Oliver McFadden, the developer behind the Revenge reverse-engineering utility for ATI Radeon GPUs, is hoping to create a free software (GPL licensed) Video BIOS for at least one ATI Radeon graphics card. This is certainly a much larger project than just reverse engineering a driver and is much more risky, but at the same time is very interesting and holds merit. Mark Shuttleworth would also like to see everything down to the device's firmware being free software. The steps Oliver is taking at this point is to examine the AtomBIOS parser, which is open-source as part of the RadeonHD driver (a partial explanation of AtomBIOS). If this open-source BIOS for ATI Radeon GPUs manages to take shape, we'll be sure to cover it here at Phoronix. More information is available on Oliver's blog.
Three days from now the new AMD-sponsored R500/600 driver being written by Novell should be unveiled, but in the meantime it's not stopping developers from continuing further work on the open-source Avivo driver. In the past two days there have been twenty-two commits to the Avivo driver's git repository. These commits fix TMDS register names and other changes based upon AMD's released specifications. As we have already shared with you before, the days of the Avivo driver are limited but the code will remain available after the new AMD open-source driver is out. We've been asked to no longer link to the FreeDesktop.org gitweb as it causes a "Phoronix Effect" with traffic that is apparently too much for their gitweb server to handle. However, hopefully you know the URL anyways so you can check out the latest open-source R500/600 driver code.
While the days of the Avivo driver are likely limited with the new open-source R500/600 driver, this driver has already improved marginally thanks to the publicly released RV630/M56 specifications. There wasn't a commit to the Avivo driver since two weeks ago, but Matthew Garrett took care of the AVIVO_VGA_MYSTERY registers with their real names and values thanks to this new documentation. The AVIVO_VGA_MYSTERY changes can be read about here. The new open-source driver that's being written so far by Novell should be released next week. In the meantime, be sure to check out the specifications (well, if 900+ pages of GPU register specifications interest you).
Ending off the X Developer Summit this year, Matthew Tippett handed off ATI's GPU specifications to David Airlie on a CD (as reported by Daniel Stone). However, the specifications are also now available on the Internet! At http://www.x.org/docs/AMD/ is the location of the documentation where you can freely download the files. Right now there is the RV630 Register Reference Guide and M56 Register Reference Guide. The RV630 Reference Guide is 434 pages long while the M56 Guide is 460 pages. Expect more documentation (and 3D specifications) to arrive shortly. The new open-source R500/600 driver will be released early next week. More information to come soon. Tell us what you think. For more information, read our ATI/AMD's New Open-Source Strategy Explained article.
AMD today has announced its first quad-core processor in the Opteron workstation series. AMD's first quad-core component has been for a while now as Barcelona , but officially it belongs to AMD's Opteron 2300 series. Accompanying their quad-core announcement is a new metric for determining power usage known as Average CPU Power (ACP). Accompanying the Opteron 2300 series are also new power-saving technologies such as AMD CoolCore Technology, Independent Dynamic Core Technology, and Dual Dynamic Power Management. Last but certainly not least, the Barcelona offers improved virtualization performance. In the near future at Phoronix we hope to be delivering Opteron 2300 benchmarks for Linux and Solaris. You can also find Linux benchmarks from Intel's first quad-core processor (known as Clovertown) in our Intel Xeon 5300 Series Preview. Find out more on this new AMD processor series in the AMD press release.
This morning at the X Developer Summit in the United Kingdom, Matthew Tippett and John Bridgman of AMD have announced that they will be releasing their ATI GPU specifications without any Non-Disclosure Agreements needed by the developers! In other words, their GPU specifications will be given to developers in the open. Therefore you shouldn't need to worry about another R200 incident taking place. The 2D specifications will be released very soon and the 3D ones will follow shortly. Specifications for ATI's R300 GPUs should also be out in the future. You may recall that we explained their new open-source strategy last week, but at that time it was still up in the air internally whether or not there would be an NDA for developers. Well, there won't be now so developers can freely access this information and use it for open-source work. Tell us what you think in the forums.
