ati开源驱动6.9.0终于发布。

zhuqin_83

用到现在了，觉得挺好。

2. SUPPORTED HARDWARE
3.        The radeon driver supports PCI and AGP video cards based on the follow‐
4.        ing ATI chips
6.        R100       Radeon 7200
8.        RV100       Radeon 7000(VE), M6, RN50/ES1000
10.        RS100       Radeon IGP320(M)
12.        RV200       Radeon 7500, M7, FireGL 7800
14.        RS200       Radeon IGP330(M)/IGP340(M)
16.        RS250       Radeon Mobility 7000 IGP
18.        R200       Radeon 8500, 9100, FireGL 8800/8700
20.        RV250       Radeon 9000PRO/9000, M9
22.        RS300       Radeon 9100 IGP
24.        RS350       Radeon 9200 IGP
26.        RS400/RS480 Radeon XPRESS 200(M)/1100 IGP
28.        RV280       Radeon 9200PRO/9200/9200SE/9250, M9+
30.        R300       Radeon 9700PRO/9700/9500PRO/9500/9600TX, FireGL X1/Z1
32.        R350       Radeon 9800PRO/9800SE/9800, FireGL X2
34.        R360       Radeon 9800XT
36.        RV350       Radeon 9600PRO/9600SE/9600/9550, M10/M11, FireGL T2
38.        RV360       Radeon 9600XT
40.        RV370       Radeon X300, M22
42.        RV380       Radeon X600, M24
44.        RV410       Radeon X700, M26 PCIE
46.        R420       Radeon X800 AGP
48.        R423/R430   Radeon X800, M28 PCIE
50.        R480/R481   Radeon X850 PCIE/AGP
52.        RV505/RV515/RV516/RV550
53.            Radeon X1300/X1400/X1500/X2300
55.        R520       Radeon X1800
57.        RV530/RV560 Radeon X1600/X1650/X1700
59.        RV570/R580  Radeon X1900/X1950
61.        RS600/RS690 Radeon X1200
63.        R600       Radeon HD 2900
65.        RV610/RV630 Radeon HD 2400/2600
67.        RV670       Radeon HD 3850/3870
69.        RV620/RV635 Radeon HD 3450/3470
71.        RS780       Radeon HD 3200

复制代码

</div>

zhuqin_83

据说已经支持hd 4800系列了，但需要一个patch，而且不支持2d和3d。

zhuqin_83

感兴趣的可以看一下：
http://www.x.org/wiki/RadeonFeature
http://www.x.org/wiki/RadeonProgram

zhuqin_83

http://lists.freedesktop.org/arc ... 08-June/000596.html

Version 6.9.0 of the radeon driver.  Lots of stuff since 6.8.0

In addition to lots of bug fixes there are several major new features:
- - Improved EXA render support for r1xx/r2xx cards
- - EXA render support for r3xx/r4xx/r5xx cards
- - Textured video support for Xv (r1xx-r5xx)

Full Changelog:

Adam Jackson (5):
      Bump CRTC size limits on AVIVO chips so 30" displays work
without tweaking.
      Add R500 unified shader register block.
      Fix R500_US_CONFIG.
      Add the R500 US index/data pair.
      Clarify R500 US setup.

Alan Coopersmith (2):
      On non-gcc compilers, use C99's __func__ instead of gcc's __FUNCTION__
      Strip ^M's from src/AtomBios/includes/ObjectID.h

Alban Browaeys (1):
      RADEON: Revert to old behavior when resetting the memmap on VT switch

