fast, stop, fast, stop, fast, crash! (OpenGL)

[shish]

I've just installed the latst drivers, and all my OGL apps have jumped from 0.05 FPS to around 50, however after about a second or running, they pause, then run fine, then pause, then run fine, and then the program crashes after recieving a SIGABRT (the "abort" signal)

All my OGL screensavers, blender, and zsnes (after being set to OGL mode) do this, whereas before they were all perfectly stable but *really* slow.

Also when this happens sometimes some of the screen (the odd row of pixels) goes staticky.

It all worked under windows quite happily with default drivers, but my disk's all fscked up and 90% of the windows fat got corrupted :mad:, so I can't boot into it to check settings.

There's nothing in the XF86 logs about it, but I doubt it's anything to do with X anyway

[Bino]

What Distro are you using, and what games are you playing when this happens?

I know for me in SuSE 8.0+ I have to ps -A in konsole to kill quake3.x86 or wolf.x86 depending on what game, to shutdown the agpart, you could see if they are running, I know for me when i forgot to close it the sound would fade out and the gl would stop after trying to reload the game again

[shish]

Dude, that... didn't seem to have anything to do with my question... let me rephrase:

On ALL OpenGL programs, they go fast, then pause. This happens a couple of times, then the program exits after recieving a SIGABRT from the operating system (the abort signal)

On default "nv" drivers everything went very slow, but was stable, now it's very fast but crashes after about 3 seconds of using OGL

It's mandrake 8.1 download edition btw

[Bino]

here is what I mean, I meant maybe you should check to see if they files are nt running in the background still after they crash or close becuase that is just goign to cause more problems, I knew you where meaning all open GL apps, but I wasn't going to sit here and type everysingle app in my list I just said quake3 and wolf becuase they use open GL, sorry for the confusion on your post, but you seemed really rude in your reply...just trying to help..

what you could try doing is changing to run lvl 3 to kill the x
then startx /"directory"/"program"/

not sure if it will help but it works for me when games are runnig a little slower then normal, i just kill the x and then start it again w/ just the game runnign in it and it gives better performance

[shish]

> here is what I mean, I meant maybe you should check to see if they files are nt running in the background

nothing abnormal...


> I just said quake3 and wolf

I don't *have* either of those (q3 on win32, but no linux binary), so it looked like you replied to the wrong thread


> but you seemed really rude in your reply...just trying to help..

eh, sorry about that. with a global post count into many thousands, I'm turning into a grumpy old man of the internet


> what you could try doing is changing to run lvl 3 to kill the x
then startx /"directory"/"program"/

the software rendering mode is slightly faster (30FPS from 25), but OGL is still fast-slow-fast-crash. And it's stuck in the top left, which gives me another idea --

Fullscreen software is a load faster, but warped unless the software resolution and X display match. zsnes only does software up to 512x4xx, while my X is 1024x768 - is there an easy way to get X into a smaller res and back without lots of logging in as root and going into the control panel?

[merlin42]

It sort of sounds like a hardware problem.

It may be that some HW problem (heat maybe?) developed and that is why windows went in the crapper, and is also why you can't get reliable 3d acceleration from your vid card.

Things I might try (not necesarily in this order):
0) reinstall newest drivers
1) try older driver versions.
2) Reinstall windows and see if it works.
3) open the case and look for things like loose wires ( I once had a video card that seemed to be bad until I opened the case and saw that the wires going to the internal speaker had come off and were touching the pins on an ic on the vid card.)
4) run with the case open
5) get lm_sensors http://secure.netroedge.com/~lm78/ or similar going and see what the internal case temp is.
6) get nvclock www.evil3d.net/products/nvclock/ and try underclocking a bit.

That being said these are all pretty much outside chances, but since it would seem you have touched on all the usual suspects this is all I can come up with.

[Andy Mecham]

Heat issues would produce symptoms like the ones you're seeing, but i'd like to exhaust other avenues first.

Can you post your log, and anything interesting in /var/log/messages?

