NVIDIA Keynote Live Blog (11 p.m. ET / 03:00 UTC)
10:58 p.m. EDT – We’re back for day two of our live blog coverage on Computex
10:59 p.m. EDT – After AMD’s big processor-centric showcase, NVIDIA is on deck tonight to deliver what promises to be a varied keynote touching on several aspects of NVIDIA’s business
11:00 p.m. EDT – The untitled keynote is scheduled to last an hour and features a who’s who of NVIDIA VPs, including Ian Buck (VP Accelerated Computing), Michael Kagan (CTO), and Jeff Fisher (SVP GeForce)
11:00 p.m. EDT – So expect the keynote to cover everything from data centers to gaming
11:01 p.m. EDT – Here we go again
11:02 p.m. EDT – Starting with a quick promo video of NVIDIA and the growth of their GPU business, from gaming to AI and more
11:03 p.m. EDT – First, Ian Buck
11:03 p.m. EDT – From AI
11:04 p.m. EDT – It requires re-imagining the data center itself
11:04 p.m. EDT – AI also opens up a new market for digital twins
11:04 p.m. EDT – A potential market of 150 billion dollars
11:04 p.m. EDT – And an additional $100 billion market for cloud game streaming
11:05 p.m. EDT – Everything NVIDIA partners can leverage using NVIDIA hardware
11:05 p.m. EDT – These tasks consume a lot of computing power – so energy efficiency is a big issue
11:06 p.m. EDT – Of course, NVIDIA is also happy to claim that a lot of power can be saved by switching from using CPUs for compute tasks to GPUs
11:07 p.m. EDT – And NVIDIA has a sizable collection of other hardware, including Grace processors and ex-Mellanox networking hardware
11:07 p.m. EDT – Now let’s recap the H100 Accelerator
11:08 p.m. EDT – NVLink 4, Multi-Instance GPU, Confidential Computing, Transformer Accelerator, etc.
11:08 p.m. EDT – And NVLink switches to form larger groups of computers
11:10 p.m. EDT – And thank you to the Taiwanese partners of NVIDIA for their contribution to the construction of these new servers
11:10 p.m. EDT – Current NVIDIA roadmap
11:11 p.m. EDT – Two-year rhythm on the material. Interleaving between x86 and Arm
11:11 p.m. EDT – Now let’s go to Grace
11:11 p.m. EDT – Grace is on track to ship next year
11:11 p.m. EDT – Grace and Grace Hopper
11:12 p.m. EDT – For CPU and CPU+GPU workloads respectively
11:12 p.m. EDT – The Grace Superchip (dual Graces) has 144 cores. It has 1 TB of LPDDR5X memory and consumes 500 Watts
11:13 p.m. EDT – And reiterating NVIDIA’s plans to put NVLink in all of their silicon, so it can be linked to other NVIDIA and third-party hardware with a consistent high-speed cache interconnect
11:14 p.m. EDT – NVIDIA has developed a lot of reference systems over the years, and they are now producing a new generation based on Grace
11:15 p.m. EDT – New CGX, OVX and GFX reference designs using Grace and/or Grace Hopper
11:15 p.m. EDT – Announcing the HGX Grace and HGX Grace Hopper systems
11:15 p.m. EDT – Both are specifically designed for 2U OEM chassis
11:46 p.m. EDT – Can leverage their existing system architectures
11:46 p.m. EDT – Grace systems will begin shipping in the first half of 2023
11:17 p.m. EDT – Now let’s move on to the Bluefield-3 NPUs
11:18 p.m. EDT – NVIDIA relies heavily on Bluefield and DPUs are becoming essential for data centers
11:18 p.m. EDT – It is silicon heavy and will be able to take advantage of NVIDIA’s GPUs for processing
11:19 p.m. EDT – Or as NVIDIA likes to put it, this networking gear will be the backbone of AI systems
11:19 p.m. EDT – Now let’s move on to NVIDIA’s enterprise software stack
11:20 p.m. EDT – As a reminder, NVIDIA has more software engineers than hardware engineers. So, at the enterprise level, software is even more of a differentiator for NV than hardware.
11:20 p.m. EDT – The next wave of AI: robotics
11:21 p.m. EDT – “We see a clear demand for automation”
11:22 p.m. EDT – Summary of NVIDIA’s Isaac robotics platform
11:23 p.m. EDT – And how Isaac interfaces with omniverse for simulation and robotics training
11:24 p.m. EDT – NVIDIA’s pre-trained models available on NGC are a big part of this activation.
