BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:Panthor: Status and future development
DTSTART;VALUE=DATE-TIME:20250929T120500Z
DTEND;VALUE=DATE-TIME:20250929T125000Z
DTSTAMP;VALUE=DATE-TIME:20260510T164505Z
UID:indico-contribution-54-407@indico.freedesktop.org
DESCRIPTION:Speakers: Liviu Dudau (Arm Ltd)\, Florent Tomasin (Arm Ltd)\nP
 resent the current status of the Panthor driver\, update on the features w
 e are working on and what the plans are for the future with regards to Arm
  Mali GPUs.\n\nhttps://indico.freedesktop.org/event/10/contributions/407/
LOCATION:
URL:https://indico.freedesktop.org/event/10/contributions/407/
END:VEVENT
BEGIN:VEVENT
SUMMARY:State of the X.Org Foundation
DTSTART;VALUE=DATE-TIME:20251001T145500Z
DTEND;VALUE=DATE-TIME:20251001T151500Z
DTSTAMP;VALUE=DATE-TIME:20260510T164505Z
UID:indico-contribution-54-430@indico.freedesktop.org
DESCRIPTION:Speakers: Lyude Paul (Red Hat)\nState of the X.Org Foundation\
 n\nhttps://indico.freedesktop.org/event/10/contributions/430/
LOCATION:
URL:https://indico.freedesktop.org/event/10/contributions/430/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Monday Opening Session
DTSTART;VALUE=DATE-TIME:20250929T070000Z
DTEND;VALUE=DATE-TIME:20250929T071000Z
DTSTAMP;VALUE=DATE-TIME:20260510T164505Z
UID:indico-contribution-54-454@indico.freedesktop.org
DESCRIPTION:Speakers: Heinrich Fink (Snap Inc.)\, Michael Wimmer (TU Wien)
 \nhttps://indico.freedesktop.org/event/10/contributions/454/
LOCATION:
URL:https://indico.freedesktop.org/event/10/contributions/454/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Tuesday Opening Session
DTSTART;VALUE=DATE-TIME:20250930T070000Z
DTEND;VALUE=DATE-TIME:20250930T071000Z
DTSTAMP;VALUE=DATE-TIME:20260510T164505Z
UID:indico-contribution-54-455@indico.freedesktop.org
DESCRIPTION:Speakers: Heinrich Fink (Snap Inc.)\nhttps://indico.freedeskto
 p.org/event/10/contributions/455/
LOCATION:
URL:https://indico.freedesktop.org/event/10/contributions/455/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Wednesday Opening Session
DTSTART;VALUE=DATE-TIME:20251001T070000Z
DTEND;VALUE=DATE-TIME:20251001T071000Z
DTSTAMP;VALUE=DATE-TIME:20260510T164505Z
UID:indico-contribution-54-456@indico.freedesktop.org
DESCRIPTION:Speakers: Heinrich Fink (Snap Inc.)\nhttps://indico.freedeskto
 p.org/event/10/contributions/456/
LOCATION:
URL:https://indico.freedesktop.org/event/10/contributions/456/
END:VEVENT
BEGIN:VEVENT
SUMMARY:PanVK: from PoC to Vulkan 1.4 in less than a year
DTSTART;VALUE=DATE-TIME:20251001T080500Z
DTEND;VALUE=DATE-TIME:20251001T082500Z
DTSTAMP;VALUE=DATE-TIME:20260510T164505Z
UID:indico-contribution-54-413@indico.freedesktop.org
DESCRIPTION:Speakers: Erik Faye-Lund (Collabora)\, Lars-Ivar Simonsen (Arm
  Ltd)\nPanVK has been improving quickly over the last year. This talk will
  go through the current status of PanVK\, what happened over the last year
 \, the challenges we've encountered along the road\, and what tasks lays a
 head!\n\nhttps://indico.freedesktop.org/event/10/contributions/413/
LOCATION:
URL:https://indico.freedesktop.org/event/10/contributions/413/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Turnip: Improving performance without compromising correctness
DTSTART;VALUE=DATE-TIME:20251001T071500Z
DTEND;VALUE=DATE-TIME:20251001T080000Z
DTSTAMP;VALUE=DATE-TIME:20260510T164505Z
UID:indico-contribution-54-409@indico.freedesktop.org
DESCRIPTION:Speakers: Danylo Piliaiev (Igalia S.L.)\nTurnip is a Vulkan dr
 iver for Adreno GPUs\, written by reverse engineering the hardware. We als
 o don’t yet have many real-world users running games\, but the driver is
  capable of running games that use D3D12\, VK\, D3D11\, and older APIs. In
  such an environment\, maintaining driver correctness while delivering per
 formance improvements is a challenge.\n\nAt previous XDCs\, I talked about
  how to debug GPU issues and how to improve the GPU debugging experience\;
  however\, the best way to do that is to avoid debugging issues altogether
 . Catching them early\, when changes are being written\, is much more prod
 uctive.\n\nI will talk about:\n- Using single frame game captures in CI to
  test correctness\;\n- Detecting performance regressions with the same CI 
 runs\;\n- Making detailed performance comparisons against the closed-sourc
 e driver.\n\nhttps://indico.freedesktop.org/event/10/contributions/409/
LOCATION:
URL:https://indico.freedesktop.org/event/10/contributions/409/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Addressing Mesa CI pain points: what we've done\, and where we are
  going
DTSTART;VALUE=DATE-TIME:20250929T145500Z
DTEND;VALUE=DATE-TIME:20250929T151500Z
DTSTAMP;VALUE=DATE-TIME:20260510T164505Z
UID:indico-contribution-54-414@indico.freedesktop.org
DESCRIPTION:Speakers: Martin Roukala (né Peres) (MuPuF TMI / Valve contra