Alex Deucher (178):
      RADEON: update man page with supported chips
      RADEON: fix DDC types 5 and 6
      RADEON: restore clock gating and CP clock errata on VT switch
      R100: fix render accel for transforms
      ATOM: properly set up DDIA output on RS6xx boards
      RS6xx: fix DDC on DDIA output (usually HDMI port)
      RS6xx: gpio entry for DDIA varies depending on the number of DFP ports
      RADEON: add textured video support for r1xx-r4xx radeons
      R500: fix typo in new r5xx regs
      RADEON: no textured video yet on XPRESS chips
      R300: fill in some more 3D bitfields
      R300: replace magic numbers in cache flush
      R300: clean up magic numbers in RADEONInit3DEngine
      R300/R500: clean up magic numbers in render code
      R300/R500: clean up magic numbers in textured video
      R300/R500: set the number graphics pipes properly
      R300: fix cordinate clamping in render code
      R300/R500: fix up clipping
      RADEON: Convert textured video to use pipelined uploads
      RADEON: fix clipping for textured video
      RADEON: fix textured video uploads
      R100/R200: move r100/r200 specific 3D setup into appropriate blocks
      R300+: update RADEONCP_REFRESH() to reflect new location of scissor regs
      AVIVO: make sure we select the right LUT for each crtc
      R300/R400: bump up the clip limits for textured video
      R300: fix up magic numbers in fragment program
      R500: fragment program clean up and magic number conversion
      AVIVO: disable dithering on DFPs
      RS6xx: fix typos in previous commit
      RS6xx: rework output parsing
      AVIVO: save/restore scaler regs
      fix off-by-one in last commit
      R5xx: fix register count when sending fragment program for textured video
      AVIVO: disable pageflipping on avivo chips until we have proper
drm support
      AVIVO: LVDS panels need dithering enabled
      AVIVO: save/restore regs by block
      RS4xx: enable exa render accel and textured video
      RADEON: only restore legacy dac regs on legacy radeons
      RADEON: memmap rework 1
      RADEON: use xf86SetDesiredModes() in screeninit and enterVT
      RS6xx: change isIGP checks to CHIP_FAMILY_RS690
      ATOMBIOS: fix atombios parser support on *bsd
      RADEON: fix fetching of dac2 adj values from newer bios tables
      AVIVO: Initial support for DCE 3.0 using atombios
      RADEON: option to override TVDAC adj values from bios with driver defaults
      DCE3.0: Minor fixups
      RADEON: take 2 on proper pragma pack support for bsds
      DCE3.0: add support for crtc memreq table
      AVIVO: fix up memsize detection for IGP chips
      AVIVO: don't add outputs for invalid connectors
      AVIVO: fix typo from a previous commit
      DCE3: Ignore outputs with DIN connectors for now
      RADEON: Fix crash in last commit
      AVIVO: clean up some unused variables
      RV550: MC setup is like RV515 not RV530
      DCE3: add output attribute to enable/disable coherent mode
      ATOMBIOS: enable load detection by default on both DACs
      DCE3: adjust PLL for DCE3 chips
      DCE3: enable DPMS on DIG ports
      ATOM: remove some cruft
      DCE3: add support for PCIEPHY (untested)
      RADEON: fix vblank interrupts after VT switch or suspend/resume
      RADEON: enable output attributes that require a modeset immediately
      R3xx/R5xx: enable blending
      R3xx: set the texture id and add some register info
      R300: setup source selects and output swizzling
      R3xx: add basic mask support
      R3xx: enable composite for non-mask cases
      R3xx: more mask work
      R3xx: attempt to setup the rasterizer properly for mask texture
      R3xx: make sure to set the FS code size correctly
      R3xx/R5xx: Fix magic numbers in vertex shaders
      R3xx/R5xx: enable VS for mask texture
      R3xx: VS WIP
      R3xx: theoretical support for component alpha
      R3xx: odds and ends...
      RADEON: make sure var is initialized properly in
RADEONAdjustMemMapRegisters()
      RADEON: fix typo in RADEONAdjustMemMapRegisters()
      R3xx: fix errant w
      R3xx: some progress
      R3xx: minor adjustments
      R3xx: get masks working and cleanup
      R3xx/R5xx: whitespace cleanup and cruft removal
      R5xx: bump tex/dst limits to 4096
      R5xx: first pass at render support (untested)
      R3xx/R5xx: remove some cruft
      R3xx: we only use 2 temps, not 3
      R3xx/R5xx: switch an ErrorF() to RADEONFALLBACK()
      R3xx/R5xx: remove extra return after last commit
      RV515: fix textured video and EXA Composite
      RADEON: fixed textured video with XAA and tiling
      RADEON: add new macros to distinguish between R3xx and R5xx 3D
      R5xx: bump textured video limits to 4096
      R3xx/R5xx: Make sure to clamp the output of the FS
      Disable the setting of HARDWARE_CURSOR_BIT_ORDER_MSBFIRST
      RS4xx: attempt to fix TMDS/DVO support
      RS4xx: attempt to set up disp/disp2 fifos correctly
      RS4xx: missed this on the last commit.
      RS4xx: more work on disp/disp2 fifo setup
      RS4xx: Revert back to previous fifo settings for now
      R3xx/R5xx: use non VAP/TCP for textured video
      R3xx/R5xx: move more VAP, etc. state setup into common init3d() function
      R3xx/R5xx: flush PVS state before enabling pvs-bypass
      RV250: disable textured video due to HW bug
      RADEON: attempt to fix lid issues
      XAA: update message about render so as to not confuse users
      AVIVO: no need to call PreinitXv() on AVIVO chips as they have no overlay
      Revert "RADEON: attempt to fix lid issues"
      RADEON: fix lid issues on AVIVO chips for real this time
      Revert "atombios: fixup the width/height to use the mode values
not the scrn ones"
      Revert "RADEON: remove driver rec copies of mc info, use save
rec directly"
      Revert "RADEON: memmap rework 1"
      Minor cleanup
      RADEON: add some quirks
      RADEON: warning fixes
      ATOMBIOS: Add support for DynamicClocks option
      R3xx+: EXA/textured video fixes
      R3xx+: minor textured video fixes
      R3xx+: use the right register for engine flush
      RADEON: store tcl status in driver rec
      Revert "R3xx/R5xx: move more VAP, etc. state setup into common
init3d() function"
      RADEON: cleanup
      R300+: textured video tcl cleanup
      R3xx+: tcl wip
      R3xx+: consolidate some common 3D code
      R3xx+: consolidate more tcl code
      R300+: pre-load vertex programs in init3D()
      Fix exa glyph corruption on newer chips
      R3xx+: more fixes to 2D/3D engine init
      R3xx+: fix XAA + textured video on non-TCL path
      R300+: consolidate some tcl/non-tcl paths
      R3xx+: 3D engine documentation and minor cleanups
      R300: move more common code to init3d()
      EXA: Don't wait for 3D idle after each Composite()
      R300+: properly setup vap_cntl
      fix up some things from the last commit
      R300+: move more common code into init3d()
      RADEON: decrease crtc/output verbosity
      RADEON: remove duplicate register define
      IGP: fix EXA composite corruption
      RADEON: assorted fixes
      RADEON: lock the cursors when updating
      RADEON: disable MMIO paths for EXA composite/texvid on IGP/R5xx
      RADEON: fixup ifdef from last commit
      radeon: add initial support for cloning outputs. (single-CRTC cards only)
      RS4xx: Split out RS400 and RS480 as separate families
      Add RS600 support
      R3/4/5xx: fix EXA rotation
      R3/4/5xx: fixup logic from last commit
      ATOM: Ignore invalid connector entries
      RADEON: enable cloning on multi-crtc cards
      RADEON: cleanup connection detection and remove duplicate code
      ATOM: remove duplicate code
      RADEON: improve support for secondary cards
      R3/4/5xx: use get_param to get the num_gb_pipes from the drm
      RADEON: only attempt to read vbios from legacy ISA space if it's primary
      R1xx: Gamma fixes for overlay
      ATOM: fail of we are not able to get a bios image
      RADEON: don't enable legacy bios table based init yet
      RADEON: add Int10 option
      RADEON: update RADEONGetVRAMType() for newer chips
      ATOM: disable TV-out for now
      RADEON: minor cleanups
      RS4xx: RC410 chips are actually RS400 based
      R5xx: fix RADEONSetAgpBase() for R5xx chips
      R3xx: use DSTCACHE_CTLSTAT rather than RB2D_DSTCACHE_CTLSTAT
      Bump for 6.9.0rc1 release
      RADEON: man page updates
      radeon: X_ERROR -> X_WARNING for num gb pipes
      RADEON: fix read past the end of an array
      0x1002 0x5657 is actually RV410
      RADEON: adjust randr crtc/output prepare/commit ordering
      IGP: attempt to fix VGA on IGP chips
      RADEON: cleanups
      RADEON: warning fix
      bump for rc release
      RADEON: adjustments to Jerome's last commit
      Bump for 6.9.0 release