11:25 p.m. EDT – Which is another example of where NVIDIA uses software to grow its ecosystem
11:26 p.m. EDT – Isaac Sim using Omniverse
11:26 p.m. EDT – NVIDIA wants to bridge the gap between sim and real
11:26 p.m. EDT – Announcement of the release of Isaac Sim 2022.1
11:27 p.m. EDT – Now rolls an Isaac Sim demo
11:29 p.m. EDT – Now let’s move on to NVIDIA hardware for robotics: Jetson
11:31 p.m. EDT – NVIDIA’s Orin SoC reportedly performs well in MLPerf benchmarks
11:31 p.m. EDT – The Jetson AGX Orin Development Kit is available now
11:32 p.m. EDT – Production systems with Jetson Orin are available now, with many being announced at Computex this week
11:33 p.m. EDT – Isaac Nova Orin. State-of-the-art architecture for autonomous mobile robots
11:34 p.m. EDT – One more thing: drive Hyperion
11:35 p.m. EDT – Hyperion version 8 will start shipping in 2024
11:35 p.m. EDT – And Hyperion 9 is in development for cars shipping from 2026
11:35 p.m. EDT – Processes twice as much sensor data as Hyperion 8
11:36 p.m. EDT – And NVIDIA signed additional vendor-partners for Hyperion
11:36 p.m. EDT – And now let’s get to the games with Jeff Fisher
11:37 p.m. EDT – “Taiwan is the cradle of the PC ecosystem”
11:37 p.m. EDT – Recap of NVIDIA’s current hardware stack, including GeForce RTX series video cards, Max-Q laptops, and G-Sync monitors
11:38 p.m. EDT – PC gaming hardware is expected to be a $67 billion market this year, with double-digit growth over the next few years
11:38 p.m. EDT – Jeff thanks NVIDIA partners for their products
11:39 p.m. EDT -“[Ampere] is the fastest GPU in the world”
11:39 p.m. EDT – Recap of Ampere family features such as RT cores, DLSS and NVENC video encoder
11:40 p.m. EDT – Ongoing deployment of a video on NVIDIA’s driver development process
11:41 p.m. EDT – This appears to be the same Game Ready (or similar) dev video they posted a few weeks ago
11:41 p.m. EDT – “Our mission with drivers is to be invisible”
11:42 p.m. EDT – There are now over 250 RTX games and apps
11:42 p.m. EDT – Hitman 3 gets an update with ray tracing and DLSS
11:43 p.m. EDT – Also announces that F1 22, which will launch in July, will also have RT and DLSS
11:44 p.m. EDT – Now let’s move on to NVIDIA Reflex
11:44 p.m. EDT – “Low system latency helps all players”
11:44 p.m. EDT – NVIDIA recently conducted a study on the impact of system latency on aiming accuracy
11:44 p.m. EDT – Kovaak’s System Latency Challenge
11:45 p.m. EDT – Launch another promotional video, this time on Reflex
11:45 p.m. EDT – Talk about reducing system latency with Reflex
11:46 p.m. EDT – Reflex is even a plugin in Unreal Engine these days
11:46 p.m. EDT – Currently up to 38 games and 22 monitors
11:47 p.m. EDT – Introducing a new Reflex compatible screen
11:47 p.m. EDT – A new Asus 500Hz 1080p 24-inch monitor
11:47 p.m. EDT – TN panel
11:47 p.m. EDT – Includes reflex analyzer
11:48 p.m. EDT – Now on to GeForce laptops with Max-Q
11:48 p.m. EDT – Summary of Max-Q feature set
11:49 p.m. EDT – New gaming laptops at Computex from MSI, Asus, Gigabyte and others
11:49 p.m. EDT – Now let’s go to NVIDIA Studio
11:49 p.m. EDT – The creative-focused NVIDIA platform
11:50 p.m. EDT – Now over 200 Studio accelerated apps
11:51 p.m. EDT – And how Omniverse will also help game developers and content creators
11:51 p.m. EDT – 10x increase in Omniverse downloads
11:51 p.m. EDT – Omniverse is also updated with new features, such as Omniverse Cloud and Audio2Face
11:53 p.m. EDT – Now let’s recap the gaming segment of this talk and the state of NVIDIA’s gaming ecosystem
11:53 p.m. EDT – And Jeff thanks the NVIDIA partners one last time
11:54 p.m. EDT – And by signing with Toy Jensen, it’s over. Thanks for join us.
Comments are closed.