ATI Radeon HD 5850/5870 On Linux?

[tuxdriver]

32xAF? Since when do ATI cards perform 32x anisotropic filtering?

[mtippett]

I'm typing this from a Debian Lenny/Sid amd64 box (AMD X4 955BE, 8GB PC2-8500, MSI K9A2 Platinum) with a brand new Radeon 5870 in it. I've got the "Unsupported hardware" banner at the bottom right, but I've not run into any problems so far. I've been running Heroes of Newerth at 1680x1050, vsync off, with 8xAA and 32xAF on at >70FPS. X-Plane runs very nicely at 1680x1050 extreme res with 4xAA and 4xAF (not nudged it any higher yet), ETQW runs very nicely with 16xAF (>60FPS constant). This is with Catalyst 9.9.

The AIB vendor (which one BTW?) has the option of including a driver on the CD. Did you check?

The hardware became "supported" between the final packaging of Catalyst 9.9 and the next release. There may some bugs that you may still encounter that were resolved prior to the final release, but the 9.9 catalyst will be 95% "there". Beyond a few small corner cases, the driver should be representative of the performance of the card.

Regards,

Matthew

[GotenXiao]

32xAF? Since when do ATI cards perform 32x anisotropic filtering?

Since the 5870, apparently

The AIB vendor (which one BTW?) has the option of including a driver on the CD. Did you check?

The hardware became "supported" between the final packaging of Catalyst 9.9 and the next release. There may some bugs that you may still encounter that were resolved prior to the final release, but the 9.9 catalyst will be 95% "there". Beyond a few small corner cases, the driver should be representative of the performance of the card.

Regards,

Matthew

XFX - I did check, but they've only got Windows 2k/XP and Vista/7 drivers on the CD.

I believe I encountered one of the bugs - this happened after I went to bed, and I've not been able to get an X server running this morning (without rebooting).

This is from kern.log:

