HD 3850 AGP lockup with OpenGL load

[maddoc]

I tried hard to make 3D acceleration work for my Powercolor HD 3850 AGP on the following system:

Intel 865 mainboard (Gigabyte GA-8I848P-G)
P4 2,8 GHz 512MB RAM

I used fglrx drivers from Catalyst versions:
8.7
8.8
8.10
8.11
8.12
9.1betaRC2

- on Ubuntu 8.10 kernel 2.6.27-9-generic
- on Debian/Lenny kernel 2.6.26-1
by building debs from the installer, plus one time direct install.
(cleanup between installs by dpkg -r).

everything installs OK, AGP initializes to 8x (intel_agp), DRI initializes, no real problems in Xorg.log visible except several warnings like
...
(WW) AIGLX: 3D driver claims to not support visual 0x72
...

now the problem is: every time I start an OpenGL app (glxglears or fgl_glxgears will do), the system freezes:
With earlier drivers from Catalyst 8.8, fgl_glxgears runs for about 10 seconds; the more recent driver I use, the faster the freeze takes place:
With 8.8 I can run
"fgl_glxgears -loop 500"
several times successfully, until system finally freezes in the 10th or so run.
With 9.1betaRC2 the cube just slightly starts turning and the system immediately locks up.
This somehow makes me unsure when hoping for upcoming driver releases to resolve the issue....

I've been trying a hell lot of xorg Options and Option combinations from
http://debian.wgdd.de/temp/fglrx_man/fglrx.4x.html
without any noticeable change.

With 8.12 I have pretty good speed in normal 2D apps, fast scrolling with Firefox, moving it around with several tabs open is no prob, too. System runs without trouble until any OpenGL app is started. (OK, I didn't test user switching and suspend - I had enough with the normal stuff...).

On this same system, Catalyst 8.12 Windows XP drivers work fine, full AGP 8x speed, 3DMark06 runs to end without any errors, so I guess there's not a heat or PSU problem (510 W, by the way).

Anybody a good idea what to try next - beside the obvious "go for nvidia"?

TIA
maddoc.

[bridgman]

That sounds like you might have AGP bus problems. The XP driver automatically slows down the AGP bus until it runs reliably; I don't think the Linux drivers do that.

Install issues don't seem like the issue here, but running with a clean install of Ubuntu 8.10 and the fglrx driver installed by the restricted driver manager would be a good way to be sure.

I don't remember the current commands for slowing down the AGP bus but that would be the first thing I would try. Take it down to 1x and see what happens.

[maddoc]

That sounds like you might have AGP bus problems. The XP driver automatically slows down the AGP bus until it runs reliably; I don't think the Linux drivers do that.

Install issues don't seem like the issue here, but running with a clean install of Ubuntu 8.10 and the fglrx driver installed by the restricted driver manager would be a good way to be sure.

I don't remember the current commands for slowing down the AGP bus but that would be the first thing I would try. Take it down to 1x and see what happens.

ty for your help bridgeman.
I tried to slow down AGP by inserting

Option "AGPMask" "0x00000006"
Option "AGPv3Mask" "0x00000003"

(both and only first one), with different values, but my card is always put in 8x mode, Xorg log shows:
...
(WW) fglrx(0): Option "AGPMask" is not used
(WW) fglrx(0): Option "AGPv3Mask" is not used
...

I tried to force the card into PCI mode with
Option "BusType" "PCI"
which seems to work, Xserver reports to run the card in PCI mode, however, kernel messages are unaffected and report AGP 8x to be used:
...
[fglrx] AGP detected, AgpState = 0x1f004a1b (hardware caps of chipset)
[fglrx] [agp] enabling AGP with mode=0x1f004b1a
agpgart-intel 0000:00:00.0: AGP 3.0 bridge
agpgart-intel 0000:00:00.0: putting AGP V3 device into 8x mode
fglrx_pci 0000:01:00.0: putting AGP V3 device into 8x mode
[fglrx] AGP enabled, AgpCommand = 0x1f004312 (selected caps)
[fglrx] Setup AGP aperture
[fglrx] Maximum main memory to use for locked dma buffers: 430 MBytes.
[fglrx] Gart cacheable size:252 M.
[fglrx] Reserved FB block: Shared offset:0, size:1000000
[fglrx] Reserved FB block: Unshared offset:7f76000, size:8a000
[fglrx] Reserved FB block: Unshared offset:1fffc000, size:4000
.

On Windows XP, SmartGart sets the card into 8x with all speed options on (fast-write etc.).

There might be an issue with intel_agp, but I didn't succeed to switch to internal agp gart usage by inserting:

Option "KernelModuleParm" "agpgart=0"

AGP doesn't work without intel_agp module, blacklisting it always results in "cannot init AGP", regardless of "KernelModuleParm" setting.

If anybody has a hint how to slow down AGP, I would like to give it a try.

I will do a fresh 8.10 install the next days, I will report results.

[bridgman]

Yeah, I think the AGP bus options are different when running a PCIE chip over an AGP bridge (you have to program the bridge not the GPU AFAIK) but I don't know what they are.

[Kano]

I highly doubt chaining the AGP rate will help. That did not help using the BIOS on a nForce2 system with a HD 2600 too. Kernel 2.6.24 + old ATI driver, something like 8-3 or 8-4 might work. The nForce2 system even required a blacklisting of snd-hda-intel to be able to load newer drivers, but still crashes.