I don't *have* either of those (q3 on win32, but no linux binary), so it looked like you replied to the wrong thread

If you have the original win32 cds, you can just install the latest linux point release (1.32b, AFAIR) then copy your .pk3 files to /where/you/installed/quake3/baseq3.

--andy

[shish]

Logs be here:

[Andy Mecham]

(WW) System lacks support for changing MTRRs

Can you add MTRR support to your kernel to see if that helps?

--andy

[shish]

MTRRs? What are they and how do I add them?

Also I'm having trouble with installing a new HD, and currently I've only got ~40MB spare. It should be sorted soon, but I don't think that there's enough room for a kernel compile quite yet...

[merlin42]

shish - Before we go around in more circles could you please give a COMPLETE description of the machine in question. I would like CPU, Motherboard, RAM, linux distro, anything else that might be pertinent.

According to the PCI IDs in your log files you are running an Aladdin V based motherboard, this is a super socket 7 board which supports many different CPUs, some of which (eg early k6-2s and regular old pentuims) do no support MTRRs.

[shish]

> a COMPLETE description
I don't know if I can get more COMPLETE than this, bar reading the ID numbers off the chips...

Misc Hardware:
~~~~~~~~~~~~~~
2GB fujitsu HD
160MB RAM (32 + 128)
Creative 32x CD drive
ASUS P5A-B rev 1.03
Award BIOS 4.51PG
Acer AcerView 34TL

Misc Software:
~~~~~~~~~~~~~~
KDE / Fluxbox

glxgears gives:
1 sec 55FPS, 2 sec pause, 1 sec 55FPS, 1 sec pause, crash (SIGABRT)
average 22FPS

XFree86 Version 4.2.0 / X Window System
(protocol Version 11, revision 0, vendor release 6600)
Release Date: 23 January 2002
If the server is older than 6-12 months, or if your card is
newer than the above date, look for a newer version before
reporting problems. (See http://www.XFree86.Org/)
Build Operating System: Linux 2.4.18-1mdksmp i686 [ELF]
Module Loader present

1024x768, 16 bit


/proc/cpuinfo:
~~~~~~~~~~~~~~
processor : 0
vendor_id : AuthenticAMD
cpu family : 5
model : 7
model name : AMD-K6tm w/ multimedia extensions
stepping : 0
cpu MHz : 267.282
cache size : 64 KB
fdiv_bug : no
hlt_bug : no
f00f_bug : no
coma_bug : no
fpu : yes
fpu_exception : yes
cpuid level : 1
wp : yes
flags : fpu vme de pse tsc msr mce cx8 mmx
bogomips : 532.48

/proc/version:
~~~~~~~~~~~~~~
Linux version 2.4.18-6mdk (quintela@bi.mandrakesoft.com)
(gcc version 2.96 20000731 (Mandrake Linux 8.2 2.96-0.76mdk)) #1
Fri Mar 15 02:59:08 CET 2002

/proc/drivers/nvidia/version:
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
NVRM version: NVIDIA Linux x86 nvidia.o Kernel Module 1.0-4349
Thu Mar 27 19:00:02 PST 2003
GCC version: gcc version egcs-2.91.66 19990314 (egcs-1.1.2 release)

/proc/drivers/nvidia/agp/card:
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Fast Writes: Not Supported
SBA: Not Supported
AGP Rates: 4x 2x 1x
Registers: 0x1f000007:0x00000000

/proc/drivers/nvidia/agp/host-bridge:
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Host Bridge: Acer Laboratories Inc. [ALi] M1541
Fast Writes: Not Supported
SBA: Supported
AGP Rates: 2x 1x
Registers: 0x1c000203:0x00000000

/proc/drivers/nvidia/agp/status:
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Status: Disabled