Oct  3 02:45:06 nightstalker kernel: [14496.680795] [fglrx:firegl_cail_init] *ERROR* CAIL: CAIL_ASICSetup failed, error 1
Oct  3 02:45:06 nightstalker kernel: [14496.680799] [fglrx:hal_init_asic] *ERROR* Failed to initialize ASIC.
Oct  3 02:45:06 nightstalker kernel: [14496.680967] BUG: unable to handle kernel paging request at ffffc200121898f0
Oct  3 02:45:06 nightstalker kernel: [14496.680970] IP: [<ffffffffa04b5815>] halReadReg32+0x5/0x10 [fglrx]
Oct  3 02:45:06 nightstalker kernel: [14496.681030] PGD 23fc6c067 PUD 23fc6d067 PMD 23cbcd067 PTE 0
Oct  3 02:45:06 nightstalker kernel: [14496.681033] Oops: 0000 [#1] SMP
Oct  3 02:45:06 nightstalker kernel: [14496.681035] last sysfs file: /sys/devices/pci0000:00/0000:00:14.4/0000:04:00.0/resource
Oct  3 02:45:06 nightstalker kernel: [14496.681037] CPU 3
Oct  3 02:45:06 nightstalker kernel: [14496.681038] Modules linked in: fuse snd_seq_dummy usb_storage udf joydev fglrx(P) binfmt_misc rfcomm l2cap bluetooth battery ppdev parport_pc lp parport cpufreq_userspace cpufreq_stats cpufreq_powersave cpufreq_conservative nf_conntrack_ipv6 ip6table_filter ip6_tables xt_tcpudp xt_state iptable_filter ipt_MASQUERADE iptable_nat nf_nat nf_conntrack_ipv4 nf_conntrack nf_defrag_ipv4 ip_tables x_tables deflate zlib_deflate ctr twofish twofish_common camellia serpent blowfish des_generic cbc cryptd aes_x86_64 aes_generic xcbc rmd160 sha256_generic sha1_generic hmac crypto_null af_key nls_utf8 nls_cp437 vfat fat nls_base ext2 dm_crypt f71882fg firewire_sbp2 loop snd_hda_codec_atihdmi snd_hda_codec_realtek snd_hda_intel snd_hda_codec snd_pcm_oss snd_mixer_oss snd_hwdep snd_pcm snd_seq_midi snd_rawmidi snd_seq_midi_event snd_seq snd_timer snd_seq_device usbhid snd hid i2c_piix4 wmi i2c_core processor button soundcore serio_raw psmouse pcspkr snd_page_alloc evdev ext3 jbd mbcache dm_mir
Oct  3 02:45:06 nightstalker kernel: or dm_region_hash dm_log dm_snapshot dm_mod ide_cd_mod cdrom ide_pci_generic sd_mod crc_t10dif atiixp ide_core ahci ata_generic libata firewire_ohci firewire_core crc_itu_t scsi_mod ohci_hcd floppy ehci_hcd r8169 mii thermal fan thermal_sys [last unloaded: fuse]
Oct  3 02:45:06 nightstalker kernel: [14496.681093] Pid: 7907, comm: Xorg Tainted: P           2.6.30-1-amd64 #1 MS-7376
Oct  3 02:45:06 nightstalker kernel: [14496.681095] RIP: 0010:[<ffffffffa04b5815>]  [<ffffffffa04b5815>] halReadReg32+0x5/0x10 [fglrx]
Oct  3 02:45:06 nightstalker kernel: [14496.681136] RSP: 0018:ffff8801d3575d20  EFLAGS: 00010216
Oct  3 02:45:06 nightstalker kernel: [14496.681137] RAX: ffffffffa0614ac0 RBX: ffff88023a4882c0 RCX: 0000000000000006
Oct  3 02:45:06 nightstalker kernel: [14496.681139] RDX: 00000000000098f0 RSI: 00000000000098f0 RDI: ffffc20012180000
Oct  3 02:45:06 nightstalker kernel: [14496.681140] RBP: ffff88021f8f8000 R08: 0000000000000000 R09: ffffffffa057d640
Oct  3 02:45:06 nightstalker kernel: [14496.681142] R10: 0000000000000009 R11: 000000003e920000 R12: 00000000017c9720
Oct  3 02:45:06 nightstalker kernel: [14496.681143] R13: ffff8801d3575dc8 R14: 00000000c0406466 R15: ffff8801fc8f33c0
Oct  3 02:45:06 nightstalker kernel: [14496.681145] FS:  00007fe0b8c72790(0000) GS:ffff880028082000(0000) knlGS:00000000f7df26c0
Oct  3 02:45:06 nightstalker kernel: [14496.681146] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
Oct  3 02:45:06 nightstalker kernel: [14496.681148] CR2: ffffc200121898f0 CR3: 000000020b12c000 CR4: 00000000000006e0
Oct  3 02:45:06 nightstalker kernel: [14496.681149] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
Oct  3 02:45:06 nightstalker kernel: [14496.681151] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
Oct  3 02:45:06 nightstalker kernel: [14496.681152] Process Xorg (pid: 7907, threadinfo ffff8801d3574000, task ffff8801dbdc8b30)
Oct  3 02:45:06 nightstalker kernel: [14496.681154] Stack:
Oct  3 02:45:06 nightstalker kernel: [14496.681155]  ffffffffa04b5f32 ffff88021f8f8000 ffffffffa04b4aaf ffff8801d3575d78
Oct  3 02:45:06 nightstalker kernel: [14496.681157]  ffffffffa049b9f6 ffff8801d3575de8 ffff880231f2ad58 ffff88021c174c80
Oct  3 02:45:06 nightstalker kernel: [14496.681159]  0000000000000001 ffff8801d3575ed8 ffffffff802939a8 ffff88021f8f8000
Oct  3 02:45:06 nightstalker kernel: [14496.681162] Call Trace:
Oct  3 02:45:06 nightstalker kernel: [14496.681164]  [<ffffffffa04b5f32>] ? __get_num_hzmem_pipes_r7xx+0x12/0x20 [fglrx]
Oct  3 02:45:06 nightstalker kernel: [14496.681204]  [<ffffffffa04b4aaf>] ? hal_get_num_hzmem_pipes+0x1f/0x80 [fglrx]
Oct  3 02:45:06 nightstalker kernel: [14496.681244]  [<ffffffffa049b9f6>] ? firegl_init_pcie+0xf6/0x3f0 [fglrx]
Oct  3 02:45:06 nightstalker kernel: [14496.681282]  [<ffffffff802939a8>] ? generic_file_aio_write+0x6f/0xc4
Oct  3 02:45:06 nightstalker kernel: [14496.681288]  [<ffffffffa015f384>] ? ext3_file_write+0x16/0x96 [ext3]
Oct  3 02:45:06 nightstalker kernel: [14496.681296]  [<ffffffffa049b900>] ? firegl_init_pcie+0x0/0x3f0 [fglrx]
Oct  3 02:45:06 nightstalker kernel: [14496.681334]  [<ffffffffa049755a>] ? firegl_ioctl+0x1ea/0x250 [fglrx]
Oct  3 02:45:06 nightstalker kernel: [14496.681372]  [<ffffffff802cc04b>] ? vfs_ioctl+0x56/0x6c
Oct  3 02:45:06 nightstalker kernel: [14496.681375]  [<ffffffff802cc48c>] ? do_vfs_ioctl+0x42b/0x464
Oct  3 02:45:06 nightstalker kernel: [14496.681377]  [<ffffffff802c13c5>] ? vfs_write+0xcd/0x102
Oct  3 02:45:06 nightstalker kernel: [14496.681380]  [<ffffffff802cc516>] ? sys_ioctl+0x51/0x70
Oct  3 02:45:06 nightstalker kernel: [14496.681382]  [<ffffffff8020fa42>] ? system_call_fastpath+0x16/0x1b
Oct  3 02:45:06 nightstalker kernel: [14496.681385] Code: 00 00 48 89 8f f8 03 00 00 48 85 c0 74 0d 48 83 f8 ff 74 07 48 89 87 f8 03 00 00 b8 01 00 00 00 c3 66 66 90 66 90 c1 e6 02 89 f2 <8b> 04 3a c3 66 66 66 90 66 66 90 c1 e6 02 89 f0 89 14 38 c3 66
Oct  3 02:45:06 nightstalker kernel: [14496.681402] RIP  [<ffffffffa04b5815>] halReadReg32+0x5/0x10 [fglrx]
Oct  3 02:45:06 nightstalker kernel: [14496.681441]  RSP <ffff8801d3575d20>
Oct  3 02:45:06 nightstalker kernel: [14496.681442] CR2: ffffc200121898f0
Oct  3 02:45:06 nightstalker kernel: [14496.681444] ---[ end trace bd5cd2ab2dcbaae5 ]---
Oct  3 02:45:06 nightstalker kernel: [14496.682163] [fglrx:firegl_release] *ERROR* device busy: 1 0
Oct  3 02:45:06 nightstalker kernel: [14496.682165] [fglrx] release failed with code -EBUSY