[maddoc]

on a fresh Ubuntu 8.10 installation, the situation is unchanged (system locked up after starting OpenGL app), but at least I could find a trace queue dump in the system log, starting with "ASIC hang"

maybe anyone here can read something interesting out of this?
does "ASIC hang" mean there is a (possible hardware) error within the GPU or gfx card?

----------------------------- begin of related kernel messages (time stamps removed)
[fglrx] ASIC hang happened
Pid: 5200, comm: Xorg Tainted: P 2.6.27-9-generic #1
[<c012989d>] ? find_busiest_group+0x15d/0x7c0
[<e11797fd>] KCL_DEBUG_OsDump+0xd/0x10 [fglrx]
[<e1189019>] firegl_hardwareHangRecovery+0x19/0x30 [fglrx]
[<e121022b>] ? _ZN4Asic9WaitUntil15ResetASICIfHungEv+0xb/0x20 [fglrx]
[<e1172c18>] ? KAS_GetTickCounter+0x8/0x10 [fglrx]
[<e12101c7>] _ZN4Asic9WaitUntil15WaitForCompleteEv+0x97/0xf0 [fglrx]
[<e120f1b3>] _ZN4Asic19PM4ElapsedTimeStampERK23PM4_TS_INTERRUPT _PARAMSj14_LARGE_INTEGER+0x1c3/0x200 [fglrx]
[<e1172c38>] ? KCL_get_pubdev+0x8/0x10 [fglrx]
[<e11a56fd>] ? firegl_trace+0x3d/0x1c0 [fglrx]
[<e1172c38>] ? KCL_get_pubdev+0x8/0x10 [fglrx]
[<e11a56fd>] ? firegl_trace+0x3d/0x1c0 [fglrx]
[<e1172c38>] ? KCL_get_pubdev+0x8/0x10 [fglrx]
[<e1201310>] _Z19uQSTimeStampRetiredjjj14_LARGE_INTEGER+0xf0/0x100 [fglrx]
[<e11fdee9>] _Z8uCWDDEQCjjjPvjS_+0x359/0x1260 [fglrx]
[<e11ea9d4>] CMMQS_uCWDDEQC+0x34/0x40 [fglrx]
[<e11a8306>] firegl_cmmqs_CWDDE_32+0x3d6/0x4b0 [fglrx]
[<c02548ca>] copy_from_user+0x3a/0x130
[<c01c64a6>] d_materialise_unique+0x156/0x370
[<e11a68aa>] firegl_cmmqs_CWDDE32+0x8a/0x140 [fglrx]
[c0211bf2>] cap_capable+0x12/0x40
[<c013a828>] capable+0x18/0x40
[<e11a6820>] firegl_cmmqs_CWDDE32+0x0/0x140 [fglrx]
[<e1183efd>] firegl_ioctl+0x22d/0x2b0 [fglrx]
[<c01c64a6>] d_materialise_unique+0x156/0x370
[<c01c64a6>] d_materialise_unique+0x156/0x370
[<e117771e>] ip_firegl_ioctl+0x1e/0x20 [fglrx]
[<c01c64a6>] d_materialise_unique+0x156/0x370
[<c01beef1>] vfs_ioctl+0x81/0x90
[<c01c64a6>] d_materialise_unique+0x156/0x370
[<c01bf086>] do_vfs_ioctl+0x66/0x1f0
[<c021007b>] install_process_keyring+0x1b/0xe0
[<c02147d8>] cap_file_ioctl+0x8/0x10
[<c01bf27b>] sys_ioctl+0x6b/0x70
[<c0103f7b>] sysenter_do_call+0x12/0x2f
[<c01c64a6>] d_materialise_unique+0x156/0x370
=======================
pubdev:0x1e134b100, num of device:-517565033 , name:<NULL>, major 56, minor -1069090554.
device 0 : 0xe1269597dd76e000 .
Asic ID:0x9505, revision:0x29, MMIOReg:0x8e1160000.
FB phys addr: 0xf0000000, MC :0xc0000000, Total FB size :0x20000000.
gart table MC:0x0, Physical:0x0, size:0x0.
mc_node :AGP, total 1 zones
MC start:0xe8000000, Physical:0xe8000000, size:0x8000000.
Mapped heap -- Offset:0x2800000, size:0x800000, reference count:1, mapping count:0,
Mapped heap -- Offset:0x2000000, size:0x800000, reference count:1, mapping count:0,
Mapped heap -- Offset:0x0, size:0x2000000, reference count:8, mapping count:0,
mc_node :FB, total 1 zones
MC start:0xc0000000, Physical:0xf0000000, size:0x8000000.
Mapped heap -- Offset:0x0, size:0x7f76000, reference count:6, mapping count:0,
Mapped heap -- Offset:0x0, size:0x1000000, reference count:1, mapping count:0,
Mapped heap -- Offset:0x7f76000, size:0x8a000, reference count:1, mapping count:0,
mc_node :INV_FB, total 1 zones
MC start:0xc8000000, Physical:0xf8000000, size:0x18000000.
Mapped heap -- Offset:0x17ffc000, size:0x4000, reference count:1, mapping count:0,
mc_node :GART_CACHEABLE, total 3 zones
MC start:0xaf300000, Physical:0x0, size:0x10d00000.
Mapped heap -- Offset:0xc00000, size:0x100000, reference count:1, mapping count:0,
Mapped heap -- Offset:0xb00000, size:0x100000, reference count:1, mapping count:0,
Mapped heap -- Offset:0xa00000, size:0x100000, reference count:1, mapping count:0,
Mapped heap -- Offset:0x900000, size:0x100000, reference count:1, mapping count:0,
Mapped heap -- Offset:0x200000, size:0x700000, reference count:2, mapping count:0,
Mapped heap -- Offset:0x0, size:0x200000, reference count:2, mapping count:0,
Mapped heap -- Offset:0xef000, size:0x11000, reference count:1, mapping count:0,
Dump the trace queue.
End of dump
syslogd 1.5.0#2ubuntu6: restart.
-------------------------------------------------- end of messages

