode to use (including refresh rate parameters) is in ScreenMode.cpp: ​https://git.haiku-os.org/haiku/tree/src/preferences/screen/ScreenMode.cpp#n628

You can use screenmode app for easy testing and debugging as it calls this class under the hood.

[rudolfc]

This is merely the 'calling code' to the driver which actually handles modesetting. If a mode can't be set therefore you need to search in the driver that's in use on your system. In this case the fault is in the nvidia driver, specifically for NV44 architecture gfx cards, that are incorporated on AMD CPU based mainboards, I guess now some ten years ago. I found the reason meanwhile: these mainboards have a different Crystal base frequency than other nVidia cards. Will patch the driver asap.

[rudolfc]

For Geforce 6000-6200, notably Nforce 4 and 4xx, but possibly more cards, added detection of a 25Mhz GPU base frequency crystal. This fixes too low refreshrates, wobbly screens, no picture at all, only part of a picture and that kind of trouble.

Fixed in hrev55039.

Closing ticket.

If the problem still exists the ticket can be reopened.

Thanks!