This is from Xorg.0.log (this morning after trying to restart kdm):

(II) fglrx(0): VESA BIOS detected
(II) fglrx(0): VESA VBE Version 3.0
(II) fglrx(0): VESA VBE Total Mem: 16384 kB
(II) fglrx(0): VESA VBE OEM: ATI ATOMBIOS
(II) fglrx(0): VESA VBE OEM Software Rev: 12.11
(II) fglrx(0): VESA VBE OEM Vendor: (C) 1988-2005, ATI Technologies Inc.
(II) fglrx(0): VESA VBE OEM Product: CYPRESS
(II) fglrx(0): VESA VBE OEM Product Rev: 01.00
(EE) fglrx(0): CAIL: CAIL_ASICSetup failed, error 1
(EE) fglrx(0): GetBIOSParameter failed
(EE) fglrx(0): PreInitAdatper failed
(EE) fglrx(0): PreInit failed
(II) fglrx(0): === [atiddxPreInit] === end
(EE) fglrx(0): PPLIB: PPLIB is not initialized!.
(EE) fglrx(0): PPLIB: swlPPLibNotifyEventToPPLib() failed!
(EE) fglrx(0):        ulEventType = 0000000c, ulEventData = 00000001

Backtrace:
0: /usr/bin/X(xorg_backtrace+0x26) [0x4ee256]
1: /usr/bin/X(xf86SigHandler+0x39) [0x4840b9]
2: /lib/libc.so.6 [0x7f3d824c0fc0]
3: /usr/lib/xorg/modules/drivers//fglrx_drv.so(swlDalHelperClose+0x87) [0x7f3d80a918a7]
4: /usr/lib/xorg/modules/drivers//fglrx_drv.so(atiddxFreeScreen+0x129) [0x7f3d80a6d589]
5: /usr/bin/X(xf86DeleteScreen+0x7e) [0x480e5e]
6: /usr/bin/X(InitOutput+0xdda) [0x46de3a]
7: /usr/bin/X(main+0x200) [0x4331d0]
8: /lib/libc.so.6(__libc_start_main+0xfd) [0x7f3d824ada8d]
9: /usr/bin/X [0x432819]

Edit: came back up fine after a reboot. I also reset the amdpcsdb to the default file to make sure there were no lingering entries from my 3870 crossfire setup.

