NVIDIA founder and CEO Jensen Huang delivered his GTC keynote on Sept 20 at 8 a.m. Pacific time, finally announcing the Ada Lovelace RTX 40 Series. Easily one of the most hotly debated and much discussed graphics card launches of all time, the RTX 40 Series is a masterclass in innovative technology, and the product of years of tireless work in pursuit of the most powerful AI-driven GPUs the world has ever seen.
In this article
MSI were one of the first to raise their head above the parapet, unveiling a line of 40 Series GPUs that are - let's be honest here - jaw-droppingly gorgeous.
Alongside MSI, GPU giants Gigabyte also barged into the fray with their own style of AORUS RTX 40 Series marketing, with a website that feels as much like a promo for a trendy nightclub than a GPU product page. And I mean that in a good way.
These colossal GPU vendors join the likes of ASUS, PNY and Zotac, and the other usual suspects in an epic launch that we all waited for - and it looks to have delivered the spectacle we dared dream of.
Yes, the wait is finally over, and the salivating masses can finally cross-reference the revealed specifications with the many leaks and spurious spec sheets we have seen over the last year. More than that, though, we also have a release date for the RTX 4080 and RTX 4090 (and, thankfully, it's not that far away).
RTX 40 Series Revealed
Jensen Huang, the founder, president and CEO of NVIDIA is certainly great to listen to as he delivers keynotes, with technical and engineering specifications rolling off his tongue with ease. Not surprising, considering he studied electrical engineering for both undergraduate and master’s degrees, and has pedigree in microprocessors with LSI Logic, and Advanced Micro Devices, Inc. - also known as the CPU titans, AMD.
The keynote kicked off with an eye candy demo - RacerX - a CGI video capturing beautifully rendered RC cars racing around a Micro Machines style gymkhana, all rendered on one ADA Lovelace GPU. The keynote also revealed the many leaps in AI and architecture design that goes into the 3rd-Gen TSMC 4N graphics cards, and some impressive claims from Huang and NVIDIA as to performance over previous generations.
NVIDIA RTX 40 Series Announcement Recap
- Ada Lovelace - or simply Lovelace - is the codename for 3rd generation RTX GPUs
- RTX 40 Series' are built on TSMC 4N process and the chip contains over 76 billion transistors.
- Lovelace brings with it new tech called SER (Shader Execution Reordering) which improves execution efficiency by rearranging and logically scheduling shading workloads on the fly, utilising the GPU resources holistically better. Lovelace and SER can improve Ray Tracing performance by up to 3x - increasing framerates in games by up to 25%.
- Ada Optical Flow Accelerator enables the all new DLSS 3 to accurately predict movement in a game, using the neural network to increase framerates without sacrificing any image quality, and with up to 4x increase in performance. DLSS 3 supports and augments CPU-limited and GPU-limited games efficiently without a system performance hit.
- Dual NVENC encoders reduce video export times by up to 50%, featuring AV1 support, similar to Intel's ARC support. AOMedia Video 1 (AV1) is an open, royalty-free video coding format that is predominantly used for streaming. NVENC AV1 is being adopted by Blackmagic Design (DaVinci Resolve etc.), OBS and Discord, with more being added to this list.
- NVIDIA Studio is back in the next generation of GPUs, allowing creators to leverage the incredible power of the Ada Lovelace AI, and deliver "unmatched performance in 3D rendering, video editing, and graphic design" through RTX accelerations.
- Portal RTX is a remastering of Portal by Valve, reimagined using NVIDIA Omniverse.
- RTX Remix is an incredible leap for game mods and offers a suite of tools and productivity for modders. RTX Remix will release soon after the Ada Lovelace launch and will enable any modder to remaster their chosen games with RTX and Ray Tracing, and take advantage of the latest visual technology in classic games. Some titles are already supporting RTX Reflex, such as Loopmancer and Scathe.
- Titles soon available for NVIDIA Reflex include:
- Call of Duty: Modern Warfare II
- Cyberpunk 2077
- Overwatch 2
- SUPER PEOPLE
- Warhammer 40,000: Darktide
- Ada Lovelace delivers 2x faster performance in rasterized games and up to 4x faster for Ray Tracing supported games.
The most incredible features of Lovelace technology is NVIDIA Omniverse™, an "easily extensible platform for 3D design collaboration and scalable multi-GPU, real-time, true-to-reality simulation". Omniverse allows design and development to coexist in harmony, offering a platform for individuals to work together as teams, and enable untethered creative possibilities. Omniverse delivers a ubiquitous platform that promotes incredible efficiencies for 3D creators, designers, developers and - at scale - enterprises.