[Kano]

If you don't know where to get a 2.6.24 kernel, just install U 8.04 if you are an U fan. To test differnt drivers, the most easy way is to use my script:

http://kanotix.com/files/install-fglrx-debian.sh

Run it like:

sh install-fglrx-debian.sh -v 8-3

[maddoc]

If you don't know where to get a 2.6.24 kernel, just install U 8.04 if you are an U fan.

Well, I'm more a Debian fan, just switched to U temporarily since I hoped to resolve these issues more quickly, or at least to see wether a solution will be in sight. I have Lenny installed on another partition. AFAIK there's no 2.6.24 kernel in official Lenny repos (?).

Moreover, my motivation of getting tied to a certain (older) kernel version is low. If there was hope that any future fglrx release would resolve the issue, it would be a nice workaround, 8.04 would be ok since it's an LTS version. But I'm afraid the AGP thing will not be on focus by AMD devs and future fglrx updates won't help me.

Generally, I'm thinking hard about throwing this 3850 AGP card onto ebay and better buy an Nvidia card, although Nvidia doesn't offer that much power on the AGP side anymore.

Is the situation with PCI-E cards and fglrx better? When reading this forum here, I wouldn't bet...

ty for your script, read a lot here about it and people really seem to love it.
I will give it a try on lenny, do you know a repo for a 2.6.24 kernel? I know how to build my own kernel (no big deal with make-kpkg), but I decided to rely on stock kernels for security issues and convenience (ok, comes first)...

[legume]

You could try changing the AGP Aperture size in bios, that sometimes helps with fglrx.

[maddoc]

You could try changing the AGP Aperture size in bios, that sometimes helps with fglrx.

tried that already, it had no effect on the problem. Somewhere I've read AGP aperture should be equal to the amount of video memory for fglrx, but my mainboard only allows up to 256M (512M on the 3850 card).

[Kano]

Did you try a 2.6.24 kernel? I compiled the latest snapshot from hardy to be used with Debian here:

http://kanotix.com/files/kernel/kernel-update-pack-generic-24/kernel/

Best put all into a dir and install em using dpkg -i *.deb.

[maddoc]

Did you try a 2.6.24 kernel? I compiled the latest snapshot from hardy to be used with Debian here:

http://kanotix.com/files/kernel/kernel-update-pack-generic-24/kernel/

Best put all into a dir and install em using dpkg -i *.deb.

ty Kano, just tried it with your supplied 2.6.24 kernel and 8-3 and 8-4 as well.

in both cases, X server starts up but screen stays black, reboot with magic sysreq necessary, dump messages in syslog from fglrx kernel module load:


Jan 5 22:29:48 dagobert dhcdbd: message_handler: message handler not found under /com/redhat/dhcp/eth0 for sub-path eth0.dbus.get.reason
Jan 5 22:29:48 dagobert kernel: [ 481.481285] ACPI: PCI Interrupt 0000:01:00.0[A] -> GSI 16 (level, low) -> IRQ 17
Jan 5 22:29:49 dagobert kernel: [ 482.503446] [fglrx] AGP detected, AgpState = 0x1f004a1b (hardware caps of chipset)
Jan 5 22:29:49 dagobert kernel: [ 482.503761] agpgart: Found an AGP 3.0 compliant device at 0000:00:00.0.
Jan 5 22:29:49 dagobert kernel: [ 482.503791] agpgart: Putting AGP V3 device at 0000:00:00.0 into 8x mode
Jan 5 22:29:49 dagobert kernel: [ 482.503976] agpgart: Putting AGP V3 device at 0000:01:00.0 into 8x mode
Jan 5 22:29:49 dagobert kernel: [ 482.504001] [fglrx] AGP enabled, AgpCommand = 0x1f004312 (selected caps)
Jan 5 22:29:49 dagobert kernel: [ 482.571939] [fglrx] Reserve Block - 0 offset = 0X1fffc000 length = 0X4000
Jan 5 22:29:49 dagobert kernel: [ 482.571948] [fglrx] Reserve Block - 1 offset = 0X0 length = 0X1000000
Jan 5 22:29:49 dagobert kernel: [ 482.571954] [fglrx] Reserve Block - 2 offset = 0X7f7f000 length = 0X80000
Jan 5 22:29:50 dagobert kernel: [ 483.232919] Pid: 5007, comm: Xorg Tainted: P 2.6.24-23-generic #1
Jan 5 22:29:50 dagobert kernel: [ 483.232949] [<c01692a4>] __report_bad_irq+0x24/0x90
Jan 5 22:29:50 dagobert kernel: [ 483.232979] [<c016959a>] note_interrupt+0x28a/0x2c0
Jan 5 22:29:50 dagobert kernel: [ 483.232986] [<e127ebd0>] IRQMGR_InterruptServiceRoutine+0x0/0x50 [fglrx]
Jan 5 22:29:50 dagobert kernel: [ 483.233145] [<c0168780>] handle_IRQ_event+0x30/0x70
Jan 5 22:29:50 dagobert kernel: [ 483.233160] [<c0169f9b>] handle_fasteoi_irq+0xbb/0xf0
Jan 5 22:29:50 dagobert kernel: [ 483.233174] [<c0106eeb>] do_IRQ+0x3b/0x80
Jan 5 22:29:50 dagobert kernel: [ 483.233196] [<c01053f3>] common_interrupt+0x23/0x30
Jan 5 22:29:50 dagobert kernel: [ 483.233225] [<e1249da0>] _ZN7PM4Ring4wptrEv+0x0/0x20 [fglrx]
Jan 5 22:29:50 dagobert kernel: [ 483.233349] [<e124d7f0>] _ZN4Asic16Is_WPTR_equ_RPTR19ConditionSuccessfulEv+ 0x20/0x50 [fglrx]
Jan 5 22:29:50 dagobert kernel: [ 483.233461] [<e124c8b8>] _ZN4Asic9WaitUntil15WaitForCompleteEv+0x38/0xf0 [fglrx]
Jan 5 22:29:50 dagobert kernel: [ 483.233572] [<e1255e56>] _ZN8AsicR60016ASICIdleInternalEN4Asic15idle_WaitMe thodE+0x96/0x1f0 [fglrx]
Jan 5 22:29:50 dagobert kernel: [ 483.233698] [<e1249a8b>] _ZN7PM4Ring8PM4queueEPPj+0x3b/0xa0 [fglrx]
Jan 5 22:29:50 dagobert kernel: [ 483.233803] [<e1255c8f>] _ZN8AsicR60010updateWptrEv+0x2f/0x50 [fglrx]
Jan 5 22:29:50 dagobert kernel: [ 483.233907] [<e1249b4a>] _ZN7PM4Ring9PM4submitEPPjb+0x5a/0x60 [fglrx]
Jan 5 22:29:50 dagobert kernel: [ 483.234040] [<e124b16c>] _ZN4Asic7PM4idleENS_15idle_WaitMethodE+0x4c/0x80 [fglrx]
Jan 5 22:29:50 dagobert kernel: [ 483.234154] [<e1255c8f>] _ZN8AsicR60010updateWptrEv+0x2f/0x50 [fglrx]
Jan 5 22:29:50 dagobert kernel: [ 483.234258] [<e124aeb0>] _ZN4Asic9assertPM4Eb+0xf0/0x250 [fglrx]
Jan 5 22:29:50 dagobert kernel: [ 483.234368] [<e1257723>] _ZN8AsicR6009assertPM4Eb+0x33/0x50 [fglrx]
Jan 5 22:29:50 dagobert kernel: [ 483.234478] [<e122dbd0>] CMMQS_Initialize+0x130/0x160 [fglrx]
Jan 5 22:29:50 dagobert kernel: [ 483.234578] [<e11f4515>] firegl_cmmqs_init+0x4b5/0x550 [fglrx]
Jan 5 22:29:50 dagobert kernel: [ 483.234655] [<e11eecbb>] MCIL_SpinLock+0x6b/0x110 [fglrx]
Jan 5 22:29:50 dagobert kernel: [ 483.234737] [<e1280677>] IRQMGR_SpinLock_Release+0x57/0x60 [fglrx]
Jan 5 22:29:50 dagobert kernel: [ 483.235110] [<c01064a0>] do_iret_error+0x90/0xa0
Jan 5 22:29:50 dagobert kernel: [ 483.235122] [<e11f3a78>] firegl_cmmqs_createdriver+0x138/0x150 [fglrx]
Jan 5 22:29:50 dagobert kernel: [ 483.235207] [<e11f3940>] firegl_cmmqs_createdriver+0x0/0x150 [fglrx]
Jan 5 22:29:50 dagobert kernel: [ 483.235283] [<e11e648e>] firegl_ioctl+0x19e/0x220 [fglrx]
Jan 5 22:29:50 dagobert kernel: [ 483.235353] [<c01064a0>] do_iret_error+0x90/0xa0
Jan 5 22:29:50 dagobert kernel: [ 483.235383] [<c01064a0>] do_iret_error+0x90/0xa0
Jan 5 22:29:50 dagobert kernel: [ 483.235391] [<e11db85c>] ip_firegl_ioctl+0x1c/0x30 [fglrx]
Jan 5 22:29:50 dagobert kernel: [ 483.235455] [<c01064a0>] do_iret_error+0x90/0xa0
Jan 5 22:29:50 dagobert kernel: [ 483.235476] [<c019e448>] do_ioctl+0x78/0x90