Avi Rozen (1):
      RADEON: fix lockup on start

Brad Smith (2):
      RADEON: fix copy/paste error in accel code
      RADEON: sys/endian.h needs sys/types.h on BSDs

Corbin Simpson (1):
      RADEON: missing stdint.h includes

Dave Airlie (43):
      r500: initial rotate support - not fully working yet.
      r500: add textured video Xv adapter support
      r500: convert fragprog to use register values
      rs690: initial rs690 rotate
      rs690: initial textured video support
      fixup check for EXA composite pointed out by Alan Swanson
      regs: fix spelling mistake
      regs: fix spelling properly
      RADEON: fix tiling/interlaced interaction with randr 1.2
      xv: fixup XAA on r500 textured video
      radeon: the 0x5974 appears to be a mobility chip...
      R5xx: fix typ in r5xx render accel
      r500: make it work from startup.
      r300: don't bother with VAP/TCL for render.
      radeon: size bios to max of bar vs 64k.
      r500/r600: fix rotation to fill screen
      atombios: fixup the width/height to use the mode values not the scrn ones
      radeon: split quirks into separate function and new quirk for IBM RN50
      radeon: use correct DDC interfaces so quirks get applied
      Revert "Revert "atombios: fixup the width/height to use the mode
values not the scrn ones""
      atombios: fix the dual-head hopefully.
      radeon: zaphod fix for cursor on second head
      radeon: zaphod fixes for pciaccess not allowing multiple MMIO maps
      radeon: fix EXA pixmap offset on zaphod
      radeon: zaphod: fix render accel for EXA
      radeon: fix textured-xv on zaphod
      avivo: fix zaphod cursor in theory
      radeon: fix zaphod EXA with texture video
      radeon: rs485 vs rs485m.. mobile? non-mobile? bi?
      radeon: fix set_cursor_colours remove fb offset
      atombios: for LVDS set use ref div as per legacy
      radeon: fix surface access on avivo chips.
      radeon: fix typo in memory calcs
      atombios: fixup atombios bitfields for endianness
      radeon: add lots of endian bits into the atombios parser
      atom: fixup includes in cd operations to get correct endian bits
in atombios.h
      atom: fix another endian define
      Revert "ATOM: disable TV-out for now"
      r600: don't add fb offset here to make shadowfb work.
      atombios: use MMIO from the entity not the info
      legacy: use entity MMIO for dpms as this can crossover between
zaphod infos
      ati: hopefully fix byteswap mess for those other OSes.
      Revert "Revert "ATOM: disable TV-out for now""