 ctor)\, Eric Engestrom (Igalia / Mesa)\, Sergi Blanch Torné (Collabora)\n
 Mesa CI has one of the most complex CI pipelines. While this is great for 
 test coverage\, this also brings challenges in infrastructure\, usability\
 , and sustainability. In this talk\, we will present what we think works\,
  what needs improvement\, and what is currently on our roadmap.\n\n\nWe've
  submitted a workshop proposal to discuss these issues further and agree o
 n the roadmap for the next 12 months. Come watch the talk and join the wor
 kshop to let us know if we've missed any pain point and influence these pl
 ans!\n\nhttps://indico.freedesktop.org/event/10/contributions/414/
LOCATION:
URL:https://indico.freedesktop.org/event/10/contributions/414/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Vulkan CTS Tips & Tricks
DTSTART;VALUE=DATE-TIME:20250929T152000Z
DTEND;VALUE=DATE-TIME:20250929T154000Z
DTSTAMP;VALUE=DATE-TIME:20260510T164505Z
UID:indico-contribution-54-416@indico.freedesktop.org
DESCRIPTION:Speakers: Ricardo Garcia (Igalia\, S.L.)\nThe Vulkan Conforman
 ce Test Suite contains millions of conformance tests to check Vulkan drive
 rs implement the specification correctly. Most people working on Vulkan dr
 ivers and related projects will have to deal with it at some point. In thi
 s talk I will present an overview of the project and some of its internals
 \, and will try to answer some practical questions such as:\n\n- How to be
 st interact with this Khronos project.\n- How to report issues and what ki
 nd of issues are considered interesting.\n- How to build the project faste
 r.\n- How to view result images easily.\n- How to locate the source code o
 f a particular test.\n- How to find tests that use a given feature or exte
 nsion.\n\nhttps://indico.freedesktop.org/event/10/contributions/416/
LOCATION:
URL:https://indico.freedesktop.org/event/10/contributions/416/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Wayback: a compatibility layer for X11 desktops
DTSTART;VALUE=DATE-TIME:20250930T092500Z
DTEND;VALUE=DATE-TIME:20250930T101000Z
DTSTAMP;VALUE=DATE-TIME:20260510T164505Z
UID:indico-contribution-54-417@indico.freedesktop.org
DESCRIPTION:Speakers: Ariadne Conill ()\nWayback is a new project which ai
 ms to provide an alternative to the X.org server and DDX drivers using Way
 land components instead.  This talk will include a review of current progr
 ess\, a discussion of the overall roadmap and a demo of wayback’s capabi
 lities at the time of the talk.\n\nhttps://indico.freedesktop.org/event/10
 /contributions/417/
LOCATION:
URL:https://indico.freedesktop.org/event/10/contributions/417/
END:VEVENT
BEGIN:VEVENT
SUMMARY:KMS offloading in KWin
DTSTART;VALUE=DATE-TIME:20251001T120500Z
DTEND;VALUE=DATE-TIME:20251001T125000Z
DTSTAMP;VALUE=DATE-TIME:20260510T164505Z
UID:indico-contribution-54-424@indico.freedesktop.org
DESCRIPTION:Speakers: Xaver Hugl (Blue Systems)\nI'll give an overview of 
 how KDE's compositor uses KMS offloading for improved efficiency and perfo
 rmance - including overlay planes\, underlays and color operations - and w
 hat problems we've faced when trying to use more of these KMS features.\n\
 nhttps://indico.freedesktop.org/event/10/contributions/424/
LOCATION:
URL:https://indico.freedesktop.org/event/10/contributions/424/
END:VEVENT
BEGIN:VEVENT
SUMMARY:A case study of core affinity for concurrent GPU workload executio
 n on ARM Mali
DTSTART;VALUE=DATE-TIME:20250930T080500Z
DTEND;VALUE=DATE-TIME:20250930T082500Z
DTSTAMP;VALUE=DATE-TIME:20260510T164505Z
UID:indico-contribution-54-421@indico.freedesktop.org
DESCRIPTION:Speakers: Heinrich Fink (Snap Inc.)\nXR compositors rely on ti
 mely execution of GPU workloads to apply a late-stage framebuffer update s
 hortly before scanout. This operation often contends with other concurrent
  rendering workloads occupying the GPU\, and its completion might get dela
 yed past its deadline\, resulting in undesired artefacts.\n\nA widely used
  solution to this problem is to apply some form of GPU-level preemption. H
 owever\, preemption latency still varies with the type of workload to pree
 mpt\, a risk of excessive delays of the time-sensitive workload remains.\n
 \nIn this talk we discuss a different approach: provided we have multiple 
 independent execution GPU cores available\, can we reserve a core for the 
 time-sensitive workload to execute independently of everything else? We pr
 esent a brief case study using the ARM Mali G610 GPU\, a four-core configu
 ration\, driven by the PanVK/panthor open-source Vulkan driver.\n\nhttps:/
 /indico.freedesktop.org/event/10/contributions/421/
LOCATION:
URL:https://indico.freedesktop.org/event/10/contributions/421/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Content Protection on Linux: Challenges\, Architectures\, and Evol
 ving Standards
DTSTART;VALUE=DATE-TIME:20250930T120500Z