While no ATI fglrx driver is available for Solaris/OpenSolaris or *BSD, now that AMD will be offering up specifications to X.Org developers and an open-source driver, it certainly is promising for any Solaris user depending upon ATI's Radeon X1000 "R500" or HD 2000 "R600" series. The open-source X.Org driver that will be released next week is far from mature, but it should be able to be ported to Solaris and other operating systems using X.Org with relative ease. What AMD announced today is targeted for the Linux community, but it can certainly help out Solaris/OpenSolaris users that use ATI hardware. Especially with "Project Indiana" coming out soon, it's only a matter of time before the open-source R500/600 driver is ported. Tell us what you think in our Solaris forum.
It's been a glorious past couple of months for ATI Radeon X1000 owners that have had very basic open-source support provided by the Avivo display driver. However, it looks like the open-source Avivo driver will soon be going away to a bit-bucket heaven. If AMD sticks to their word, Jerome Glisse (the Avivo lead developer) will discontinue all work on the xf86-video-avivo driver. In fact, he said he "kind of stopped doing real work on it" already and is partially the reason why the Avivo v0.1 driver hasn't been released. He is, however, looking forward to contributing to a new yet to be named open-source driver. Jerome is doing this because he believes AMD is starting to truly work with the open-source community. We will disclose more information on the new open-source driver in just a few hours.
Well, by now you've probably seen our articles today -- AMD 8.41 Display Driver Preview, ATI R300/400 Linux Performance, ATI R500 Linux Performance, ATI Radeon HD 2900XT On Linux, and AMD: Accelerating Open-Source Drivers? -- but the fun is certainly not over. In the coming hours and days we will be delivering additional articles that talk about AMD/ATI's latest efforts in the Linux arena. These articles include Radeon HD 2400 and HD 2600 performance metrics, a Linux versus Windows performance comparison with the new driver, and other articles sought after by the community. There has also been some talk on the Internet about ATI specifications being released at the Linux Kernel Summit, and once our embargo expires (or we're otherwise permitted to talk about it) we will be covering what's up with ATI's open-source side as well. In the meantime be sure to take part in one of our active discussions taking place in the Phoronix Forums. If you have any questions about the new driver don't be afraid to ask.
Revenge, a clean-room reverse engineering utility being developed by Oliver McFadden for Radeon graphics cards, is nearing its 1.0 release. This utility is designed for reverse engineering the ATI graphics cards and their binary driver. Oliver is finishing up work on hardware and software identification, a revenge.sh script, and adding more tests/analysis work before releasing it as Revenge 1.0. Already on the road-map for Revenge 1.1 is texture dumping support. You can find Oliver's announcement on his Live Journal.
Dodji Seketeli and Jerome Glisse have completed porting the open-source "Avivo" R500 driver to libpciaccess, the new PCI infrastructure. Libpciaccess allows X.Org to access the PCI bus and devices with platform independence. Keith Packard has documented on the X.Org Wiki how to implement this new PCI infrastructure with X.Org graphics drivers. The git commit marking the completion of the libpciaccess changes for the Avivo driver can be viewed via gitweb.
With most of the open-source ATI Radeon driver development efforts focused on the RandR 1.2 branch, David Airlie has decided to merge the RandR 1.2 branch with the master tree. All of the Radeon driver activity has been around such things as TV-Out support and all of the RandR 1.2 goodies which resulted in the merge to master. The color problem on the R100/200 series with TV-Out was also fixed as well as other TV-Out fixes and a code cleanup.
Earlier today we told you about a TMDS fix to address some of the display issues present with the open-source "Avivo" R500 driver. Based upon the feedback in our ATI Linux forum this latest work was successful in eliminating some of the TMDS problems. Work has, however, continued in the day with additional fixes. Some bits of the Avivo driver code was cleaned up and the driver can now differentiate between connected digital and analog monitors, a BIOS initialization bit with TMDS_CNTL, and improving TMDS TMDS 0x7880 knowledge. The Avivo utility has also been updated with the latest TMDS register changes. Grab the latest source and technical changes from FreeDesktop.org and stop by the open-source forums for any troubleshooting or to report on the Avivo driver successes and failures.