[chaos386]

have you read the test on anandtech? The 5850 is blazing fast. There is no reason to buy a 4870 - except the prize...

Price *is* a big factor. The 5850 is around twice as fast as the 4850 in theory (a little less in practice, aside from in Far Cry 2), but 2.6x the price. I'm willing to pay a little extra for the performance, but if the 5800 series isn't well supported right away, I might as well get a 48X0 card and enjoy fast open-source 3D drivers sooner.

I'm typing this from a Debian Lenny/Sid amd64 box (AMD X4 955BE, 8GB PC2-8500, MSI K9A2 Platinum) with a brand new Radeon 5870 in it. I've got the "Unsupported hardware" banner at the bottom right, but I've not run into any problems so far. I've been running Heroes of Newerth at 1680x1050, vsync off, with 8xAA and 32xAF on at >70FPS. X-Plane runs very nicely at 1680x1050 extreme res with 4xAA and 4xAF (not nudged it any higher yet), ETQW runs very nicely with 16xAF (>60FPS constant). This is with Catalyst 9.9.

Thanks for braving ahead and getting a card you didn't even know would be functional under Linux!

The AIB vendor (which one BTW?) has the option of including a driver on the CD. Did you check?

The hardware became "supported" between the final packaging of Catalyst 9.9 and the next release. There may some bugs that you may still encounter that were resolved prior to the final release, but the 9.9 catalyst will be 95% "there". Beyond a few small corner cases, the driver should be representative of the performance of the card.

Regards,

Matthew

By "next release", I assume you mean Catalyst 9.10, right? I basically just want to know if the driver packaged with Ubuntu 9.10 will officially support the 5850.

[mtippett]

Thanks for braving ahead and getting a card you didn't even know would be functional under Linux!

Since the 4800 launch, Linux has generally been either included in the release CDs (as mentioned before, AMD doesn't control this), or included in the follow-up catalyst release.

By "next release", I assume you mean Catalyst 9.10, right? I basically just want to know if the driver packaged with Ubuntu 9.10 will officially support the 5850.

Well, that's an extra variable.

Yes, Catalyst 9.10 should have the hardware enabled by default.

Now, for distributions, we typically do a 1-2 apprach. We look to ensure that at release there is a driver that _mostly_ works (in this case, the 8.660 driver pushed a month or so ago. There may be some quirks when the distro is released, but that is because the distribution may change in the intervening months or we may not have completed all development.

When the distro is formally released, within a month or two we have the "final" release, which contains our full development for that distribution _and_ should be matched to the final release of the distribution.

This does means things like projective transforms in RANDR1.3 get rolled in without much fanfare. (look at the --transform option in xrandr).

Regards,

Matthew

[rbmorse]

I just installed an HD5850 onto Ubuntu 9.10 and loaded the FGLRX driver from repository. Works fine with my Apple 23 Inch Cinema display (1920 X 1200/60) so far. Any way to remove the "unsupported hardware" watermark?

[bridgman]

The Catalyst 9.10 driver should include "official" support and remove the "unsupported HW" watermark.

[Munchkin]

I'm on the fence of buying a 5870...

Can anyone please confirm current status with drivers?

I need to run:

- two monitors @ 23" @ 2048 x 1152
- compiz fusion

I can wait with 3D gaming until that is solved. But if I can't get my two monitors to work with compiz I can't work.

My alternative is to buy a 4890 and upgrade within a few months.

[Kano]

Save your money and get a GT300 next year if you really use it mainly for Linux and not only part time.

[energyman]

Save your money and get a GT300 next year if you really use it mainly for Linux and not only part time.

if it ever comes out, you mean?