AWS for M&E’s Post

It's #ibc2023 show time and at Spin Digital we are very happy to share some exciting news. Together with Intel Corporation we are officially presenting our new live 8K VVC/H266 encoder. The results are ready to be shared: using our optimized software encoder (Spin Enc Live) running on a 4th Generation Intel Xeon Processor we can encode 8K HDR video with 60-fps with #VVC in real-time, in broadcast-grade quality, and live stream it over the Internet. And it is not "just" real-time 8K VVC encoding, it also includes other cutting-edge features such us: motion-compensated temporal filtering, perceptually-optimized encoding, screen content coding, content-aware encoding, and live adaptive bitrate streaming. Behind PR announcements there is always a (more important) human story. This one started a few years ago when we met with our partners at Intel (Ravi Velhal) and started a collaboration around #8K video coding and streaming technologies. This partnership has evolved and survived over generations of video coding standards and processor architectures, but it is maintained by trust, technological craftmanship, and having fun doing all this. You can find more technical details in our joint announcement. Or come and visit us at IBC in Amsterdam: Booth 1.B32

META LEADS IN H100 DEPLOYMENTS Meta is making leaps with its AI infrastructure. Mark Zuckerberg recently shared that by the end of this year, they will have about 350k H100s including other GPUs that'll be around 600k H100 equivalents of compute. This makes Meta likely one of the largest GPU cluster operators in the world. Their investments in AI Research SuperCluster from 2 years ago are expanding further. With world-class compute infrastructure, Meta is well-positioned to support Reels and other Reels-sized AI services expected to emerge. #aihardware #NVIDIA #GPU #Semicondutors

Please to announce and present our new live 8K/60FPS/HDR #VVC/H266 Encoder at #ibc2023 here in #amsterdam this week. During #NAB2023 in Vegas we launched Intel 13th gen Core based 8K decoder player. Versatile Video Codec or H.266 standards based pioneering 8K live implementation brings unparalleled immersiveness and stunning 8K live with much reduced bitrate while preserving broadcast grade quality. Very important aspect to preserve creative intent while we pursue democratization of 8K for wider acceptance from 8K file based OTT as well as live performance and sports broadercasting. Having started my Intel career developing H.264 in 1999, later H.265 and now H.266, this is cross industry relentless pursuit of excellence over the last three decades to drive content ecosystem forward. Powered by Intel’s 4th Gen #xeon Scalable processor based server, real time 8K #VVC software encoder leverages AVX-512/VNNI DL boost advanced CPU instruction set level #AI technologies for perceptual optimized, content aware encoding and live adaptive bit rate streaming with low latency. Thank You Mauricio Alvarez-Mesa and #SpinDigital team for pursuing innovation together. Please come visit us at IBC Hall1.B32 to experience it yourself. 8K VVC Live IBC technical paper: https://lnkd.in/gkeWTSxF Press release : https://lnkd.in/gJ6pdWpM #SeeingIsBelieving #8k #8KLive #VVC #iAmIntel

Yikes! The Zuck is spending $10.5 Billion on Nvidia processors to make his AI work! That is a lot of cabbage!! Gonna take a mind-boggling amount of energy to run it too! And they said the disc had an environmental impact? <<The 350,000 number is staggering, and it’ll also cost Meta a small fortune to acquire. Each H100 can cost around $30,000, meaning Zuckerberg’s company needs to pay an estimated $10.5 billion just to buy the computing power, not to mention paying all the electricity costs.>> https://bit.ly/3vXiyJK

Llama 3 is Meta's latest LLM release. The difference with some of its competitors is that it's fully open source. That means there is more public data available, including the infrastructure that was used for training its 70B parameter model (just an average size, these days). It took a cluster of over 24,000 GPUs. I did some back of the envelope calculation, and that ends up being just about 7.2MW of power, assuming about 300W per GPU (Nvidia A100). If we assume a 5 week training cycle, that adds up to an impressive 6 GWh of electricity ⚡ , not including the power required to cool the cluster. 😲 Next step for Meta: Gen AI. To achieve their goal, they are planning to run a cluster of 350k GPUs!...Will this datacenter use 100% clean energy? Let's hope so. https://lnkd.in/gQJPaduh

New meta.ai image generation updates as you type. By the end of 2024, Meta plans to grow their infrastructure build-out to 350,000 NVIDIA H100 GPUs with compute power equivalent to nearly 600,000 H100s. I imagine these will likely be used mostly for training, but also to help with the large-scale deployment such as this web interface competitor to Google, Microsoft, and OpenAI. #generativeAI #meta #llama3

🌟 Exciting News! 🌟 Delve into the competitive landscape of the Big 3 in market capitalization: NVIDIA, Microsoft, and Apple in my latest blog post! Explore how these tech giants are reshaping the tech industry and the implications for consumers like you. 🌐 Read the full blog here: https://lnkd.in/d6qb93ux Discover: - NVIDIA’s AI and GPU innovations - Microsoft’s cloud and software revolution - Apple’s groundbreaking consumer technology From enriched gaming graphics to state-of-the-art cloud solutions and seamless device integration, these companies are steering the course of technology's future. Check out the blog and share your thoughts! 🚀✨ #TechInnovation #NVIDIA #Microsoft #Apple #MarketCapitalization #CloudComputing #ArtificialIntelligence #ConsumerTech #NewBlog #PacConsulting #StayInformed #TechTrends

🚀 Exciting News! Listen2it is Now NVIDIA Cloud Validated! 🎉 We're thrilled to announce a major milestone in our journey towards delivering cutting-edge audio technology to our users. Listen2it has officially achieved NVIDIA Cloud Validation, cementing our commitment to excellence and performance in the realm of audio processing. What does this mean for you, our valued users? 1️⃣ Unparalleled Performance: With NVIDIA Cloud Validation, Listen2it is now equipped to provide you with lightning-fast and highly efficient text-to-audio processing. Whether you're converting text documents into speech, generating natural-sounding voiceovers, or utilizing Listen2it for any text-to-audio tasks, you can expect an exceptional level of performance. 2️⃣ Enhanced Reliability: While we have always prioritized reliability, this validation from NVIDIA adds an extra layer of confidence. You can rely on Listen2it to consistently deliver high-quality text-to-audio conversions with precision. 3️⃣ Future-Proof Technology: By harnessing the power of NVIDIA's GPU acceleration, we're future-proofing Listen2it in the domain of text-to-audio conversion. This ensures that Listen2it continues to lead the way in this evolving technology field. 4️⃣ Better User Experience: Above all, this validation is designed to elevate your user experience. You will now enjoy smoother, faster, and more efficient text-to-audio conversions, making your tasks easier and more enjoyable. Thank you for being a part of the Listen2it community. Your future audio adventures just got even more extraordinary! 👂#Let'sListen2It🌟 #NVIDIAValidation #AudioExcellence