Phoronix News Bot
Samsung Has G2D Driver, Virtual Display For Linux 3.4
Phoronix: Samsung Has G2D Driver, Virtual Display For Linux 3.4
Besides the DRM work already piling up for Linux 3.4, there's more. The Samsung developers responsible for the Exynos graphics driver have sent in their "-next" pull request, which brings several new features, including the basis of 2D acceleration for this open-source ARM graphics driver. There's also a virtual display driver that could be used for handling wireless displays...
http://www.phoronix.com/vr.php?view=MTA3MDE
Senior Member
Ah ha... so they're adding open 2D acceleration for a product that will be used almost exclusively with Android (now requiring 3D acceleration), in order to get the interface for their 3D driver added to kernel.
Senior Member
Huh? They have no 3d core, it's ARM Mali for now.
Senior Member
What are you talking about? All they are IS a 3d core...Originally Posted by curaga
Senior Member
Samsung is taking over
Own system-on-chip solutions, kernel contributions, open source device drivers.
Samsung is taking over the smartphone game!
Senior Member
Posted by curaga
Huh? They have no 3d core, it's ARM Mali for now.droidhacker Ah ha... so they're adding open 2D acceleration for a product that will be used almost exclusively with Android (now requiring 3D acceleration), in order to get the interface for their 3D driver added to kernel.
im not sure what you two mean ?. the mali provides all these 2D, 3D and video encode/decode core code blocks in SOC and has since its initial Falanx beginings, a long time ago now.
also you can get Linaro Android Builds as well as generic Linaro ARM Linux builds today and switch between them at will with a rooted device OC.
there's also a Samsung Exynos 5250 Dual-core Cortex A15 and GAIA SMDK Development Board available for anyone wanting to test that now/soon ReadAnnounced in November 2011
more: http://www.cnx-software.com/2012/03/...#ixzz1p5yR3p7B
", the new Samsung Exynos 5250 processor features 2 Cortex A15 clocked at 1.7 GHz or 2 GHz, with a Mali-T604 GPU and support for dual-channel 800 MHz LPDDR3 RAM that allows for a data bandwidth of up to 12.8 GB/s.
The processor is manufactured using 32-nm HKMG (High-K Metal Gate) technology that reduces leakage by 30% proving lower power consumption.Samsung claims the new Exynos 5 processors are twice as fast and consume twice as less power than its previous Exynos 4 application processor based on Cortex A9.
This seems to confirm TI OMAP 5 vs. Nvidia Tegra 3 benchmark results.The Exynos 5250 supports embedded DisplayPort (eDP) interface up to WQXGA resolution (2560×1600), which is handy for Samsung since they are also working on a WQXGA 10.1″ LCD Panel for tablets. This interface is also compatible with PSR (Panel Self Refresh) technology which instructs the application processor not to send image data to the LCD panel when the set is displaying still image in order to reduce power consumption.
Samsung Application Processor also support different booting method via SATA, UART, USB 3.0 and eMMC 4.5 which implies that the Exynos 5 processors will be used beyond smartphones and tablets.The Exynos 5 processors also have pretty decent multimedia hardware support with a 8 MP image signal processor and support for 1080p60 video playback with the most common video codecs (h.264, h.263, VC1, MPEG2 and MPEG4).
Samsung will also start to manufacture the Exynos 5450 quad-core Cortex A15 allegedly with an ARM Mali-T658 GPU later this year. That’s the first quad-core Cortex A15 processor I’ve ever heard of.ARM’s Katie Morgan interviewed Samsung during MWC 2012 and you can have a peak at the Exynos 5 development board in the video below.
Read more: http://www.cnx-software.com/2012/03/...#ixzz1p5yt2DHq"
Last edited by popper; 03-14-2012 at 08:18 AM.
Senior Member
Just like every other GPU, the 2D hardware has been dropped. 2D acceleration is done by the 3D core. This new driver provides 2D acceleration on Exnos/Mali, it is thus accessing the 3D core.
You will recall that the the kernel devs have previously REJECTED interfaces that are to be used exclusively by closed blob drivers, excuse being just that.
My point is that it appears that they are building a 2D driver **AS AN EXCUSE** in order to get a kernel interface that they could ALSO access with their blob 3D drivers. I.e., satisfy the kernel requirements that there be some usable open source driver that utilizes that interface.
Their objective here does NOT seem to be to provide open source 2D acceleration. Its to gain the interface for their 3D driver.
Note that I'm NOT saying that there is anything wrong with this approach. In fact, the opposite, since they ARE complying with upstream requirements.
Posting Permissions
Reply With Quote