DTEND;VALUE=DATE-TIME:20250930T125000Z
DTSTAMP;VALUE=DATE-TIME:20260510T164505Z
UID:indico-contribution-54-419@indico.freedesktop.org
DESCRIPTION:Speakers: Suraj Kandpal (Intel)\nWith the rapid proliferation 
 of high-quality digital media\, effective content protection mechanisms ha
 ve become essential to prevent unauthorized copying\, redistribution\, and
  piracy. This presentation explores the architecture and implementation of
  content protection (CP) technologies in the Linux graphics stack\, focusi
 ng primarily on High-bandwidth Digital Content Protection (HDCP) and Prote
 cted Audio Video Path (PAVP).\nWe begin by examining the motivations behin
 d content protection: protecting creators’ intellectual property and mee
 ting stringent compliance requirements imposed by content providers and li
 censing authorities. The talk outlines the historical limitations of HDCP 
 1.x—its reliance on aging cryptographic algorithms such as RC4\, limited
  interface support\, and vulnerability to reverse engineering—and descri
 bes how HDCP 2.x addresses these shortcomings through modern encryption (R
 SA\, AES)\, enhanced revocation mechanisms\, and expanded compatibility wi
 th newer interfaces like HDMI 2.0\, DisplayPort\, and USB Type-C.\nThe pre
 sentation then details how content protection flows from userspace to the 
 kernel driver. We cover how userspace software negotiates and sets content
  protection properties (CONTENT_PROTECTION and CONTENT_TYPE) via DRM/KMS A
 PIs\, and how kernel drivers enforce these policies\, maintain state trans
 itions\, and ensure authenticated link establishment between the GPU and d
 isplay sink. Special emphasis is placed on how HDCP is typically used only
  for external displays\, while PAVP provides hardware-protected sessions i
 nternally and can be sufficient for trusted embedded panels.\nAttendees wi
 ll gain insight into the specific driver responsibilities—such as signal
 ing userspace when protection states change—and the integration of HDCP 
 and PAVP to achieve a secure end-to-end content pipeline. The session will
  also discuss the practical challenges and trade-offs associated with impl
 ementing CP\, including usability\, latency\, power consumption\, and the 
 complexity of supporting multiple evolving standards in an open-source env
 ironment.\nThe presentation aims to provide developers\, driver maintainer
 s\, and architects with a clear understanding of the current state of cont
 ent protection on Linux\, the reasons behind specific design choices\, and
  the practical considerations when enabling CP in production systems.\n\nh
 ttps://indico.freedesktop.org/event/10/contributions/419/
LOCATION:
URL:https://indico.freedesktop.org/event/10/contributions/419/
END:VEVENT
BEGIN:VEVENT
SUMMARY:RADV: Are we there yet?
DTSTART;VALUE=DATE-TIME:20251001T092500Z
DTEND;VALUE=DATE-TIME:20251001T101000Z
DTSTAMP;VALUE=DATE-TIME:20260510T164505Z
UID:indico-contribution-54-428@indico.freedesktop.org
DESCRIPTION:Speakers: Timur Kristóf (Valve)\, Daniel Schürmann (Valve)\n
 RADV is the Mesa Vulkan driver for AMD GPUs. It started in 2016 as an expe
 riment for GCN 1. Now RADV supports the very latest GPUs and offers cuttin
 g-edge features such as ray tracing\, mesh shading\, DGC and more. We spea
 rheaded a custom compiler backend called ACO. Thanks to the ubiquity of Vu
 lkan on the Linux desktop\, today RADV is the most widely used open source
  Vulkan driver for gaming.\n\nAfter so many years of development\, it's ti
 me to ask\, are we there yet?\n\nThis talk presents the current direction 
 of RADV development\, as well as some of the experience we gained on the w
 ay. It's also a status update on what's been happening in RADV over the pa
 st couple of years.\n\nI'll show the latest developments\, how we maintain
  the code\, how we are endeavouring to share code with RadeonSI\, how we u
 se NIR to improve our compiler stack and how we've retire thousands of lin
 es of legacy code from both drivers. I'll also touch on the topics we're c
 urrently working on and the issues we still face.\n\nhttps://indico.freede
 sktop.org/event/10/contributions/428/
LOCATION:
URL:https://indico.freedesktop.org/event/10/contributions/428/
END:VEVENT
BEGIN:VEVENT
SUMMARY:What’s new in ir3
DTSTART;VALUE=DATE-TIME:20250929T080500Z
DTEND;VALUE=DATE-TIME:20250929T082500Z
DTSTAMP;VALUE=DATE-TIME:20260510T164505Z
UID:indico-contribution-54-422@indico.freedesktop.org
DESCRIPTION:Speakers: Job Noorman (Igalia S.L.)\nir3\, the compiler backen
 d for Adreno GPUs used by Freedreno and Turnip\, has recently gained suppo
 rt for some interesting hardware features. From vectorized instructions to
  register aliasing\, this talk will cover some of those features and the c
 hallenges we faced while implementing them. To lay the groundwork\, I will
