Go Back   nV News Forums > Linux Support Forums > NVIDIA Linux

Newegg Daily Deals

Reply
 
Thread Tools
Old 10-07-04, 07:00 PM   #49
Soluzar
Registered User
 
Join Date: Oct 2004
Posts: 34
Default Re: Finally stable 61.11? [patch inside]

Quote:
Originally Posted by PrakashKC
The AGP Drive Strength is only customizable (otherwise one would never be able to get it stable) with "fantastic" VIA chipsets and using AGP 4x.... Other chipsets know how to implement AGP 4x properly...
That's... interesting. I don't know why everyone knocks VIA chipsets so much. I haven't had the slightest cause for complain stability-wise, other than before I realised I had to change the AGP Driving Value, and in that case, the problem was between the chair and the keyboard, since the nVidia README file tells you to do so. :P

The reason I say interesting is because I my card is AGP 8x and I still found that tweaking the AGP Driving Value made a significant difference to the stability of my box. You sure it only affects 4x, PrakashKC?
Soluzar is offline   Reply With Quote
Old 10-07-04, 07:43 PM   #50
ScofieldKid
Registered User
 
Join Date: Aug 2004
Posts: 21
Default Re: Finally stable 61.11? [patch inside]

OK. I'm thinking this is an AGP bug with the K8W or a power issue. I even get some artifacting / flashes on this RADEON board I have in the system now, in 2D xorg driver.

I did notice there is now an even newer BETA for the K8W BIOS, if anyone else wants to play. For me, the obvious workaround is to yank out an opteron and find a K8T800 or whatever single-proc board. K8W is never going to be stable doing AGP video with anything... as far as I can tell...

ftp://ftp.tyan.com/bios_beta/2885_202w_beta.zip
ScofieldKid is offline   Reply With Quote
Old 10-11-04, 01:16 PM   #51
thedarkavenger
Registered User
 
Join Date: Oct 2004
Posts: 1
Default Re: Finally stable 61.11? [patch inside]

Ok so a long time ago when the 4600ti had just hit the shelves i ran along and bought one. plugged it in to my beasty (at the time) board and had a agp 4x party in my pants going on. this card broke down to the fact that my mate just can't catch but it apears that the floor can.

so little while ago i got myself a mx440 with a whole 64mb and 8x. *Sigh* lack of money but it was either that or sell my soul to the ati devil. now i have some groovy epox nforce 2 board and still the same card but that time has come to buy a new one and my question is this.

im running linux-2.4.27 slackware-current the latest xorg mondey can't buy. now of course from what i could see the pm timer overide was only availible in the 2.6 series of kernel so i recompiled that and still have the same problem as a lot of you.

mouse moves but everything else aint ..

in other words if i buy a geforce FX will i get the same lame ability to just use the mouse?

using the open source drivers is all very well and good but i can feel the card open up and run like a speed demon when its got the proper drivers. of course this is for 10 mins.

there anything i should either be trying or do i have to look at ati cards?
thedarkavenger is offline   Reply With Quote
Old 10-12-04, 04:22 AM   #52
PrakashP
gentoo ~x86_64 user
 
PrakashP's Avatar
 
Join Date: Jul 2004
Location: Germania
Posts: 213
Default Re: Finally stable 61.11? [patch inside]

So, after some weeks usage (though I don't run the machine 24/7) I can say, that this combination is stable for me on nforce2 with gf4ti (all settings are a must, changeing one leads to unstability):

AGPGART
Fastwrites OFF
PM Timer

Using NvAGP with above, leads to the illfamous mouse moves, system frozen, xorg/xfree taking 100% CPU bug.
Using Fastwrites or tsc as timing source leads to hard lock ups.

I run AGP at 8x with SBA enabled (because of AGP 3.0, this is mandatory).

As said, I can even use the experimental NForce2 CPUFreq driver which changes the FSB on-the-fly. But this likes to lock-up without applied delays (look tino source). But using delays the system gets sluggish during FSB changes. So instead of changing the FSB dynamically, I made three profiles for minimal, nominal and gaming performance. Then the delay doesn't make problems and I still can save energy when I want to.
PrakashP is offline   Reply With Quote
Old 11-06-04, 04:13 PM   #53
ScofieldKid
Registered User
 
Join Date: Aug 2004
Posts: 21
Default Re: Finally stable 61.11? [patch inside]

Found this interesting errata on the AMD 8151 chip today. This is the chip that implements the AGP-to-HypterTransport bridge on the Tyan Thunder K8W S2885. I was trying to figure how they architected AGP into HT, when I expected that both HT channels would be used up by 1) CPU-to-CPU and then 2) by CPU-to-8131 for I/O.

In any case, it seems that they are suggesting that there are some problems with the AMD8151. Check out the errata here:

http://www.amd.com/us-en/assets/cont...docs/25912.pdf

I don't have an NVIDIA card in my box anymore. But in any case, I've set all of the HT link settings I can to 600MHz. I'm just going to wait and see if that makes the box have a good uptime. I would be interested if anyone else moves to 600MHz if they see any of these lockup/crash/black screen problems go away.

Note: for any other frustrated dual-proc users, I can see a few alternatives for moves. The ASUS SK8V looks like a nice Socket-940
http://www.asus.com/prog/spec.asp?m=SK8V&langs=09
So that would get you there, with the K8T800 and VIA chipset, and it appears it is registered ECC so you could move your memory as well. Another way to go would be to move to a non-AGP board, like the ARIMA boards. And finally, it may be the case that punting the AGP and using a PCI slot for the graphics card would get you there. I wouldn't be surprised if that worked as well. Just tossing out some alternatives for anyone else who has been frustrated by the Tyan Thunder K8W.

