[Mesa-dev] Distros: Mesa 20.0 defaulting to the 'iris' driver instead of 'i965'
Hello, As most of you know, we've been developing a new 3D driver called 'iris' for recent Intel hardware (Gen8+), based on the Gallium driver framework used by the rest of the Mesa community. After two years of development, we think that it's ready for production, and recommend it as the default driver starting with Mesa 20.0. Building and Selecting Drivers -- We recommend building and installing both drivers side-by-side: meson ... \ -Dgallium-drivers=iris, ...other drivers... \ -Ddri-drivers=i965, ...other drivers... \ -Dprefer-iris=true A new build option, -Dprefer-iris, will change Mesa's DRI driver loader to select 'iris' as the default driver for all hardware it supports, rather than i965. Starting with Mesa 20, this will default to 'true', but it's likely wise to set it explicitly. To continue defaulting to i965, you can set -Dprefer-iris=false, but still build and ship both. Users can also override the driver selection at run-time. To change it for all applications, they can create a ~/.drirc file containing: This can also be set via an environment variable for one-off testing: dri_driver=iris ./app This way, if users encounter a bug with iris, they can fall back to i965. (Or, if you ship i965 as the default, they can try iris.) Gen12+ (Tigerlake) hardware is only supported by iris. Dependencies on the X Server We recommend upgrading to xserver 1.20.7 and libepoxy 1.5.4 before shipping iris. The modesetting driver in 1.20.7 will consult Mesa to choose the DRI driver, instead of hardcoding i965. So whatever decision you make for Mesa will apply to the whole system. With older X servers, you may have a case where apps use iris for GL, but modesetting+glamor uses i965. Although this can work, we don't recommend shipping such a configuration. Why Switch? --- The new iris driver is significantly more efficient than i965. Recent benchmarking [1] shows that it's roughly 5-10% faster at many actual applications. In simple CPU overhead tests, iris can issue 5-20x as many draw calls per second as i965. Our hope is that this lower overhead will translate to better laptop battery life and higher performance. [1] https://www.phoronix.com/scan.php?page=news_item=Intel-9900KS-Gallium3D-Perf Future optimization work will primarily target iris (GL) and anv (Vulkan), with very little performance work happening on i965. We have no plans to drop support for i965 - it continues to be the only 3D driver for Intel Gen4-7.5 hardware, as well as Cherryview/Braswell. Feedback & Bug Reporting We are definitely interested in any issues you encounter with iris. Please report bugs as usual to Mesa's Gitlab issue tracker: https://gitlab.freedesktop.org/mesa/mesa/issues Please let us know if you have any questions. Thanks! --Ken signature.asc Description: This is a digitally signed message part. ___ mesa-dev mailing list mesa-dev@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/mesa-dev
[Mesa-dev] Merging experimental r600/nir code
Dear all, has anybody any objections if I merge the r600/NIR code? Without explicitely setting the debug flag it doesn't change a thing, but it would be better to continue developing in-tree. Best, Gert ___ mesa-dev mailing list mesa-dev@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/mesa-dev
Re: [Mesa-dev] [ANNOUNCE] Mesa 20.0 branchpoint planned for 2020/01/29, Milestone opened
On Thu, Jan 23, 2020 at 9:44 AM Jason Ekstrand wrote: > > Should we make iris the default for 20.0? I think it's time. Sadly, gitlab > milestones don't let you comment on them. It does seem like it's time to flip that switch. ___ mesa-dev mailing list mesa-dev@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/mesa-dev
Re: [Mesa-dev] [ANNOUNCE] Mesa 20.0 branchpoint planned for 2020/01/29, Milestone opened
Should we make iris the default for 20.0? I think it's time. Sadly, gitlab milestones don't let you comment on them. On Wed, Jan 22, 2020 at 12:26 PM Dylan Baker wrote: > Hi list, due to some last minute changes in plan I'll be managing the 20.0 > release. The release calendar has been updated, but the gitlab milestone > wasn't > opened. That has been corrected, and is here > https://gitlab.freedesktop.org/mesa/mesa/-/milestones/9, please add any > issues > or MRs you would like to land before the branchpoint to the milestone. > > Thanks, > Dylan > ___ > mesa-dev mailing list > mesa-dev@lists.freedesktop.org > https://lists.freedesktop.org/mailman/listinfo/mesa-dev > ___ mesa-dev mailing list mesa-dev@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/mesa-dev
Re: [Mesa-dev] -fno-common build failures (default from upcoming gcc release 10)
I've fixed most of the problems in 2385, but the problem still remains in omx. Could someone who's familiar with omx please take a look? Alternatively, maybe we can try adding -fcommon to just omx, until this issue is resolved. On Tue, 2020-01-21 at 18:49 +0100, Bas Nieuwenhuizen wrote: > I think this ended up spawning a bunch of work in > https://gitlab.freedesktop.org/mesa/mesa/issues/2385 > > On Mon, Jan 20, 2020 at 3:41 PM Stefan Dirsch > wrote: > > Hi > > > > Starting from the upcoming GCC release 10, the default of -fcommon > > option will > > change to -fno-common. Due to this we're going to see a lot of > > build failures > > in Mesa drivers like > > > > multiple definition of `syncobj_handle'; src/amd/vulkan/9198681@@ > > vulkan_radeon@sha/meson-generated_.._radv_entrypoints.c.o (symbol > > from plugin):(.text+0x0): first defined here > > [ 213s] > > /usr/lib64/gcc/x86_64-suse-linux/9/../../../../x86_64-suse- > > linux/bin/ld: > > src/amd/vulkan/9198681@@vulkan_radeon@sha/radv_wsi_wayland.c.o > > (symbol from > > plugin): in function > > `radv_GetPhysicalDeviceWaylandPresentationSupportKHR': > > > > I'm wondering if there is already anybody working on fixing these > > issues or is > > it recommended to workaround it by setting -fcommon manually > > somehow? How can > > the latter be done when using meson/ninja as build tool? > > > > Thanks, > > Stefan > > > > Public Key available > > -- > > Stefan Dirsch (Res. & Dev.) SUSE Software Solutions Germany GmbH > > Tel: 0911-740 53 0Maxfeldstraße 5 > > FAX: 0911-740 53 479 D-90409 Nürnberg > > http://www.suse.deGermany > > > > (HRB 36809, AG Nürnberg) Geschäftsführer: Felix Imendörffer > > > > ___ > > mesa-dev mailing list > > mesa-dev@lists.freedesktop.org > > https://lists.freedesktop.org/mailman/listinfo/mesa-dev > ___ > mesa-dev mailing list > mesa-dev@lists.freedesktop.org > https://lists.freedesktop.org/mailman/listinfo/mesa-dev ___ mesa-dev mailing list mesa-dev@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/mesa-dev