Jan 5 22:29:50 dagobert kernel: [ 483.235503] [<c019e4bc>] vfs_ioctl+0x5c/0x290
Jan 5 22:29:50 dagobert kernel: [ 483.235511] [<c0191be0>] do_sync_write+0x0/0x110
Jan 5 22:29:50 dagobert kernel: [ 483.235538] [<c019e758>] sys_ioctl+0x68/0x80
Jan 5 22:29:50 dagobert kernel: [ 483.235565] [<c01043a2>] sysenter_past_esp+0x6b/0xa9
Jan 5 22:29:50 dagobert kernel: [ 483.235575] [<c01064a0>] do_iret_error+0x90/0xa0
Jan 5 22:29:50 dagobert kernel: [ 483.235622] =======================
Jan 5 22:30:02 dagobert kernel: [ 494.992422]
Jan 5 22:30:02 dagobert kernel: [ 494.992429] Pid: 5007, comm: Xorg Tainted: P (2.6.24-23-generic #1)
Jan 5 22:30:02 dagobert kernel: [ 494.992436] EIP: 0060:[<e124d812>] EFLAGS: 00003202 CPU: 0
Jan 5 22:30:02 dagobert kernel: [ 494.992588] EIP is at _ZN4Asic16Is_WPTR_equ_RPTR19ConditionSuccessfulEv+ 0x42/0x50 [fglrx]
Jan 5 22:30:02 dagobert kernel: [ 494.992593] EAX: e0dd8080 EBX: 007ffff9 ECX: 00000000 EDX: e1442000
Jan 5 22:30:02 dagobert kernel: [ 494.992597] ESI: 00000000 EDI: dea89b74 EBP: dea89adc ESP: dea89ac4
Jan 5 22:30:02 dagobert kernel: [ 494.992601] DS: 007b ES: 007b FS: 00d8 GS: 0033 SS: 0068
Jan 5 22:30:02 dagobert kernel: [ 494.992605] CR0: 8005003b CR2: e1411fbc CR3: 1e8d5000 CR4: 000006d0
Jan 5 22:30:02 dagobert kernel: [ 494.992610] DR0: 00000000 DR1: 00000000 DR2: 00000000 DR3: 00000000
Jan 5 22:30:02 dagobert kernel: [ 494.992615] DR6: ffff0ff0 DR7: 00000400
Jan 5 22:30:02 dagobert kernel: [ 494.992648] [<e12400d8>] _Z19uQSTimeStampElapsedjj14_LARGE_INTEGER+0xc8/0xf0 [fglrx]
Jan 5 22:30:02 dagobert kernel: [ 494.992753] [<e124c8b8>] _ZN4Asic9WaitUntil15WaitForCompleteEv+0x38/0xf0 [fglrx]
Jan 5 22:30:02 dagobert kernel: [ 494.992871] [<e1255e56>] _ZN8AsicR60016ASICIdleInternalEN4Asic15idle_WaitMe thodE+0x96/0x1f0 [fglrx]
Jan 5 22:30:02 dagobert kernel: [ 494.993002] [<e1249a8b>] _ZN7PM4Ring8PM4queueEPPj+0x3b/0xa0 [fglrx]
Jan 5 22:30:02 dagobert kernel: [ 494.993106] [<e1255c8f>] _ZN8AsicR60010updateWptrEv+0x2f/0x50 [fglrx]
Jan 5 22:30:02 dagobert kernel: [ 494.993217] [<e1249b4a>] _ZN7PM4Ring9PM4submitEPPjb+0x5a/0x60 [fglrx]
Jan 5 22:30:02 dagobert kernel: [ 494.993352] [<e124b16c>] _ZN4Asic7PM4idleENS_15idle_WaitMethodE+0x4c/0x80 [fglrx]
Jan 5 22:30:02 dagobert kernel: [ 494.993464] [<e1255c8f>] _ZN8AsicR60010updateWptrEv+0x2f/0x50 [fglrx]
Jan 5 22:30:02 dagobert kernel: [ 494.993569] [<e124aeb0>] _ZN4Asic9assertPM4Eb+0xf0/0x250 [fglrx]
Jan 5 22:30:02 dagobert kernel: [ 494.993680] [<e1257723>] _ZN8AsicR6009assertPM4Eb+0x33/0x50 [fglrx]
Jan 5 22:30:02 dagobert kernel: [ 494.993795] [<e122dbd0>] CMMQS_Initialize+0x130/0x160 [fglrx]
Jan 5 22:30:02 dagobert kernel: [ 494.993895] [<e11f4515>] firegl_cmmqs_init+0x4b5/0x550 [fglrx]
Jan 5 22:30:02 dagobert kernel: [ 494.993974] [<e11eecbb>] MCIL_SpinLock+0x6b/0x110 [fglrx]
Jan 5 22:30:02 dagobert kernel: [ 494.994056] [<e1280677>] IRQMGR_SpinLock_Release+0x57/0x60 [fglrx]
Jan 5 22:30:02 dagobert kernel: [ 494.994425] [<c01064a0>] do_iret_error+0x90/0xa0
Jan 5 22:30:02 dagobert kernel: [ 494.994441] [<e11f3a78>] firegl_cmmqs_createdriver+0x138/0x150 [fglrx]
Jan 5 22:30:02 dagobert kernel: [ 494.994536] [<e11f3940>] firegl_cmmqs_createdriver+0x0/0x150 [fglrx]
Jan 5 22:30:02 dagobert kernel: [ 494.994616] [<e11e648e>] firegl_ioctl+0x19e/0x220 [fglrx]
Jan 5 22:30:02 dagobert kernel: [ 494.994689] [<c01064a0>] do_iret_error+0x90/0xa0
Jan 5 22:30:02 dagobert kernel: [ 494.994719] [<c01064a0>] do_iret_error+0x90/0xa0
Jan 5 22:30:02 dagobert kernel: [ 494.994744] [<e11db85c>] ip_firegl_ioctl+0x1c/0x30 [fglrx]
Jan 5 22:30:02 dagobert kernel: [ 494.994813] [<c01064a0>] do_iret_error+0x90/0xa0
Jan 5 22:30:02 dagobert kernel: [ 494.994832] [<c019e448>] do_ioctl+0x78/0x90
Jan 5 22:30:02 dagobert kernel: [ 494.994852] [<c019e4bc>] vfs_ioctl+0x5c/0x290
Jan 5 22:30:02 dagobert kernel: [ 494.994869] [<c0191be0>] do_sync_write+0x0/0x110
Jan 5 22:30:02 dagobert kernel: [ 494.994890] [<c019e758>] sys_ioctl+0x68/0x80
Jan 5 22:30:02 dagobert kernel: [ 494.994915] [<c01043a2>] sysenter_past_esp+0x6b/0xa9
Jan 5 22:30:02 dagobert kernel: [ 494.994933] [<c01064a0>] do_iret_error+0x90/0xa0
Jan 5 22:30:02 dagobert kernel: [ 494.994980] =======================



