At OSCON 2008, Linux on mobile devices has been an extremely hot topic. This morning Intel's Dirk Hohndel had keynoted about their Atom-based mobile devices and netbooks and this afternoon he had a smaller session where he talked in greater detail about Moblin and what will be known as Moblin 2.0.
It's been about three months since the xf86-video-intel 2.3.0 driver was released and today they have now issued their next quarterly update. Version 2.4.0 of this Intel X.Org graphics driver brings GMA 4-Series support (including their new flagship X4500 IGP), improved i965 EXA Render performance, integrated HDMI support, and SDVO-HDMI support. In addition, there are many more minor fixes and changes, which are detailed in the xf86-video-intel 2.4.0 release announcement. At Phoronix we shared last month that this new driver update would be available in July.
This week Intel has unveiled its Centrino 2 platform that had been codenamed Montevina. The Centrino 2 notebook platform supports 45nm Penryn processors, GMA 4-Series graphics, up to DDR3-1333MHz support, DisplayPort/HDMI/DVI/VGA connectivity, 802.11g/n WiFi with an add-on card for 802.16 WiMax, and many other Intel innovations. Montevina will be the competition to AMD's Puma Platform that was introduced last month at Computex Taipei.
Back in May while open-source X developers were bickering about their TTM memory manager concerns, Intel's Keith Packard chimed in and announced Intel's Graphics Execution Manager, or GEM for short. The Graphics Execution Manager is Intel's replacement of Tungsten's TTM for their xf86-video-intel graphics driver. GEM seems to make memory management simpler and address some of the TTM shortcomings.
If you use the xf86-video-intel driver in a media box or just commonly watch movies using XvMC (X-Video Motion Compensation) for video acceleration of MPEG-2 files, you may want to check out the latest work going on to the Intel driver. No, XvMC doesn't yet support more video formats like Intel's Keith Packard had talked about at FOSDEM. Committed last night were nine XvMC-related patches to the xf86-video-intel git master branch for this open-source graphics driver. There's nothing groundbreaking about these updates, but a number of fixes and the batch buffer size has been increased.
Earlier this week support for Intel's GMA 4-Series Chipsets was committed to the xf86-video-intel DDX driver and OpenGL support in Mesa. In preparation for the forthcoming Intel GMA X4500, the Intel Linux driver has received some added work last night. In particular, HDMI should now be working with Intel hardware. In addition, DisplayPort registers have been added to the Intel driver, but it doesn't look like that connection is fully supported yet. These features should be part of the xf86-video-intel 2.4.0 driver release.
Committed this afternoon to the xf86-video-intel driver development branch at FreeDesktop.org is support for Intel's GMA 4-Series Chipsets. The 4-Series IGPs (namely the Intel X4500) are brand new and is designed to be three times faster than the current Intel GMA X3100. The commit for the G45 Linux graphics support can be viewed here. More information on the Intel G45 chipset can be read at Intel's website. Kudos to the Intel developers for delivering prompt support.
Intel's Eric Anholt has announced on the dri-devel mailing list that they are getting close to merging GEM to master. There was one last bug hitting the GEM folks but that seems to be worked out and the Intel engineers will proceed to merge GEM to master in all three repositories.
PowerTOP is an incredibly handy open-source utility of Intel's that had first premiered a year ago. As of this week, however, PowerTOP has been ported over to OpenSolaris. PowerTOP 1.0 was released for OpenSolaris, which is more or less an equivalent to the Linux version (which is currently at v1.9). This power conservation utility is for both x86 and SPARC architectures. This OpenSolaris port does, however, have some outstanding bugs and currently it doesn't have the power-savings recommendation feature. If you're interested, check out its project page.
Having been in development for a few months now and going through a couple development releases, version 2.3.0 of the xf86-video-intel driver has been released. The 2.3 release represents Intel's Q2'08 open-source driver update and the biggest change is the addition of XvMC (X-Video Motion Compensation) support for the Intel 915/945 IGPs. This graphics driver also has improved stability and a horde of bug-fixes. The release announcement and source download links can be found on the xorg-list. xf86-video-intel 2.3.0 will be part of the forthcoming X.Org 7.4 release.
Intel's Zhenyu Wang has announced the release of xf86-video-intel 126.96.36.1993, which is another Intel X.Org driver test release prior to version 2.3.0. xf86-video-intel 2.3 will officially introduce XvMC support, code cleaning, bug-fixes, improved quirk handling, and other changes.
In Linux graphics news unrelated to the recent ATI/AMD excitements, Intel has just released the xf86-video-intel 188.8.131.521 driver. This driver is meant to serve as the first release candidate for what will become xf86-video-intel 2.3.0. The biggest change since the 2.2 release is that the XvMC support has been merged to master! Intel's open-source XvMC implementation currently works with the Intel 915, 945 and G33 IGPs, but the 965 XvMC support is currently problematic. Intel's XvMC support was previously found in a separate development branch. Aside from these video playback improvements, the Intel 2.3 RC1 driver has many bug-fixes, a few added quirks, code cleanups, and other improvements.