Yesterday we told you about TMDS fixes for the Avivo driver to hopefully correct display-related problems. Well, a bug crept into the system when the TMDS1 registers were being adjusted when it should have been the TMDS2 registers. This bug was corrected this morning so check out the latest development code if you're still running into problems. Report your results in our open-source ATI/AMD forum with the results, where the lead Avivo developer is an active member of the Phoronix community.
Jerome Glisse (the lead Avivo driver developer) has fixed TMDS issues with the ATI Radeon X1000 series hardware. The latest Avivo git driver changes how the TMDS is programmed (see the git commit for more information). Another git commit updated the supported chipset list against AMD's official list, so all products in the ATI Radeon X1000 "R500" series should now have a PCI ID entry. Those of you experiencing any problems with your display in the past, be sure to try out this latest code and report back with your findings. Discuss in the forums.
With a security vulnerability in ATI's Catalyst driver installer for Microsoft Windows Vista being exposed (dubbed the "purple pill"), engineers have been scurrying to address this problem. The Inquirer is reporting that the patched driver will be released tomorrow (Monday) and that it will be Catalyst 7.8 that gets released. Going by past tradition and what was shared in The Truth About ATI/AMD and Linux, if the ATI Catalyst 7.8 driver for Windows is released tomorrow, so will the next ATI Linux "fglrx 8.40" driver.
A month ago we told you that Avivo v0.1.0 (the open-source ATI R500 X1000 Linux driver) was coming soon, but still we have not seen this release. Talking with Jerome Glisse (the main Avivo developer), the v0.1.0 release is postponed until he or another willing open-source developer fixes issues with the Radeon X1200, X1300, X1400, and X1900 to allow for properly setting the TMDS on these graphics cards. The meaning of a few important registers are still not known. Jerome hasn't had time to work on the Avivo driver recently nor does he know when he will the available time. It looks like the Avivo open-source project could use some extra help.
At SIGGRAPH in San Diego, AMD has introduced five new ATI FireGL workstation graphics cards. These new R600-based FireGL products support Microsoft DirectX 10.0, OpenGL 2.1, up to 320 stream processors, AutoDetect functionality, and a 2GB frame-buffer. In 3D workstation benchmarks the new FireGL graphics cards are over 300% faster than their previous generation GPUs. These new AMD/ATI FireGL cards include the V3600, V5600, V7600, V8600, and the V8650. The FireGL V3600 will sell for $299 USD while the flagship V8650 will cost $2,799 USD. Availability is expected next month and for these new FireGL graphics cards you can expect Linux support to come with the new fglrx driver.
Earlier today we mentioned the recent xf86-video-ati driver commits and now David Airlie has announced on his blog that version 6.6.193 of the open-source Radeon driver has been released. Fixed in the xf86-video-ati 6.6.193 driver is 3D acceleration support for the ATI RS480 chipsets when using the newer version of Mesa as well as better VBL support for lowering power consumption. Like most software releases, there is also a number of bug fixes and code cleanups.
Aside from minor changes to the ATI Radeon driver man-page last month, there hasn't been too much to report on with this open-source X.Org driver (though there has been a wealth of happenings with the Avivo R500 driver). Today, however, there was some Radeon driver updates to the git tree. Luc Verhaegen had cleaned up PortInfo to CRTC mapping and sanitizing blanking and DPMS functions. Meanwhile, David Airlie cleaned up pieces of the DPMS/blank register programming, added "-Wall" for the GCC argument, removed unused variables, and updated the configuration file for what will be the ATI 6.6.193 driver release. All this work had occurred in the xf86-video-ati tree.
The GATOS project has long been known as the path to take for enabling TV output support with the open-source X.Org driver on the Radeon 8500 to 9250 (R200) series, but soon you may no longer need to worry about the GATOS patch. Hanno has announced on the project mailing list that all of those involved with the GATOS project have agreed to move away from the GPL in favor of the MIT/X11 license. This relicensing makes it possible for the GATOS code to be merged with the X.Org Radeon R200 driver.