no warning or error messages in Xorg log, everything seems to initialize fine (forgot to save log :()

think I'll switch to radeonhd for the moment, support for RV670 seems to be in the queue...works reasonable for 2D already (soft accel).

[Kano]

Show me:

cat /proc/asound/cards
cat /proc/asound/modules

if you dont use snd_hda_intel in there, then just do:

echo backlist snd-hda-intel > /etc/modprobe.d/snd

that would disable the sounddriver for the gfx card. Sometimes

irqpoll

as bootcheat helps too.

[maddoc]

Show me:

cat /proc/asound/cards
cat /proc/asound/modules

if you dont use snd_hda_intel in there, then just do:

echo backlist snd-hda-intel > /etc/modprobe.d/snd

that would disable the sounddriver for the gfx card. Sometimes

irqpoll

as bootcheat helps too.

sry Kano, but I gave up last night. I had the feeling to waste too much time on this issue, and I apologize for consuming yours, now that I surrendered.
I will now go for nvidia, there will be a "cheap" HD3850 AGP on ebay soon...
thank you again for your help!

it would really like to investigate this further, I am still curious wether a working solution does exist, but I already plugged my old nvidia card in again and did a clean up of both U and lenny from extra packages.

I do expect open source drivers to deliver a lot in 2009, but have to admit that I'm not patient enough to wait.

[zondajag]

sry Kano, but I gave up last night. I had the feeling to waste too much time on this issue, and I apologize for consuming yours, now that I surrendered.
I will now go for nvidia, there will be a "cheap" HD3850 AGP on ebay soon...
thank you again for your help!

it would really like to investigate this further, I am still curious wether a working solution does exist, but I already plugged my old nvidia card in again and did a clean up of both U and lenny from extra packages.

I do expect open source drivers to deliver a lot in 2009, but have to admit that I'm not patient enough to wait.

Um hey man. I'm relatively new to the Linux scene but i like to think i have things under control atm. I have a Gecube agp 3850 512mb and im running Compiz Fusion on it fine on Ubuntu 8.1. I had to install linux in safe graphics mode (xforcevesa which didnt allow desktop effects from compiz) but once installed ubuntu found a proprietry driver that works. it was of the version 8.53 and it installed with ati catalyst 2.1.
I thought about looking for newer catalyst drivers but after reading this thread im not sure i want to mess with this since everythings running nicely.
Plus i downloaded catalyst 8.12 and the software requirements in the release notes were very obscure. I have no idea if i already had what was needed to install it. None of the stuff on the list was easy to find and i was unsure if installing the possibly 'newer' catalyst and drivers would mess things up.

The only problem im having is that when accessing ntfs drives and partitions on the machine causes linux to freeze after a short time. Tried installing newer ntfs 3g drivers but that didnt fix a thing. Its a bit like the problem u were having in a way, is there a possible link?

[maddoc]

@zondajag: I would tend to assume you have a different problem - if you can reliably connect your freezes to ntfs activities, I would bet there's something else cooking.
Have you tried running any OpenGL app, e.g. glxgears (contained in package mesa-utils)? If this is running without trouble, your problem is hardly related to what I've been describing here.

[zondajag]

@zondajag: I would tend to assume you have a different problem - if you can reliably connect your freezes to ntfs activities, I would bet there's something else cooking.
Have you tried running any OpenGL app, e.g. glxgears (contained in package mesa-utils)? If this is running without trouble, your problem is hardly related to what I've been describing here.

Erm i dont think the ntfs is all that reliable but i would assume its a different problem, though i think it happens after a set time which is a bit strange.

I havent tried glxgears but is it a visual thing in linux someplace? Is it related to the gears animation in the 3d cube desktop in compiz fusion?

[maddoc]

Erm i dont think the ntfs is all that reliable but i would assume its a different problem, though i think it happens after a set time which is a bit strange.

I havent tried glxgears but is it a visual thing in linux someplace? Is it related to the gears animation in the 3d cube desktop in compiz fusion?

you have to open a terminal and type "glxgears" at the command prompt. If you get something like " bash: glxgears: command not found" then you have to install the package mesa-utils first by typing:
"sudo aptitude install mesa-utils"
which supplies the glxgears program.
glxgears opens a window and displays 3 animated gears, yes, pretty like those in the gears animation with compiz fusion gears. In addition, the frame rate is displayed every 5 seconds in the terminal.
Thus theglxgears program provides a real life test wether OpenGL works on your system, and gives some hint about 3D performance, although it is not intended to be a 3D benchmarking tool.

