Intel just ATE Nvidia's Lunch! AV1 Hardware Encoding is HERE

The Future of Video Encoding: Intel's Arc Technology and Its Implications

In recent years, the video encoding landscape has undergone significant changes with the advent of new technologies like AV1, which is poised to become the new standard for video compression. One company that's been at the forefront of this technology is Intel, which has recently announced its Arc platform, a range of integrated graphics processing units (iGPUs) and discrete GPUs designed specifically for gaming, content creation, and streaming applications.

Intel's Arc platform is not just about gaming; it's also focused on video encoding and decoding. The company claims that its new technology will provide a significant boost in performance and power efficiency compared to traditional CPU-based encoders. This is particularly exciting news for streamers and video editors who have long been waiting for faster and more efficient ways to encode their content.

The slide mentioned in the transcript highlights Intel's claim that its Arc platform can decode up to 8K60 12-bit HDR, while encoding up to 8K 10-bit HDR. While these specifications sound impressive, it's essential to note that actual performance may vary depending on various factors, including the specific hardware configuration and software implementation.

To test the limits of Intel's AV1 encoder, we'll need to conduct further testing to see how well it performs compared to CPU-based encoders like those enabled in recent OBS updates. However, with the support of popular video editing software like DaVinci Resolve and Premiere Pro, which have already been updated to encode with AV1 on Intel Arc GPUs, the prospects for improved performance look promising.

Another significant development is Intel's "Deep Link" technology, which allows the CPU, iGPU, and dGPU to work together to speed up workflows. This dynamic power sharing feature can be particularly useful in laptops, where power management is a critical aspect of battery life. By dynamically allocating power between the CPU and GPU, Deep Link enables both components to remain fed and working at their best.

The "Hyper Encode" feature takes this technology even further by splitting video files into GOPs (groups of pictures) and having the dGPU and iGPU encode alternating frames if both have an AV1 encoder relevant to the work. This process can bring a massive performance improvement, up to 60% transcoding performance jump, which is a significant advantage for streamers and video editors.

For computer workloads, Intel's "Hyper Compute" feature provides a similar solution by dynamically balancing workloads between CPU, iGPU, and dGPU using Machine Learning services. This technology can make a substantial difference in workflows that involve tasks like video enhancement, where the AI-powered Video EnhanceAI application was 2.5x faster on the new Arc GPU than integrated graphics.

One of the most significant aspects of Intel's Arc platform is its focus on workloads traditionally associated with gaming, which often overlap with those used for streaming and video editing. The "Creator Studio" feature within the Arc Control app provides tools like automatic highlight creation, sharpening filters, AI webcam processing (including background removal, blurring, and replacement), and auto framing.

The final note worth mentioning is Intel's Arc Control app itself, which can be seen as a clone of Nvidia GeForce Experience. The app offers features like driver management, performance tuning, hardware monitoring, game performance stats, and even AI-based features like sharpening filters and AI webcam processing. While some might view this as a slight duplication, the inclusion of features like automatic highlight creation and AI-powered video enhancement make Arc Control a more comprehensive solution for content creators.

As we look ahead to the future of video encoding, it's clear that Intel's Arc technology is going to play a significant role. With its promising performance capabilities and innovative features, this platform has the potential to revolutionize workflows in streaming, video editing, and computer workloads alike. Whether you're a seasoned streamer or an emerging content creator, staying informed about Intel's latest developments can help you make the most of this exciting new technology.

Get Subscribed for More Content

Make sure you don't miss any future updates on AV1, Arc Technology, and other related topics by getting subscribed to my channel. I'll be sharing more in-depth guides, tutorials, and analysis as we continue to explore the world of video encoding and content creation.

Lighting and Streaming Audio Guides

If you're new to streaming or looking to improve your setup, be sure to check out my comprehensive lighting guide and audio guide. These resources will provide you with a solid foundation for creating high-quality streams that engage and retain your audience.