  also talk about the internals of ir3\, and about the debugging and revers
 e engineering tools we have at our disposal to help us understand new hard
 ware features.\n\nhttps://indico.freedesktop.org/event/10/contributions/42
 2/
LOCATION:
URL:https://indico.freedesktop.org/event/10/contributions/422/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Support for VM_BIND and sparse textures in Freedreno
DTSTART;VALUE=DATE-TIME:20250929T092500Z
DTEND;VALUE=DATE-TIME:20250929T101000Z
DTSTAMP;VALUE=DATE-TIME:20260510T164505Z
UID:indico-contribution-54-427@indico.freedesktop.org
DESCRIPTION:Speakers: Connor Abbott (Valve)\, Rob Clark (Qualcomm)\nUntil 
 recently\, freedreno\, the upstream kernel driver for Qualcomm Adreno GPUs
 \, has been using the same legacy memory management paradigm as other upst
 ream DRM drivers. While this worked well for the gallium/GL driver\, it wa
 s a poor fit for Vulkan\, where memory is managed explicitly by the user\,
  resulting in unnecessary complexity and CPU overhead\, and it couldn't ha
 ndle the more sophisticated use-cases required by "sparse textures" that a
 re used by some of the latest games. In order to fix this we embarked on a
  major project across the entire stack to rewrite the internals of the ker
 nel driver to use the recently-upstreamed gpuvm subsystem\, expose a compl
 etely new "VM_BIND" uAPI for managing memory\, and take advantage of it in
  turnip\, the upstream Vulkan driver. This then finally allowed us to impl
 ement sparse textures in turnip. In this talk we'll go over the challenges
  of retrofitting VM_BIND into a mature production driver and dealing with 
 the quirks of sparse textures on Adreno.\n\nhttps://indico.freedesktop.org
 /event/10/contributions/427/
LOCATION:
URL:https://indico.freedesktop.org/event/10/contributions/427/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Adding New YUV Formats to Mesa
DTSTART;VALUE=DATE-TIME:20250930T143000Z
DTEND;VALUE=DATE-TIME:20250930T145000Z
DTSTAMP;VALUE=DATE-TIME:20260510T164505Z
UID:indico-contribution-54-423@indico.freedesktop.org
DESCRIPTION:Speakers: Eric Smith (Collabora\, Ltd.)\nThere are a plethora 
 of YUV formats found in the wild\, and it seems like more are added every 
 day. Adding a new format to Mesa requires several steps that may not alway
 s be obvious. In this talk we'll discuss how to add support for a new YUV 
 texture format to OpenGL\, and some of the recent infrastructure work that
  should make this easier.\n\nOur focus is on OpenGL and EGL\, but if time 
 permits we'll briefly discuss how things change for Vulkan.\n\nhttps://ind
 ico.freedesktop.org/event/10/contributions/423/
LOCATION:
URL:https://indico.freedesktop.org/event/10/contributions/423/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Display Next Hackfest Update
DTSTART;VALUE=DATE-TIME:20251001T143000Z
DTEND;VALUE=DATE-TIME:20251001T145000Z
DTSTAMP;VALUE=DATE-TIME:20260510T164505Z
UID:indico-contribution-54-431@indico.freedesktop.org
DESCRIPTION:Speakers: Harry Wentland (AMD)\nIn early July AMD is hosting a
  Display Next Hackfest where compositor and driver developers come togethe
 r to discuss features and pain points and work toward solutions across the
  stack\, including compositors\, wayland protocols\, DRM/KMS interfaces\, 
 and kernel drivers. Previous year's Hackfests have been hosted by Red Hat 
 and Igalia.\n\nProposed topics can be found at https://github.com/hwentlan
 d/2025linuxdisplayhackfest/wiki/Topics-&-Talks\n\nI will attempt to summar
 ize the key discussions/outcomes from the Hackfest\, touching on Color Man
 agement and HDR\, Commit Failure Feedback\, Real-Time Scheduling\, and pos
 sibly more\, depending on time.\n\nhttps://indico.freedesktop.org/event/10
 /contributions/431/
LOCATION:
URL:https://indico.freedesktop.org/event/10/contributions/431/
END:VEVENT
BEGIN:VEVENT
SUMMARY:GPU SVM in DRM / Xe
DTSTART;VALUE=DATE-TIME:20250929T101500Z
DTEND;VALUE=DATE-TIME:20250929T103500Z
DTSTAMP;VALUE=DATE-TIME:20260510T164505Z
UID:indico-contribution-54-437@indico.freedesktop.org
DESCRIPTION:Speakers: Matthew Brost (Intel)\nHeterogeneous Memory Manageme
 nt (HMM) has been part of the Linux kernel for some time\, with prior GPU-
 focused implementations in AMDKFD\, Nouveau\, and OpenNV. However\, integr
 ating HMM correctly—ensuring proper API usage\, race handling\, and lock
 ing—is notoriously complex. This talk introduces GPU SVM\, a new abstrac
 tion layer that streamlines interactions between HMM and the core memory m
 anagement (MM) subsystem\, offering a more accessible and safer API for DR
 M driver developers. We will walk through the design of GPU SVM\, its init
 ial deployment in the Xe driver\, key challenges and learnings from the in
 tegration process\, and discuss future directions for shared virtual memor
 y support in the DRM ecosystem.\n\n[1] https://www.kernel.org/doc/html/v5.