[zondajag]

doesnt compiz fusion use the opengl features on the 3850 card anyway or does it use some other form of messing with the graphics?

[monomakh]

I have the same videocard and the same problem - hangs on 8x agp mode.
My system is:
hardware
CPU:Athlon XP 2200+
Mainboard:Soltek SL-75FRN3-L on nForce2 chipset
RAM:768Mb
Videocard: Sapphire HD 3850 agp
Powerblock just 300w, but videocard doesn't start with another mine 400w block.

Previous videocard: GeForce 4200 Ti 8x that works fine with mine configuration on 8x.

OS: OpenSuse 11.1 x86 (2.6.27.7-9)
Catalyst 8.12 (8.10 and previous cant to be installed on OpenSuse 11.1. 9.1 beta have the same problem).

My hardware configuration is very different from author of this topic, but i have the same problems with this videocard.

AGP aperture size set on 512Mb AGP Fast write is off, but system hangs if AGP 8x is enabled after start any of opengl applications.
System works more stable (but sometimes can hangs too) on 4x agp mode.
But I thihk that card works very slow on 4x
Glxgears get me no more than 3000FPS (3500FPS on 9.1beta)
Fgl_glxgears get me no more than 716FPS (700FPS on 9.1beta)

[bridgman]

I doubt you will see any significant performance difference between AGP8x and AGP4x. Even 2x won't make that much difference.

And GLXGEARS IS NOT A FREAKIN' BENCHMARK :D

Seriously, all glxgears does is draw flat-shaded triangles and clear the screen a lot. No textures, no shaders, none of the things that modern GPUs spend their transistor budget on. An old card will usually outrun a new card on that kind of test.

[Kano]

Well it is a thruput benchmark. Just raw power but nothing special.

[monomakh]

By the way, mine GPU speed is always 324Mhz, although normal speed must be 670Mhz. Can I set GPU speed to 670 manually?

[bridgman]

Well it is a thruput benchmark. Just raw power but nothing special.

Agreed, but "throughput" means different things to different people. Glxgears only really measures ROP performance, and for the last few years the trend has been towards relatively fewer ROPs and relatively more shaders and texture engines. Driver optimizations have also tended towards more code which improves performance with complex workloads at the expense of longer code paths for trivial workloads.

I think we're already at the point where software rendering is faster than low end GPUs when running glxgears. On more complex workloads, of course, it's a different story.

By the way, mine GPU speed is always 324Mhz, although normal speed must be 670Mhz. Can I set GPU speed to 670 manually?

The chip should switch to full speed automatically when it sees a significant 3D workload. It's possible that glxgears doesn't use enough of the chip to trip the clock switching logic, not sure though. I guess that would be the ultimate insult for a benchmark; when the driver doesn't think it's even worth speeding up the clocks ;)

[legume]

I have the same videocard and the same problem - hangs on 8x agp mode.
My system is:
hardware
CPU:Athlon XP 2200+
Mainboard:Soltek SL-75FRN3-L on nForce2 chipset
RAM:768Mb
Videocard: Sapphire HD 3850 agp
Powerblock just 300w, but videocard doesn't start with another mine 400w block.

OS: OpenSuse 11.1 x86 (2.6.27.7-9)
Catalyst 8.12 (8.10 and previous cant to be installed on OpenSuse 11.1. 9.1 beta have the same problem).


I've got the same card on an Abit nforce2 and it doesn't hang - there are differences, though.

You are 150W short of PSU power and also I've never seriously used my card with kernels > that 2.6.24.7 as I get glitching.

I also don't run gl apps with composite on as that was a bit flakey - glxinfo | grep render should say direct YES

[monomakh]

Composite was off, when I run all my tests. Today I've tried to enable composite and gets some slower FPS but difference is not serious.
direct rendering is set to YES.
As I said before, my system don't start with mine 400W PSU and this videocard, and I've back to 300W PSU. Although,maybe I will try to power on videocard to 300W PSU and motherboard and other devices to 400W PSU, but before I want to start some tests and even install Windows. If all will be works at windows - then problem not in PSU.

[bridgman]

Your 400W supply may have more +5 and even less power on the +12 rail; that's not unusual. The board draws most of its power from the +12 rail.

You need to check the per-rail amperage ratings on each P/S and compare that to what the board requires. There's a bit of wiggle room if you have a really basic PC (1 hard disk, no floppy etc..) but not that much.

[monomakh]

I've downloaded and run under wine GL Excess (OpenGL benchmark program).
I have the notebook with Quadro NVS 140M (much slower than HD 3850) and the same OS and GL Excess installed.
GL Excess can start only in window mode, although no problem with fullmode with Nvidia.
Summaries for HD 3850(1st rate) was much better than NVS 140M(2nd rate) for FILL RATE(21553/2829), polygon count(13522/3893),vram tests(4812/2607), only cpu-fpu(1763/3849) tests was failed (it's normal, my CPU is slower than on notebook).
Instantly after test, I go to Catalyst Control Center but GPU works just only on 324Mhz.
I will try to see how my card is works on windows... ughh... i need to find free space on hdd to install it :)

[legume]

Instantly after test, I go to Catalyst Control Center but GPU works just only on 324Mhz.

You'll not see the full rate after the test, you can make a little script like -


while [ 1 ] ; do
aticonfig --adapter=0 --od-getclocks
aticonfig --adapter=0 --od-gettemperature
sleep 3
done


paste that into an editor save it as, say, poll-card.
chmod +x poll-card and run it from an xterm/konsole/whatever
./pollcard > card-out
run your bench then <ctrl> c it when you've finished and grep xyz card-out for whatever lines you are interested in.