"WEBVTTKind: captionsLanguage: enIntel just ATE NVIDIA’S LUNCH for creators!From outright cloning GeForce Experience,a DLSS competitor, to adaptive sync, to beatingthem to the punch with AV1 encoding, the newArc Alchemist line of dGPUs from Intel areabsolutely wild.Intel had their announcement presentationtoday for their new dedicated GPU hardware,which they’re calling Arc, shipping nowin laptops with higher end chips and add-in-cardsteased for later in the year.Let’s touch on hardware, then talk aboutgaming, and wrap up with content creation.The hardware itself seems like a substantiallyimpressive leap over Intel’s previous integratedGPU tech.The graphics chips, specifically called XeHPG, feature new processing cores, a new encodeand decode media engine, a new display outputengine, and graphics pipeline.The Xe HPG hardware is broken up into whatIntel is calling “Render Slices” - scalableup to 8 on the higher end cards.Render Slices are comprised of their “Xe-cores”which could be seen as normal GPU processingcores on any GPU, plus dedicated hardwarefor thread sorting, geometry sample processing,as well as dedicated hardware raytracing unitswhich support DirectX Raytracing or DXR, andVulcan RT, as well.These cores are optimized for all of the featuresof DirectX12 Ultimate for gaming, which includesfeatures like Variable-Rate Shading, Raytracing,Mesh Shading, Sampler Feedback, and so on.This is kinda great to hear as the originalrun of dedicated GPUs from Intel we heardabout were primarily for server workflowsand not gaming parts, there is a nearly completefocus on gaming in this product stack.AI processing both for workloads and gamingcame up here with each Xe-core getting 161024-bit “Matrix Engines” which are supposedto be optimized for AI, along with 192KB ofshared Level 1 Cache or shared local memoryon each core.The AI gets involved specifically with gamingin the form of AI upscaling - yes, Intel isreleasing their own DLSS competitor.But instead of being temporal AA upscalinglike AMD’s FSR, it’s supposed to actuallybe leveraging the AI hardware as with Nvidia’ssolution.This is coming early summer and they alreadyhave a small list of games integration thetech.Unfortunately, XeSS, as they’re callingit, does need to be integrated on a per-gamebasis, which will slow down its adoption - butI do believe this kind of tech is going tobe a default expectation for gaming as timegoes on, so it’s great to know Intel hassolutions, too.To output your games, Intel CLAIMS they haveoutput capabilities “for the future” intheir new Xe Display Engine - but by defaultwe’re looking at HDMI 2.0b and DisplayPort1.4a.There is a note in the slides that it’sDisplayPort 2.0 “ready”, but nothing aboutHDMI 2.1.This doesn’t sound very future-proof tome, especially when they’re talking about360hz at 1440p, and 8K60 and 4K120 in HDR.I really don’t get why they wouldn’t haveHDMI 2.1 ready to go at this point.Regardless, the chips can handle 4 displaypipes at full res, which is very handy.To handle all of this, 3 types of displaysyncing are supported.There’s adaptive sync, which is great.But then Intel also announced their 2 new“sync” modes - Speed Sync and Smooth Sync.Speed Sync sounds like what Nvidia Fast Syncis supposed to do, but doesn’t always do- maintain low latency, zero-tearing monitorsync, without enabling VSync.It just does a faster and better job of flippingframes over as they become available.Smooth Sync on the other hand, apparentlyaims to hide screen tearing a bit by blurringand dithering the differences between thetwo frames being displayed at the same time?In their sample screenshot it literally justlooks like they’re just applying a blurfilter at the edges of the tear, which I’mnot sure how they’re doing that withoutlatency, and it’s just a simulation, sowe’ll have to test more closely once wehave hands on the hardware.Gaming is cool, but the new Media Engine iswhere my interest really lies.The new Intel Arc GPUs give us the first hardwareencoders for AV1 - the open source video encoderthat I’ve been hyping up for YEARS on thechannel.AV1 enables vastly higher qualities at lowerbitrates and once more widely adopted willallow for high resolution and high frameratestreaming on sites like YouTube, AND allowfor better-looking, lower-bandwidth playbackfrom websites like Netflix.YouTube is already COVERED in AV1 videos,if you enable it in your playback preferencesand have