Intel's Eric Anholt has been working on pulling Render improvements for the Intel 965 IGP chipset out of the intel-batchbuffer branch of the xf86-video-intel driver and pushing them into the mainline driver. These mainline Render improvements aren't yet there, but Eric has the code right now in his personal Intel git repository on FreeDesktop.org. Stopping this work right now are color correctness issues.
The xf86-video-intel 2.2.1 driver update has been in the cooker now since this past December, and today on the eve of FOSDEM it has finally been released. The Intel 2.2.1 X.Org video driver is a maintenance release that addresses a large number of bugs since the release of the 2.2.0 driver. Not all blocker bugs were addressed for this release, but an Intel 2.2.2 driver update is possible (permitting enough interest). The next major xf86-video-intel driver update will come in the way of version 2.3. This next update is expected to reduce mode-setting flickers and deliver on other improvements. The xf86-video-intel 2.2.1 driver release announcement can be read on the X.Org mailing list.
We've known since last month that Intel's X.Org display driver would be having a v2.2.1 release soon, and it now looks like it will be released this week. No new features will be introduced until the Intel 2.3.0 driver, but the 2.2.1 release will address a variety of outstanding bugs. Jin Gordon is calling for those in Intel's new driver testing program to test out the latest code in the xf86-video-intel-2.2-branch while using the latest DRM, Mesa, and Xserver components. Presently there are three bugs attached to the Intel 2.2.1 blocker bug.
Last May Intel had introduced PowerTOP as an open-source utility for monitoring a system's power consumption and making recommendations on how to reduce the power consumption thereby extending the battery life for mobile devices. Today, however, Intel has announced a new sibling for PowerTOP: LatencyTOP. Intel's LatencyTOP is a tool geared for developers that are interested in visualizing system latencies. LatencyTOP is able to detect where system latencies are originating and displaying the maximum and average time needed for each task. We'll be exploring this new open-source utility further and will report our results on LatencyTOP in the near future. More information is available from LatencyTOP.org.
Intel's Gordon Jin has announced today that the Intel graphics driver team is looking to establish a closer relationship with the Intel X.Org community of both end-users and engineers. To do this, Intel has created a new community testing program. This Intel testing program is designed to help in the testing of platforms not officially supported by the Intel validation team and bugs that are only experienced in very specific hardware/software configurations.
Intel's Zhenyu Wang has announced the creation of the "xvmc" branch in the xf86-video-intel driver. This new branch features code clean-ups relating to X-Video Motion Compensation (XvMC) and a new framework that will support greater hardware media decoding functionality in the future. The xvmc branch is designed to replace the earlier "xvmc-i915" branch. In addition, libIntelXvMC.so replaces libI915XvMC.so. Along with this new branch and framework, an XvMC option has been added to Intel's driver for the xorg.conf to disable this feature, which comes enabled by default.
The Intel X.Org 2.2.0 Driver was released just about a month ago, but a bug-fix v2.2.1 release is planned for the near future. Jesse Barnes mentioned on the FreeDesktop.org mailing list that the 2.2.1 release will address some of the recent regressions in this open-source Intel video driver. There is a 2.2.1 blocker bug, which currently has five open bugs (another five are already closed), but the two major regressions that this release will cure is X-Video crashing and a pipe enable bug. The Intel EXA performance issues will not be fixed until a later release.
We knew that a new Intel driver was coming and now the xf86-video-intel 2.2.0 driver has finally arrived. This new Intel driver for X.Org closes over 160 bugs (with 96 bugs left to squash), but xf86-video-intel 2.3 should address even more bugs. This new Intel display driver requires X server 1.4, Mesa 7.0, and the DRM module. The release announcement can be read on the X.Org mailing list.
It's been six months since Intel had introduced their xf86-video-intel 2.0 driver for graphics adapters on X.Org and two months since their most recent release (v2.1.1), but we are on the heels of a new Intel driver release. The Intel X.Org video driver will soon reach version 2.2.0 and contains a number of fixes and other enhancements since the xf86-video-intel 2.1 release. Among the many changes include defaulting to EXA as opposed to XAA, 965 IGP fixes, and improved support for G33/35 integrated graphics. Most of the changes, however, are just fixes. Just about 15 minutes ago, Jesse Barnes of Intel had bumped the driver version to 2.1.99, which will be the testing pre-release prior to the xf86-video-intel 2.2 introduction. Jesse had confirmed this information on the X.Org mailing list. Of course, once the Intel 2.2 X.Org driver is introduced, we'll be sure to let you know (you can also be notified by subscribing to our RSS feed).