[monomakh]

aticonfig --adapter=0 --od-getclocks
ERROR - Could not find library: libatiadlxx.so

ll /usr/lib/libatiadlxx.so
-rw-r--r-- 1 root root 164529 Jan 25 20:30 /usr/lib/libatiadlxx.so

By the way, I have installed windows xp sp2. GPU always works at 669Mhz by default, neither 324Mhz. The results of benchmark was really better than at linux (except VRAM test). The most difference was at polygon count and CPU/FRU tests.

[legume]

aticonfig --adapter=0 --od-getclocks
ERROR - Could not find library: libatiadlxx.so

I think that's down to which fglrx you are using - it works with 8-10

I have installed windows xp sp2. GPU always works at 669Mhz by default, neither 324Mhz.

Strange, mine shows 324 when there is no GPU load.

[monomakh]

I've installed 9.1 and its works really better than 9.2 beta. aticonfig --adapter=0 --od-getclocks works without errors and I've seen 669Mhz. I get really better result by benchmark tool and no problems with fullscreen.

[monomakh]

Could not find library: libatiadlxx.so was at 9.2 beta.
At windows drivers version was 8.12.

Although 9.1 is better than 8.12, I still have critical issues:
Hangs at 8x agp speed after few seconds or one when opengl application started. I have not these issue at windows.
Hangs at 4x agp speed often but randomly. I can work hour or two minutes without this hangs. Computer never hangs if radeonhd used (but 3d is not supported :( )

Ogh, hangs now have some change. With 9.1, now my cursor is not hanging :)
I want to check tv-out support. My monitor dont works with tv-out at 8.12, 9.1 beta or 9.2 beta... I don't think that something have been changed.

My card just 1 month old... It's time to think to sale it...
I have problems with nvidia, but they are just nothing comparing with ati.

[bridgman]

The Windows drivers stress-test the AGP bus and automatically downgrade it to a slower speed if necessary. They also turn off fast writes (and a couple of other things I forget) if required to get the bus running reliably.

I suspect the Windows driver is automatically slowing the AGP bus to make your system run reliably; sounds like that would be 2X for your system. The Linux driver does not have this mechanism (called SmartGART).

Does the system run reliably at 2X ?

[monomakh]

Tv-out problem is remains. Just connect tv-out, goes to runlevel 3, configure display to fglrx and start runlevel 5... All hangs and nothing at output... Just reboot... oh, reboot to runlevel 3 or it hangs again.
Or just disconnect s-video and connect it to my new notebook with nvidia. Or just disconnect this card and install back old nvidia card, that I changed and all problem will be solved.
Brrr...
Looks like the only way to solve problem is to buy nvidia and forget amd forever.

[monomakh]

My motherboard doesnt support 2x agp mode. I have option to disable or enable AGP at 8x or 4x. But my old Nvidia GeForce Ti 4200 with 8x support works with 8x normally at the same linux or windows versions. My old GF 4200 supports 3d, tv-out, 8x, OpenSuse at the same hardware...
Looks like HD 3850 doesn't support any of these.

[bridgman]

I don't know the details, but my understanding is that the drivers can adjust the bus speed even if you don't have a BIOS option for the lower speeds.

[monomakh]

I found a workaround how to watch dvb-s using xine (I have not this issue on nvidia cards). I need to run watching dvb-s, press alt+F11 (goes to console), and back from it (alt+F7) - than video looks like normal.
This is not a very good workaround, because after switch a channel or stop playback - all hangs and monitor goes to black screen.
I've switched back to stable but 3d and tv-out unsupported radeonhd and start search of ways to sold this unstable for linux card.

[monomakh]

Maybe radeonhd works not so good as it can?
(--) PCI:*(0@1:0:0) unknown vendor (0x1002) unknown chipset (0x9505) rev 0, Mem @ 0xb0000000/0, 0xc2000000/0, I/O @ 0x0000c000/0, BIOS @ 0x????????/131072

[monomakh]

fglrx works really better with tv-out after I change from one DVI connector to another

[monomakh]

Plus another minus this card vs my old nvidia: my computer didn't shut down with this card. I hear that HDD is stopped, but video is on and I need to power off the computer hard. I have not this problem when I shutting down computer from windows or when using nvidia card.

[Swoopy]

tried that already, it had no effect on the problem. Somewhere I've read AGP aperture should be equal to the amount of video memory for fglrx, but my mainboard only allows up to 256M (512M on the 3850 card).

Does your mainboard allow changing the voltage for your AGP card?
I had to up it to 1.65V on my Abit Av8 with H.I.S. RadeonHD "Turbo Edition" 3850 AGP before 3d apps would run stable in either Windows or Linux. 2D apps and windows itself caused no problems at all at any voltage.(This required going to the voltages page in the BIOS and changing from "auto" to "manual" after which I could set the memory, CPU core and AGP voltages by hand.)

At 1.55V or lower, the "VPU Recover" interrupts would happen about once every 5 minutes while 3D benchmarking or gaming in Windows XP.

[monomakh]

Yes, my motherboard have this option. But this option didnt help me to power off videocard at shutdown automatically, didnt help me with tv-out, didnt help me with hangs at 8x

[monomakh]

Thanks for all. Videocard have been sold