If you've been unable to run the new fglrx 8.39.4 driver with Fedora 7 x86_64, mmastrac on our forums has written a guide for fixing the fglrx driver with 64-bit Fedora 7. A bit of hex editing is all that's involved for this support. The 32-bit version of Fedora 7 should work "out of the box" with the fglrx 8.39.4 driver. For help and discussion, check out our AMD/ATI Linux Forum.
With the fglrx 8.39.4 driver (or newer), there's a chance you may run into a watermark (similar to the AMD testing watermark) saying that your hardware is unsupported when in fact it is supported. If you run into this unsupported hardware watermark, it's likely that /etc/ati/control is missing. If that's the case for you, try reinstalling the driver or manually copying the control file. The driver contents can be extracted using the --extract argument and the control file is located in common/etc/ati/. If you experience this problem, create a thread in the Phoronix Forums with what distribution you are using and how you had installed the driver initially (and whether adding the control file had removed the watermark).
This morning AMD has re-released the 8.39.4 fglrx driver after it was taken down last week over watermark problems. This latest 8.39.4 revision available from the AMD website should properly produce the needed signature to remove this small annoyance from the display. No other changes had went into this release. If you run into any problems with the 8.39.4 driver, be sure to stop by the Phoronix Forums.
If you've been experiencing the AMD watermark issue with the 8.39.4 Linux driver, there is an unofficial workaround to correct this problem. As is outlined in this Phoronix Forums post, you can obtain the signature value from the ATI/AMD driver package and then manually place it in /etc/ati/signature to remove the "testing use only" watermark. This is not the official workaround sanctioned by AMD and the 8.39.4 fglrx driver was removed from the AMD website earlier today. Find out more in the Phoronix Forums.
Last month was an odd month where an aticonfig issue had caused two fglrx display drivers to be released in the same month (8.38.6 and 8.38.7). Though it seems the AMD release train has run off the tracks once more and we'll probably see another driver out of the AMD camp in the coming days. If you're using the distribution-specific packaging scripts or obtaining the fglrx driver from a distribution repository, it's recommended to not upgrade to fglrx v8.39.4. While not an issue that will harm you, you'll be stuck with an AMD watermark in the bottom right hand corner of the screen that reads "Testing use only".
Today AMD had released the 8.39.4 display driver, which offers support for Fedora 7 and also corrects some bugs. However, at least one serious issue has crept into this driver. If you are running the fglrx driver right now, chances are there's a black and green watermark in the lower right hand corner of your screen that says "AMD Testing Use Only". This watermark is only supposed to display on internal/beta fglrx builds.
Over the past several days we have seen some interesting developments with the Avivo driver due to the increased performance (attributed to shadow frame-buffer support). In the past five hours we have seen six commits to the xf86-video-avivo driver. These commits all pertain to avivotool with some of the changes being to dump 128 bytes of EDID data instead of 64 bytes, verbose i2c information, and making i2c more reliable. Additional information is available from the driver gitweb page.
Last week we reported that version 0.1.0 of the Avivo driver would soon be released and that following day were several noteworthy commits to the driver such as new PCI IDs added and cleaning up the code. Over the weekend, Alexander Larsson has added shadow frame-buffer support. To use the Shadow frame-buffer with the Avivo driver using the latest code from git, the ShadowFB boolean option needs to be added to the device section of the xorg.conf. Jerome Glisse has also fixed the 15 and 16 bits screen color depth with the driver.
Yesterday we told you that the Avivo 0.1 driver will be out shortly, and in the past 24 hours now we have seen a number of commits to the Avivo git. There have been eight commits in the past day that range from adding a number of new PCI IDs, removing the FB manager, cleaning up the initialization code, and properly using the Avivo scaler. In total there are now 68 ATI Radeon X100 "R500" GPUs supported. Last month we posted an article on the open-source ATI R500 driver. You can discuss this driver in the Phoronix Forums.
773 AMD news articles published on Phoronix.