 0/vm/hmm.html\n\nhttps://indico.freedesktop.org/event/10/contributions/437
 /
LOCATION:
URL:https://indico.freedesktop.org/event/10/contributions/437/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Tyr: a new Rust GPU driver
DTSTART;VALUE=DATE-TIME:20250929T125500Z
DTEND;VALUE=DATE-TIME:20250929T131500Z
DTSTAMP;VALUE=DATE-TIME:20260510T164505Z
UID:indico-contribution-54-432@indico.freedesktop.org
DESCRIPTION:Speakers: Daniel Almeida (Collabora)\nThis talk will cover the
  current state of Tyr\, a new GPU kernel mode driver written in Rust for A
 RM Mali GPUs. We will discuss the status of the abstractions in the Rust-f
 or-Linux ecosystem and lay out the roadmap to upstream the driver. We will
  touch upon its relationship to the current C implementation in the kernel
  (Panthor) and how we plan to have these two drivers co-exist for the fore
 seeable future.\n\nhttps://indico.freedesktop.org/event/10/contributions/4
 32/
LOCATION:
URL:https://indico.freedesktop.org/event/10/contributions/432/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Finally fixing Multi-GPU on Linux
DTSTART;VALUE=DATE-TIME:20250929T143000Z
DTEND;VALUE=DATE-TIME:20250929T145000Z
DTSTAMP;VALUE=DATE-TIME:20260510T164505Z
UID:indico-contribution-54-425@indico.freedesktop.org
DESCRIPTION:Speakers: Victoria Brekenfeld (System76)\, Sebastian Wick (Red
  Hat)\nCompositors have long relied on ignoring additional GPUs on a syste
 m as far as they can\, while application configuration is left to users wi
 th obscure environment variables and config files.\n\nWe propose a new ser
 vice with dynamic configuration and policies similar to what other OSes pr
 ovide already and accompanying changes to desktop environments and composi
 tors to finally allow deterministic and user-visible gpu selection.\n\nhtt
 ps://indico.freedesktop.org/event/10/contributions/425/
LOCATION:
URL:https://indico.freedesktop.org/event/10/contributions/425/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Nouveau/NVK Update
DTSTART;VALUE=DATE-TIME:20251001T101500Z
DTEND;VALUE=DATE-TIME:20251001T103500Z
DTSTAMP;VALUE=DATE-TIME:20260510T164505Z
UID:indico-contribution-54-401@indico.freedesktop.org
DESCRIPTION:Speakers: Faith Ekstrand (Collabora)\nA lot has happened in No
 uveau in the last year\, including Vulkan 1.4 support in NVK\; Vulkan conf
 ormance on Maxwell\, Pascal\, Volta\, and Kepler\; and the switch to Zink 
 for OpenGL.  Faith will talk about the advances made in NVK in the last ye
 ar as well as plans going forward.\n\nhttps://indico.freedesktop.org/event
 /10/contributions/401/
LOCATION:
URL:https://indico.freedesktop.org/event/10/contributions/401/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Adapting etnaviv to contemporary Vivante hardware
DTSTART;VALUE=DATE-TIME:20251001T083000Z
DTEND;VALUE=DATE-TIME:20251001T085000Z
DTSTAMP;VALUE=DATE-TIME:20260510T164505Z
UID:indico-contribution-54-415@indico.freedesktop.org
DESCRIPTION:Speakers: Christian Gmeiner (Igalia)\nThe etnaviv driver is be
 ginning to support the latest features added to the newer GPU generations 
 of Vivante hardware. This talk will present the challenges that we’re fa
 cing in adapting the driver and how we’ve improved the CI system notably
  to make these changes possible. A significant focus will be on our stride
 s towards full GLES3 support: what’s landed\, what’s in progress\, and
  what’s next. Attendees will leave with a clear picture of etnaviv’s c
 urrent state\, roadmap\, and how they can get involved or benefit from ong
 oing work.\n\nhttps://indico.freedesktop.org/event/10/contributions/415/
LOCATION:
URL:https://indico.freedesktop.org/event/10/contributions/415/
END:VEVENT
BEGIN:VEVENT
SUMMARY:MSM display and graphics update
DTSTART;VALUE=DATE-TIME:20251001T125500Z
DTEND;VALUE=DATE-TIME:20251001T131500Z
DTSTAMP;VALUE=DATE-TIME:20260510T164505Z
UID:indico-contribution-54-442@indico.freedesktop.org
DESCRIPTION:Speakers: Dmitry Baryshkov (Qualcomm Technologies Inc.)\, Rob 
 Clark (Google)\nMSM display and graphics drivers (both kernel and user spa
 ce) continue to evolve over the years both in terms of feature set \, chip
 sets supported and also complexity.\n\nThis talk focuses on the new featur
 es gained by the drivers (new chipsets\, improved resource allocation for 
 display planes\, VM_BIND\, RustiCL support) and on lessons learned during 
 this development.\n\nhttps://indico.freedesktop.org/event/10/contributions
 /442/
LOCATION:
URL:https://indico.freedesktop.org/event/10/contributions/442/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Machine learning with NIR
DTSTART;VALUE=DATE-TIME:20250930T071500Z
DTEND;VALUE=DATE-TIME:20250930T080000Z
DTSTAMP;VALUE=DATE-TIME:20260510T164505Z
UID:indico-contribution-54-403@indico.freedesktop.org
DESCRIPTION:Speakers: Tomeu Vizoso (Independent contractor)\, Philipp Zabe
 l (Pengutronix)\nExtensions to NIR that allow to express operations on ten
 sors will be explained. New intrinsics will be introduced and there will b
 e quick walkthroughs on how to optimize a NIR ML program with hardware-ind
 ependent and hardware-specific passes. Machine code generation will be als
 o quickly covered\, both to GPGPU units and to fixed-function accelerator 
 hardware.\n\nhttps://indico.freedesktop.org/event/10/contributions/403/
LOCATION:
URL:https://indico.freedesktop.org/event/10/contributions/403/
END:VEVENT
BEGIN:VEVENT
SUMMARY:The Power of Video