/proc/drivers/nvidia/cards/0:
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Model: GeForce2 MX/MX 400
IRQ: 11
Video BIOS: 03.11.01.44.00
Card Type: AGP

glxinfo:
~~~~~~~~
name of display: :0.0
display: :0 screen: 0
direct rendering: Yes
server glx vendor string: NVIDIA Corporation
server glx version string: 1.3
server glx extensions:
GLX_EXT_visual_info, GLX_EXT_visual_rating, GLX_SGIX_fbconfig,
GLX_SGIX_pbuffer
client glx vendor string: NVIDIA Corporation
client glx version string: 1.3
client glx extensions:
GLX_ARB_get_proc_address, GLX_ARB_multisample, GLX_EXT_visual_info,
GLX_EXT_visual_rating, GLX_EXT_import_context, GLX_SGI_video_sync,
GLX_SGIX_swap_group, GLX_SGIX_swap_barrier, GLX_SGIX_fbconfig,
GLX_SGIX_pbuffer, GLX_NV_float_buffer
GLX extensions:
GLX_EXT_visual_info, GLX_EXT_visual_rating, GLX_SGIX_fbconfig,
GLX_SGIX_pbuffer, GLX_ARB_get_proc_address
OpenGL vendor string: NVIDIA Corporation
OpenGL renderer string: GeForce2 MX/PCI
OpenGL version string: 1.4.0 NVIDIA 43.49
OpenGL extensions:
GL_ARB_imaging, GL_ARB_multitexture, GL_ARB_point_parameters,
GL_ARB_texture_compression, GL_ARB_texture_cube_map,
GL_ARB_texture_env_add, GL_ARB_texture_env_combine,
GL_ARB_texture_env_dot3, GL_ARB_texture_mirrored_repeat,
GL_ARB_transpose_matrix, GL_ARB_vertex_buffer_object,
GL_ARB_vertex_program, GL_ARB_window_pos, GL_S3_s3tc, GL_EXT_abgr,
GL_EXT_bgra, GL_EXT_blend_color, GL_EXT_blend_minmax,
GL_EXT_blend_subtract, GL_EXT_clip_volume_hint,
GL_EXT_compiled_vertex_array, GL_EXT_draw_range_elements,
GL_EXT_fog_coord, GL_EXT_multi_draw_arrays, GL_EXT_packed_pixels,
GL_EXT_paletted_texture, GL_EXT_point_parameters, GL_EXT_rescale_normal,
GL_EXT_secondary_color, GL_EXT_separate_specular_color,
GL_EXT_shared_texture_palette, GL_EXT_stencil_wrap,
GL_EXT_texture_compression_s3tc, GL_EXT_texture_cube_map,
GL_EXT_texture_edge_clamp, GL_EXT_texture_env_add,
GL_EXT_texture_env_combine, GL_EXT_texture_env_dot3,
GL_EXT_texture_filter_anisotropic, GL_EXT_texture_lod,
GL_EXT_texture_lod_bias, GL_EXT_texture_object, GL_EXT_vertex_array,
GL_IBM_texture_mirrored_repeat, GL_KTX_buffer_region, GL_NV_blend_square,
GL_NV_fence, GL_NV_fog_distance, GL_NV_light_max_exponent,
GL_NV_packed_depth_stencil, GL_NV_pixel_data_range, GL_NV_point_sprite,
GL_NV_register_combiners, GL_NV_texgen_reflection,
GL_NV_texture_env_combine4, GL_NV_texture_rectangle,
GL_NV_vertex_array_range, GL_NV_vertex_array_range2, GL_NV_vertex_program,
GL_NV_vertex_program1_1, GL_NVX_ycrcb, GL_SGIS_generate_mipmap,
GL_SGIS_multitexture, GL_SGIS_texture_lod
glu version: 1.3
glu extensions:
GLU_EXT_nurbs_tessellator, GLU_EXT_object_space_tess

visual x bf lv rg d st colorbuffer ax dp st accumbuffer ms cav
id dep cl sp sz l ci b ro r g b a bf th cl r g b a ns b eat
----------------------------------------------------------------------
0x21 16 tc 0 16 0 r y . 5 6 5 0 0 16 0 16 16 16 16 0 0 None
0x22 16 dc 0 16 0 r y . 5 6 5 0 0 16 0 16 16 16 16 0 0 None
0x23 16 tc 0 16 0 r . . 5 6 5 0 0 16 0 16 16 16 16 0 0 None
0x24 16 tc 0 16 0 r y . 5 6 5 0 0 0 0 16 16 16 16 0 0 None
0x25 16 tc 0 16 0 r . . 5 6 5 0 0 0 0 16 16 16 16 0 0 None