Mobility talks definitely dominated Intel's Developer Forum last month with information from Montevina and Moorestown to the next-generation Mobile Internet Devices from various vendors. However, talks on multi-core computing definitely ranked high (along with PCI Express 2.0, PCI Express 3.0, and USB 3.0). With technologies like Intel Threading Building Blocks 2.0, Pervasive DataRush, it's time to prepare for a world of massively multi-core computing. Multi-core computing is here and it's here to stay. From assisting in real-time 3D surgery simulations to on-the-fly computerized language translations, many cores will be required.
Have you had your eyes on a computer upgrade with Intel's forthcoming X38 Chipset? Over the course of the next week we will be looking at Intel's new X38 Chipset under Linux and Solaris as we look at the ASUS P5E3 Deluxe and the Gigabyte X38-DQ6 (this motherboard isn't even being announced until the 10th). Among the Intel X38 features include PCI Express 2.0, dual PCI Express x16 slots, Intel Core 2 Quad 1333MHz 45nm CPU support, and removing the overclocking protection from the motherboard. We'll tell you in these first two articles how Intel's X38 works with Linux and whether it's worth the upgrade over Intel's P35 Chipset.
In addition to the PCI Express 3.0 details, information on the next generation USB standard -- USB 3.0 -- were also shared by Intel in San Francisco. The USB 3.0 specifications will be finalized in the first half of 2008, but it will offer backward compatibility with existing USB 1.1/2.0 devices, ten times the performance of USB 2.0, energy efficiency improvements, and support for copper and optical interconnects. A USB 3.0 Promoters Group was also announced by Intel, which includes names like Hewlett-Packard, NEC, Microsoft, and Texas Instruments. The first USB 3.0 adapter was also showcased at IDF Fall 2007.
While running through the information we recorded from the Intel Developer Forum this past week in San Francisco, we didn't comment really on PCI Express 3.0 but wanted to share some additional details from IDF. While native PCI Express 2.0 graphics cards aren't even available yet, PCI Express 3.0 will over twice the bandwidth of PCI Express 2.0, which makes PCI-E 3.0 four times the bandwidth of PCI-E 1.0. PCI Express 3.0 also supports data re-usage, dynamic power management, backwards compatibility, and atomic operations. The release specifications for PCI Express 3.0 will be available in 2009 while PCI-E 3.0 products aren't expected until 2010. Meanwhile, Intel's Xeon 5400 Chipsets and Intel's X38 Express Desktop Chipset will support PCI Express 2.0 this year.
Next week Phoronix will be providing live coverage from the Fall 2007 Intel Developer Forum (IDF) taking place at the Moscone Center. We'll be covering everything from Harpertown / Stoakley benchmarks to Intel's Moblin project and Ian Murdock's speech on Solaris (Project Indiana). There will also be a technology showcase. However, we'd like to hear from you what else you would like covered at the Intel IDF this fall. Check out the Fall 2007 IDF website and then let us know in the Phoronix Forums what you would like to see covered.
On Wednesday we reported on Intel driver fixes for graphics tiling on the 965 chipset (GMA 3000/X3000). Today another batch of commits hit the xf86-video-intel driver (there were 10 commits in the past 6 hours) that should finally fix tiling on the 965 and enabling it by default on this video chipset. Other fixes and code cleaning had also went into this latest round of open-source work. See all of the latest details at FreeDesktop.org.
Two commits to the Intel X.Org driver address previous issues with this open-source driver and the Intel 965 (GMA 3000/X3000) Chipsets. Copying of memory blocks of the sf_kernel when a mask is needed has been corrected by David Airlie. Intel 965 Chipsets with the latest Intel X.Org git code can now handle composite acceleration to A8 destinations. There is, however, high overhead with the Intel compositing setup that results in a text-rendering slowdown. All of the latest Intel X.Org information is available from the xf86-video-intel git.
On the Phoronix Forums we have now created a forum dedicated to the discussion and support for Intel's X.Org driver and their other Linux software projects (such as PowerTOP). This is similar to our AMD (Open and Closed) and NVIDIA forums. Previously discussion had taken place in our other sub-forums. We have started moving the appropriate posts into this new Intel Linux forum and you can check out the new forum here. Contact us with any comments or suggestions you may have.
Yesterday we told you about the latest work with the Intel X.Org driver and today there is another batch of commits to the xf86-video-intel driver. INTEL_VERSION_MAJOR / MINOR / PATCH has been defined using PACKAGE_VERSION_*, fixing pitch in Prepare* functions, fix EXA rendering with tiled front buffer on pre-965 GPUs, and merging branch "origin".
954 Intel news articles published on Phoronix.