DTSTART;VALUE=DATE-TIME:20250930T140500Z
DTEND;VALUE=DATE-TIME:20250930T142500Z
DTSTAMP;VALUE=DATE-TIME:20260510T164505Z
UID:indico-contribution-54-434@indico.freedesktop.org
DESCRIPTION:Speakers: Leo Li (AMD)\nFor the past two years\, there has bee
 n some great work done across display components to optimize power consump
 tion for video playback. This talk will highlight some of this work using 
 Firefox/MPV\, Weston\, and amdgpu as an example. The goal is to demonstrat
 e real power savings\, and to show what can be done today to improve video
  playback power.\n\nhttps://indico.freedesktop.org/event/10/contributions/
 434/
LOCATION:
URL:https://indico.freedesktop.org/event/10/contributions/434/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Display Engine Based Adaptive Sharpness Filter
DTSTART;VALUE=DATE-TIME:20251001T132000Z
DTEND;VALUE=DATE-TIME:20251001T133000Z
DTSTAMP;VALUE=DATE-TIME:20260510T164505Z
UID:indico-contribution-54-447@indico.freedesktop.org
DESCRIPTION:Speakers: Swati Sharma ()\, Nemesa Garg ()\, Adarsh GM ()\nIn 
 this talk\, we will present an adaptive sharpness filter implemented in th
 e Intel graphics display engine [1]\, designed to intelligently enhance im
 age clarity based on content characteristics. Unlike conventional sharpeni
 ng techniques that apply uniform enhancement\, our solution preserves visu
 al quality by adapting dynamically to scene details\, with minimal power a
 nd performance overhead.\nWe will begin with the motivation behind introdu
 cing this feature in the display engine\, followed by a deep dive into the
  kernel-level implementation and its architectural considerations. Next\, 
 we will discuss the validation methodology\, including our use of the IGT 
 GPU Tools framework and Chamelium-based testing [4] to ensure correctness 
 and robustness. Finally\, we will explore the current state of userspace i
 ntegration efforts in open-source compositors such as GNOME [2] and KDE (K
 Win) [3].\nAs part of the upstream journey\, we are currently facing two m
 ain challenges: acceptance of the proposed UAPI and gaining support for th
 e feature at the compositor level. This session aims to open up discussion
  around these issues and encourage collaboration. The implementation is av
 ailable in open source\, and we welcome contributions from the broader com
 munity to help drive this forward.\n\n[1] KMD: https://patchwork.freedeskt
 op.org/series/138754/\n[2] Mutter: https://gitlab.gnome.org/GNOME/mutter/-
 /merge_requests/3665\n[3] Plasma:  https://invent.kde.org/plasma/kwin/-/me
 rge_requests/7689\n[4] IGT: https://patchwork.freedesktop.org/series/13021
 8/\n\nhttps://indico.freedesktop.org/event/10/contributions/447/
LOCATION:
URL:https://indico.freedesktop.org/event/10/contributions/447/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Improvements to the Raspberry Pi driver stack
DTSTART;VALUE=DATE-TIME:20250929T083000Z
DTEND;VALUE=DATE-TIME:20250929T085000Z
DTSTAMP;VALUE=DATE-TIME:20260510T164505Z
UID:indico-contribution-54-449@indico.freedesktop.org
DESCRIPTION:Speakers: Juan A. Suarez (Igalia\, S.L.)\, Ella Stanforth ()\n
 This talk will deliver an overview of the recent development efforts in th
 e Raspberry Pi driver stack focusing on three areas: bugfixing and stabili
 ty improvements\, better performance and OpenGL 3.2 compliance.\n\nThe fir
 st part of the presentation will focus on new features to better align wit
 h OpenGL 3.2 requirements\, namely seamless cubemaps\, 16-bit normalized f
 ormat suppport as well as other features such as robust context support an
 d more.\n\nThe second part will focus on performance improvements. Here we
  will discuss the various improvements implemented in the GL driver and sh
 ow a comparative analysis to illustrate the impact of the optimization wor
 k.\n\nhttps://indico.freedesktop.org/event/10/contributions/449/
LOCATION:
URL:https://indico.freedesktop.org/event/10/contributions/449/
END:VEVENT
BEGIN:VEVENT
SUMMARY:KosmicKrisp\, a Vulkan to Metal layered driver
DTSTART;VALUE=DATE-TIME:20250930T125500Z
DTEND;VALUE=DATE-TIME:20250930T131500Z
DTSTAMP;VALUE=DATE-TIME:20260510T164505Z
UID:indico-contribution-54-410@indico.freedesktop.org
DESCRIPTION:Speakers: Aitor Camacho Larrondo (LunarG)\nThis talk will intr
 oduce the KosmicKrisp Vulkan to Metal driver LunarG has been creating. We 
 will discuss why the driver was created\, and how the Mesa Vulkan driver f
 ramework enabled a fast\, Vulkan conformant implementation.\n\nhttps://ind
 ico.freedesktop.org/event/10/contributions/410/
LOCATION:
URL:https://indico.freedesktop.org/event/10/contributions/410/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Descriptors are Hard
DTSTART;VALUE=DATE-TIME:20250929T071500Z
DTEND;VALUE=DATE-TIME:20250929T080000Z
DTSTAMP;VALUE=DATE-TIME:20260510T164505Z
UID:indico-contribution-54-402@indico.freedesktop.org
DESCRIPTION:Speakers: Faith Ekstrand (Collabora)\nOne of the chief challen
 ges in modern graphics drivers is accessing resources that live outside th
 e shader.  While shader compilation is largely a well-understood problem\,
  the interface between the shader and the rest of the API is still a chall
 enge.  OpenGL and Vulkan\, and D3D12 all have vastly different approaches 
 to solving this problem.  In this talk\, Faith will give an overview of re
 source mapping\, both from an API and a hardware perspective\, and talk ab