0x26 16 dc 0 16 0 r . . 5 6 5 0 0 16 0 16 16 16 16 0 0 None
0x27 16 dc 0 16 0 r y . 5 6 5 0 0 0 0 16 16 16 16 0 0 None
0x28 16 dc 0 16 0 r . . 5 6 5 0 0 0 0 16 16 16 16 0 0 None

[bwkaz]

Why are you using egcs to compile the driver?

Note that you have a compiler mismatch between your kernel and the nVidia module. This is known to not work in a lot of cases. Let me guess -- you used IGNORE_CC_MISMATCH, right? Don't do that -- fix the problem by installing the right compiler.

[shish]

> Why are you using egcs to compile the driver?
> Note that you have a compiler mismatch between your kernel and the nVidia module. This is known to not work in a lot of cases. Let me guess -- you used IGNORE_CC_MISMATCH, right? Don't do that -- fix the problem by installing the right compiler.

Wha? I am? I thought I just ran the installer and it popped up "installed ok"

echo $CC
/usr/bin/gcc

actually I just made that up, but I've been using gcc for everything, and I'm pretty sure I don't even *have* egcs

[mtrr]

shish: the module is compiled with egcs, I assume the kernel also,
do a cat /proc/version and you'll see that. and also do a
gcc --version

Andy: MTRR on a K6? do they exist there? IIRC not in the same
way as on intel chips and later K7...


cheers

[bwkaz]

Originally posted by shish
/proc/drivers/nvidia/version:
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
NVRM version: NVIDIA Linux x86 nvidia.o Kernel Module 1.0-4349
Thu Mar 27 19:00:02 PST 2003
GCC version: gcc version egcs-2.91.66 19990314 (egcs-1.1.2 release) This is why I think you're using egcs -- it says so. ;) Your kernel, however, was actually compiled with gcc. Note your /proc/version:

Linux version 2.4.18-6mdk (quintela@bi.mandrakesoft.com)
(gcc version 2.96 20000731 (Mandrake Linux 8.2 2.96-0.76mdk)) #1
Fri Mar 15 02:59:08 CET 2002

My /proc/driver/nvidia/version looks like:

NVRM version: NVIDIA Linux x86 nvidia.o Kernel Module 1.0-4349 Thu Mar 27 19:00:02 PST 2003
GCC version: gcc version 3.2 and my /proc/version has this in it:

Linux version 2.4.20 (avatar@beta) (gcc version 3.2) #5 Thu Mar 20 13:24:20 EST 2003 Hold on, wait a minute. We both compiled nvidia.o at the same time? (Mar. 27, 7 PM Pacific Time, before the release?) That doesn't make any sense... umm... I don't know. Maybe that file (/proc/driver/nvidia/version) is reported by the precompiled nv-linux.o file then... hmm.

[Andy Mecham]

Maybe that file (/proc/driver/nvidia/version) is reported by the precompiled nv-linux.o file then...Yes.

--andy

[bwkaz]

Ah, that makes more sense then. Are there gcc 2.95 versions of nv-linux.o anywhere?

shish -- is the nVidia card sharing an IRQ with anything else? Check /proc/interrupts to see if any other kernel modules have registered interest in IRQ 11.

[shish]

Nope, nothing else on IRQ 11. I've tried rebuilding the kernel with MTRR support, but I can't get it to boot:

Loading linux-shish......................
<screen goes blank>

the command being

make menuconfig
make dep && make bzImage && make modules && make modules_install
<move kernel to /boot/, edit lilo.conf>
lilo