Last edited by ScofieldKid; 11-06-04 at 04:49 PM.
ScofieldKid is offline   Reply With Quote
Old 11-24-04, 11:16 AM   #54
ScofieldKid
Registered User
 
Join Date: Aug 2004
Posts: 21
Default Re: Finally stable 61.11? [patch inside]

Thought I would followup to this last test. Initial testing showed a working system. But after a day, the black screen behavior showed again. If I were a betting man, I would guess that the AMD8151 is the cause, but that is just a guess.

I'm moving off of TYAN, and switching to some ASUS K8T800 platforms.

AGP 3D Graphics + Tyan K8W + Linux = black screen.
ScofieldKid is offline   Reply With Quote
Old 12-29-04, 05:12 AM   #55
ScofieldKid
Registered User
 
Join Date: Aug 2004
Posts: 21
Default Re: Finally stable 61.11? [patch inside]

OK. Having some good news right now. It appears that modding the BIOS settings for AGP Drive Strength, I think they call it "P Strobe" or "N Strobe", seems to have me finally up and running solidly. Here's the background on this:

http://www.rojakpot.com/default.aspx?location=9
http://graphics.tomshardware.com/gra...html#solutions
http://www6.tomshardware.com/howto/2...tuning-18.html
http://www.ertyu.org/~steven_nikkel/...atibility.html

So... I can run solid 2D-only for a month, no problem. Initiate 3D, whamo.
Setting all those values to the MAX produced screen tear and lockup. So just
on a whim, I set them all at the midpoint values. If the max was 31, I set it to 15.
If the max was 63, I set it to 31.

Eureka. I believe this is the fix. Note: the default setting in the TYAN K8W BIOS for these is quite vague, hence the links provided above. It's still a bit of a guessing game even when you know what these are. The default is "AUTO". That was the setting I had during all of my previous badness. Now with these fixed settings, it looks solid for 3D. I'll keep watching and report if anything bad happens.

Long hard road.

Also note: the "nvclock" utility was helpful in pulling out basic info. That's a cool
think and it is sourceforge CVS available, so get it through CVS. The CVS/latest version is much more up to date about which cards it can see.

(P.S.) Might be nice for someone at TYAN to clue in on this problem. The MOBO is wmsensormon capable under Linux, so it seems like somehow we could get the voltages and find out if there is an under-voltage condition on the AGP bus. If the
AUTO function is too weak, that might be it. Or if the fact that they are trying to build
a slot that can do too many things, i.e. AGP v. AGP Pro v. AGP XXX, maybe they should clarify these issues a bit more in their documentation.

I like the platform. Just wasn't fun using voodoo to divine the secrets of the AGP settings.
ScofieldKid is offline   Reply With Quote
Old 01-10-05, 05:24 PM   #56
ScofieldKid
Registered User
 
Join Date: Aug 2004
Posts: 21
Default Re: Finally stable 61.11? [patch inside]

Another followup. Well, no, drive strength didn't help. It is pretty weird that I can reboot, and be good for a day, but then the next day it will die.

I'm trying a different card now. This time a ASUS 6800 GT, the wimpy one. So far so good. I'll post back if this goes south. Of course, if this really is an AGP Hyptertransport bridge issue on the K8W, it won't matter which card I use.

I have some other boards to test with now... different chipset single-proc. So if this goes south, I'll test with some of those.

NVIDIA + LINUX is still the only game in town, so I'll keep plugging away. It is very cool that we are finally getting some good games and some solid drivers...
ScofieldKid is offline   Reply With Quote

Old 01-11-05, 08:23 PM   #57
ScofieldKid
Registered User
 
Join Date: Aug 2004
Posts: 21
Default Re: Finally stable 61.11? [patch inside]

OK. Testing now on a single-proc Opteron 146.
Motherboard is ASUS SK8V.
Graphics now ASUS 6800 GT (the whimpy one...

For other folks who have moved to FC3, I got the new kernel and compiled, and turned off all of the AGP modules in the build, which succeeded in preventing a foreign AGP module load of agpgart. Kernel compile steps were:

http://download.fedora.redhat.com/pu...37_FC3.src.rpm
rpm -ivh *src.rpm
cd /usr/src/redhat/SPECS
rpmbuild -bp --target=i686 kernel-2.6.spec
cd /usr/src/redhat/BUILD/kernel-2.6.10/linux-2.6.10
make xconfig
make
make modules_install
make install

Note: getting about 9920 FPS in glxgears.
That seems like a very good thing...
ScofieldKid is offline   Reply With Quote
Old 01-12-05, 08:14 PM   #58
ScofieldKid
Registered User
 
Join Date: Aug 2004
Posts: 21
Default Re: Finally stable 61.11? [patch inside]

Yeah, this ASUS SK8V + Opteron 146 is working quite well.
No problems at all now. As usual, ASUS working well. I've
had a lot of good experience with their stuff. End of story.
ScofieldKid is offline   Reply With Quote
Old 01-13-05, 03:42 AM   #59
zander
NVIDIA Corporation
 
zander's Avatar
 
Join Date: Aug 2002
Posts: 3,740
Default Re: Finally stable 61.11? [patch inside]

@ScofieldKid: thanks for the update; it's good to hear that you've found a working configuration, at last.
zander is offline   Reply With Quote
Reply


Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Forum Jump


All times are GMT -5. The time now is 01:05 AM.


Powered by vBulletin® Version 3.7.1
Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
Copyright 1998 - 2014, nV News.