 out some of the ways that Vulkan has advanced in recent years to try and m
 ake the situation better.\n\nhttps://indico.freedesktop.org/event/10/contr
 ibutions/402/
LOCATION:
URL:https://indico.freedesktop.org/event/10/contributions/402/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Demo: KosmicKrisp\, a Vulkan to Metal layered driver
DTSTART;VALUE=DATE-TIME:20250930T132000Z
DTEND;VALUE=DATE-TIME:20250930T133000Z
DTSTAMP;VALUE=DATE-TIME:20260510T164505Z
UID:indico-contribution-54-411@indico.freedesktop.org
DESCRIPTION:Speakers: Aitor Camacho Larrondo (LunarG)\nThis demo will show
 case the KosmicKrisp Vulkan to Metal driver LunarG has been creating.\n\nh
 ttps://indico.freedesktop.org/event/10/contributions/411/
LOCATION:
URL:https://indico.freedesktop.org/event/10/contributions/411/
END:VEVENT
BEGIN:VEVENT
SUMMARY:GPU Job Scheduling in DRM: Past\, Present and Future
DTSTART;VALUE=DATE-TIME:20250929T140500Z
DTEND;VALUE=DATE-TIME:20250929T142500Z
DTSTAMP;VALUE=DATE-TIME:20260510T164505Z
UID:indico-contribution-54-433@indico.freedesktop.org
DESCRIPTION:Speakers: Philipp Stanner (Red Hat)\nAll GPU drivers need a lo
 ad balancing mechanism for submitting jobs (i.e.\, command buffers) to the
 ir GPU. The shared infrastructure for this in the Linux kernel's DRM subsy
 stem is the DRM GPU Scheduler. This component has accumulated a large numb
 er of problems over the course of a decade. The talk shall give an overvie
 w over the history of the scheduler and how the problems came to be. Moreo
 ver\, it shows which problems were solved during the last 12 months\, and 
 which still persist. This part of the talk will also serve as an RFC for h
 ow we as kernel developers can work towards avoiding the most significant 
 issues of the category the scheduler experienced.\n\nAnother important asp
 ect of the GPU Scheduler that shall be highlighted is the addition of supp
 ort for firmware scheduling: For GPUs whose firmware performs the actual s
 cheduling (instead of the kernel being responsible for that)\, the DRM GPU
  Scheduler is not an actual scheduler anymore\, but a mere load balancer. 
 This motivates the introduction of a new\, significantly simplified piece 
 of DRM infrastructure: the DRM Jobqueue. The Jobqueue is intended to serve
  as a simpler\, more robust replacement of the DRM GPU Scheduler. First in
 tended users are upcoming drivers such as Nova\, and\, possibly\, later al
 so existing users of the GPU Scheduler.\n\nhttps://indico.freedesktop.org/
 event/10/contributions/433/
LOCATION:
URL:https://indico.freedesktop.org/event/10/contributions/433/
END:VEVENT
BEGIN:VEVENT
SUMMARY:An upstream-first approach to application virtualization
DTSTART;VALUE=DATE-TIME:20250930T083000Z
DTEND;VALUE=DATE-TIME:20250930T085000Z
DTSTAMP;VALUE=DATE-TIME:20260510T164505Z
UID:indico-contribution-54-436@indico.freedesktop.org
DESCRIPTION:Speakers: Alyssa Ross ()\n[Spectrum](https://spectrum-os.org) 
 is a Linux-based desktop operating system which aims to provide improved s
 ecurity by isolating each application running on the system (as well as dr
 ivers where possible) into its own lightweight virtual machine\, while sti
 ll providing a integrated desktop system.  In contrast to previous virtual
 ization-based desktop compartmentalization efforts\, Spectrum takes an ups
 tream-first approach\, building on top of mechanisms like Wayland security
  contexts\, XDG Desktop Portals\, and virtio-gpu.  This talk is an overvie
 w of how various freedesktop.org interfaces\, including those which were n
 ot designed with virtualization in mind\, can be used to enable a much mor
 e cohesive and sustainable compartmentalized system than has been possible
  in the past\, and aims to start a discussion about how virtualization mig
 ht be further accommodated in freedesktop.org protocols and implementation
 s going forward.\n\nhttps://indico.freedesktop.org/event/10/contributions/
 436/
LOCATION:
URL:https://indico.freedesktop.org/event/10/contributions/436/
END:VEVENT
BEGIN:VEVENT
SUMMARY:A Vulkan renderer for Weston
DTSTART;VALUE=DATE-TIME:20250930T101500Z
DTEND;VALUE=DATE-TIME:20250930T103500Z
DTSTAMP;VALUE=DATE-TIME:20260510T164505Z
UID:indico-contribution-54-406@indico.freedesktop.org
DESCRIPTION:Speakers: Erico Nunes ()\nNew Mesa drivers are starting to def
 ault to Vulkan-first (with OpenGL provided as a layer by Zink). One import
 ant part of enabling a new Vulkan driver is to enable WSI (Window System I
 ntegration) and support compositors through e.g. Wayland. The options to s
 upport this WSI development for new Vulkan-first drivers haven't been too 
 many or require significant enablement upfront\, so these platforms may si
 gnificantly delay being able to run a first Wayland compositor which can r
 un some simple Vulkan applications on a display.\nThe Weston Wayland compo
 sitor has been used for this purpose during enablement of OpenGL drivers a
 nd has useful features/backends to support that\, it would be great if Wes
 ton had a native Vulkan renderer with that use case in mind.\nWell\, now i
 t does! https://gitlab.freedesktop.org/wayland/weston/-/merge_requests/167
 9\n\nThis talk aims to introduce the state and design choices of the Westo
 n Vulkan renderer\, bring up some of the current challenges\, and start a 
 discussion on its applications (Vulkan-first platforms\, development\, tes
 ting and CI for Vulkan drivers\, actually using it as a compositor\, etc).