the hardware to decode it - but wehaven’t had accessible encoders yet.This is exciting.I think the GPU can also encode VP9, too?That’s also nice.The slide just says “up to 8K60 12-bit HDRdecode” and “up to 8K 10-bit HDR encode”so we’ll have to do some testing to seejust how high the AV1 encoder can handle,specifically, and how the quality comparesto the CPU encoders enabled in the recentOBS update and so on.Get subscribed so you don’t miss it - AV1is really the future.Apparently FFMPEG, Handbrake (which just usesFFMPEG), XSplit, DaVinci Resolve, and PremierePro are already all updated to encode withAV1 on Intel Arc GPUs, so that’s awesome.Premiere usually takes years to implementnew encoders, after all.Relating to encoding and general content creationworkflows, Intel described their “Deep Link”technology.This allows the CPU, iGPU, and Arc dGPU toall work together to speed up workflows anymore.The dream of multi-GPU encoding and videoediting seems to be here with this new lineup.Dynamic Power Share is specifically relevantto laptops, but allows the GPU to be polledat 100ms intervals to dynamically allocatepower between the CPU and GPU to keep bothfed and working at their best, and this canbe slowed down when on battery to optimizefor battery life.Then there’s “Hyper Encode” - and thisis literally just splitting up video filesinto GOPs or groups-of-pictures, just havingthe dGPU and iGPU encoding alternating frames,if both the dGPU and iGPU have the same encoderrelevant to the work - to encode much fasterand bring it together at the end.For H264 and HEVC this will be a MASSIVE performanceimprovement that streamers and video editorseverywhere would IMMEDIATELY be able to feelthe difference in.For AV1, the iGPU won’t have an AV1 encoder,so this won’t get used.This can bring up to a 60% transcoding performancejump, which is massive, honestly.All those questions I get about using multi-GPUsfor streaming or video editing...now have an answer.This workflow similarly applies to computerworkflows, too, with “Hyper Compute.”This uses a Machine Learning service to dynamicallybalance workloads between CPU, iGPU, and dGPU,to make sure the right workloads are assignedto the right places.Specifically with this tech, Video EnhanceAI - which I previously showed off upscalingold family VHS tapes with - was 2.5x fasteron the new Arc GPU than integrated graphics,and 2.9x faster when leveraging both the ArcGPU AND integrated graphics.I use Video Enhance AI a lot, but less thanI’d like due to speed - I’m excited tosee how this can help that workflow.Overall with all of this technology, it isVERY laptop focused - which is neat, but Ireally hope a lot of this workload sharingcan still work out nicely on desktop, as well.Lastly, I wanted to mention Intel’s newArc Control app.There’s no easy way to put this, it’sjust straight-up a clone of Nvidia GeForceExperience.Perhaps better, as it provides some updatesthat many of us wish GFE would have gottenby now.Arc Control is a “one stop shop” app thatallows you to manage and update drivers, tuneand tweak performance, view hardware and gameperformance stats, manage and optimize gamesettings - you know everything GeForce Experiencedoes, even with an in-game overlay.But it goes beyond that and integrates whatwould be the Nvidia Broadcast app, too.Arc Control has a “Creator Studio” featureallowing you to stream or record your gameplayand create automatic highlights.It looks fairly basic, but there seems tobe a “Sharp Stream” feature that theydidn’t mention, I can assume is either asharpening filter or an AI Upscale feature,along with AI webcam processing includingbackground removal, blurring, and replacement,along with auto framing.The only missing feature would be AI backgroundnoise removal on your mic!And of course, they had to add a jab thatArc Control does NOT require a login - whichwe always complain about being required forGeForce Experience.Again, the quality and performance of allof these features remains to be seen, we stillhave to get our hands on the tech, and onlythe first lower-end laptops are releasingnow.More on the actual add-in cards and more powerfullaptops is coming later in the year.But let’s just say, it’s an exciting timefor technology.Make sure you’re ready to stream with thisnew tech by watching my streaming audio guidehere, or my lighting guide, here.I’m EposVox the stream professor and remember:Be kind, rewind.\n"