Doug Chapman (1):
      Bug #14826: Fix a bogus check around xf86SetOperatingState.

Egbert Eich (1):
      Change RMX code to follow the programming algorithm suggested by ATI.

George Sapountzis (7):
      drop mach64 (build system)
      drop mach64
      drop r128 (build system)
      drop r128
      add warning for mach64
      add warning for r128
      radeon: pass distcheck

Jerome Glisse (1):
      radeon: flush & wait for 2d & dma idle after 2d blit

Jiří Paleček (2):
      RADEON: PLL tweaks
      RADEON: remove extraneous line from new pll code

Julien Cristau (1):
      Add a test for __GLIBC__ to CD_Common_Types.h.

Maciej Cencora (1):
      RADEON: remove driver rec copies of mc info, use save rec directly

Matt Turner (4):
      Fix a few warnings
      Properly fix uninitialized variables warnings
      Enable BSR in Log2 functions
      Replace CARD{8,16,32} with uint{8,16,32}_t

Matthieu Herrb (1):
      Makefile.am: nuke RCS Id

Michael Babej (1):
      r600: fix vt switch issue

Michel Dänzer (7):
      radeon: Fix typo flagged by gcc -Wall.
      Fix 16 bit packed YUV XVideo playback on big endian systems with
DRI disabled.
      Handle EXA coordinate limits more cleverly.
      Include config.h, so FGL_LINUX can actually be defined when it's tested...
      Increase default CP timeout.
      Reinstate CARD* types that are part of external interfaces.
      Call DRM_IOCTL_MODESET_CTL ioctl to avoid problems with DRM post
vblank-rework.

Owen Taylor (3):
      Radeon: Omit mask coordinates
      Turn on wrapping when repeating on R100 + R200
      Emulate repeats by drawing in tiles

Owen W. Taylor (2):
      R3xx/R5xx: Fix pitch and clamp mode for repeating textures
      Fix rendering of transformed sources for REPEAT_NONE with EXA on >= R300.

Paulo Cesar Pereira de Andrade (2):
      Ensure symbols used by other modules are visible.
      Compile warning fixes.

Peter Zubaj (1):
      R300: Fix clamping and pitch for textured video

Stephan Wolf (1):
      R3xx+: further fix for IGP chips

Tilman Sauerbeck (1):
      R3xx/R5xx: fix up a8-src-something_with_colors

Tobias Diedrich (1):
      R200/R300: fix gamma setup for overlay

thegraze (1):
      ATOM: add support for DragonFlyBSD

git tag: xf86-video-ati-6.9.0

http://xorg.freedesktop.org/arch ... o-ati-6.9.0.tar.bz2
MD5: afa4873f58ae6fa31fedeb9ed31531db  xf86-video-ati-6.9.0.tar.bz2
SHA1: 558f94b76e1b4c3c4429ab1fdc3e3c8523e414dc  xf86-video-ati-6.9.0.tar.bz2

http://xorg.freedesktop.org/arch ... eo-ati-6.9.0.tar.gz
MD5: 11dd6a53fd98aa32c6f9b518e841df9c  xf86-video-ati-6.9.0.tar.gz
SHA1: 69a7ea39fa3f54b87fee577475ebfc7857525566  xf86-video-ati-6.9.0.tar.gz

jarryson

一直再用那什么git版本，觉得很不错。

james6688

3D支持怎样。。。

zhuqin_83

如果你指compiz，那肯定比官方的好。

zhuqin_83

我上面给出来的两个链接可以研究一下，至少我们可以发现视频回放还有更好的加速方式xvmc，但目前还没开始做。r200以上显卡的opengl加速也是开始没多久，不久的将来应该会有不俗的表现。

jarryson

也许等官方更新了xf86-video-ati我就不用编译了git版本了，呵呵

oydl82

Post by zhuqin_83;1868177
如果你指compiz，那肯定比官方的好。

我的M200主板从来就没有有使用开源驱动开启过3D！