 \n\nhttps://indico.freedesktop.org/event/10/contributions/406/
LOCATION:
URL:https://indico.freedesktop.org/event/10/contributions/406/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Remarks on 2D Hardware Acceleration Support
DTSTART;VALUE=DATE-TIME:20250930T145500Z
DTEND;VALUE=DATE-TIME:20250930T151500Z
DTSTAMP;VALUE=DATE-TIME:20260510T164505Z
UID:indico-contribution-54-418@indico.freedesktop.org
DESCRIPTION:Speakers: Paul Kocialkowski (sys-base)\nNumerous chips still c
 ome with dedicated 2D hardware accelerators that are often not part of big
 ger gpu unit. These devices are usually very efficient for simple composit
 ing and transform works and can be useful to accelerate display compositin
 g or be used as part of multimedia pipelines.  They also typically offer l
 ow latencies and low power cost.\n\nUnfortunately they are mostly unsuppor
 ted in Linux and there is no standard userspace API and implementation to 
 leverage them. As of today\, most users looking to use these units are lef
 t with proprietary APIs and binary libraries.\n\nThis talk will present a 
 brief overview of 2d acceleration hardware\, exisiting proprietary APIs an
 d the current state of things in mainline Linux. Some thoughts and proposi
 tions to properly support these units will then be discussed.\n\nhttps://i
 ndico.freedesktop.org/event/10/contributions/418/
LOCATION:
URL:https://indico.freedesktop.org/event/10/contributions/418/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Driving multiple displays with Superbuffer composition
DTSTART;VALUE=DATE-TIME:20251001T140500Z
DTEND;VALUE=DATE-TIME:20251001T142500Z
DTSTAMP;VALUE=DATE-TIME:20260510T164505Z
UID:indico-contribution-54-435@indico.freedesktop.org
DESCRIPTION:Speakers: Shashank Sharma ()\nNow a days many consumer electro
 nics segments demand several display outputs coming out of a single Graphi
 cs card. Like modern Automotive infotainment segment demands 12-15 HDMI di
 splay outputs being driven from a single GPU\, so that they can be used to
  display navigation panel\, control panel\, mirrors\, multiple media playe
 rs and several independent status readers to show speedometer\, maps\, odo
 meters\, temperature\, music etc. But due to limited number (3-4) of displ
 ay pipelines available in its Display engine\, the GPU can only drive 3 or
  4 independent output displays per card\, which leaves a gap for a solutio
 n in this market.\n\nThe method proposeed here presents a unique method co
 mbining the synergy of SW and HW to solve this problem. The software creat
 es graphics superbuffers to encode multiple display outputs in a single la
 rge framebuffer\, and drives it in a single display pipeline by using GPU'
 s capacity to drive high resolutions and then uses a smart splitter HW at 
 the end of the display pipeline to splits and push these splitted buffers 
 into multiple display outputs.\n\nIf implemented precisely\, this method w
 ill allow a single GPU to drive 12-16 output displays.\n\nhttps://indico.f
 reedesktop.org/event/10/contributions/435/
LOCATION:
URL:https://indico.freedesktop.org/event/10/contributions/435/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Graphic Testing without hardware: discovering the power of VKMS!
DTSTART;VALUE=DATE-TIME:20250929T132000Z
DTEND;VALUE=DATE-TIME:20250929T133000Z
DTSTAMP;VALUE=DATE-TIME:20260510T164505Z
UID:indico-contribution-54-448@indico.freedesktop.org
DESCRIPTION:Speakers: Louis Chauvet ()\nAs a userspace developer\, creatin
 g graphical applications is a common task. While software and frameworks l
 ike Weston or Qt make it straightforward to display content on a screen\, 
 ensuring compatibility across various displays poses significant challenge
 s. Users may have different screen sizes\, resolutions\, and capabilities\
 , making comprehensive testing complex. How will your application perform 
 on a small screen? What about an extremely large one? How does it react to
  displays being connected/disconnected? Moreover\, if your application mus
 t run on diverse devices\, how will it leverage or adapt to varying hardwa
 re capabilities?\n\nVKMS addresses these challenges by enabling graphic te
 sting without the need for physical hardware. VKMS is a Linux kernel DRM d
 river that allows you to emulate a wide range of display configurations on
  any device. This talk will introduce VKMS\, highlight its current capabil
 ities\, demonstrate practical test examples\, and provide details on addit
 ional features we are currently working on upstreaming into VKMS.\n\nhttps
 ://indico.freedesktop.org/event/10/contributions/448/
LOCATION:
URL:https://indico.freedesktop.org/event/10/contributions/448/
END:VEVENT
END:VCALENDAR