I'm a gamer. Why should I care about Omniverse?
All gamers should care about NVIDIA's commitment to Omniverse. With Omniverse, NVIDIA has clearly seen the future of game design and development, and that lies in seamless, globalised collaboration of professional teams. Imagine a game studio developing a game in Blender, Reallusion Character Creator and Unreal Engine and co-existing in the same development environment, despite being on three different continents.
Imagine smaller studios leveraging lower costs of offshore developers, keeping full oversight and control in-house, with a small staff of lead game designers and devs. This "develop anywhere" approach now fits comfortably with the current economy and remote working ethos that many companies have already adopted. With NVIDIA Omniverse, the transfer of assets is no longer required, and the testing, debugging and ongoing development can take place in real time - in one place, from anywhere. For gamers, this means shorter development timelines and much better beta & alpha testing pipelines, mitigating launch disasters and blunders we've seen in recent years.
TL;DR - In short, you get beautiful games faster, with a centralised QA process that could not exist previously. This is the future of game development, and you read it here first (probably).
Omniverse Application Support
Compatible with numerous applications and creative suites, Omniverse offers "true simultaneous multi-app and multi-user collaboration" and provides order to complex workflows, no matter where in the world team members are. This is made possible by Pixar’s Universal Scene Description (USD), an extensible and open-source 3D "scene description and file format" that is designed for content creation, promoting collaboration and easy interchange between different tools and applications.
There have been early adopters and champions of Omniverse already, with the following projects: -
Omniverse is on the very thin line between innovation and the Uncanny Valley, which NVIDIA is navigating extremely well, bringing the real world and their version of the Metaverse closer together. What Mr. Zuckerberg hoped his company's VR Metaverse would be has been delivered by NVIDIA's technologies and advancements with Lovelace. Omniverse is enabling teams to create a replica of the real world inside Omniverse, and design, test, develop and deploy to their hearts content at a fraction of the real world cost.
RTX 40 Series release date and MSRP
- GeForce RTX 4090 (MSRP £1679) UK Release: October 12th 2022 @ 2pm BST
- GeForce RTX 4080 16GB (MSRP £1269) UK Release: November 2022
- GeForce RTX 4080 12GB (MSRP £949) UK Release: November 2022
These prices will - naturally - be subject to change, availability and stock levels. The RTX 4080 has been considered as a "paper launch", and is expected to sell stocks out very quickly, with a re-release happening in the following months.
RTX 40 Series Specifications
| ||GeForce RTX 4090 ||GeForce RTX 4080 (16GB) ||GeForce RTX 4080 (12GB) |
|NVIDIA CUDA Cores ||16384 ||9728 ||7680 |
|Boost Clock (GHz) ||2.52 ||2.51 ||2.61 |
|Memory Size ||24 GB ||16 GB ||12 GB |
|Memory Type ||GDDR6X ||GDDR6X ||GDDR6X |
To take advantage of the technology in the new NVIDIA GPUs, the new breed of PSU - ATX 3.0 - has been developed. Offering many efficiencies over the previous generation, and delivering the type of power required for RTX 40 Series graphics cards. ATX 3.0 power supplies will become available in October from leading brands such as ASUS, FSP, Cooler Master, Gigabyte, MSI, Silverstone, and ThermalTake.
RTX 40 Series vs 30 Series
Available October 12th, 2022, and starting at £1679, the GeForce RTX 4090 will be the ultimate GeForce GPU, delivering unrivalled performance at the highest level for gamers and creators. Incredibly, the RTX 4090 has bucked leaks and rumours, being up to 2x as fast as previous generation (RTX 30 Series) GPUs, but requiring the same power as the GeForce RTX 3090 Ti.
RTX 4080 vs RTX 3080 Ti
RTX 4090 vs RTX 3090 Ti
Should you buy the RTX 4080 or RTX 4090?
Make no mistake, you are bearing witness to the next generation in gaming and creative possibilities. The RTX 40 Series represents the most significant leap forward in visual performance and raw power since the first generation of RTX in 2018. But, more than that, the performance gains mean so much more to the wider world of RTX users.
Developers, data scientists and gamers alike have access to an Enterprise level of power and processing that will catapult the human race into the next era of entertainment, innovation and creativity.
In our previous article, we asked if you should wait for the RTX 40 Series. We said yes, depending on your budget. That was before we knew how much (or how little) NVIDIA were selling the new RTX 40 Series for.
The question now is, should you buy one? The answer is a categorical yes. If you are building or upgrading a PC, or plan on buying a pre-built gaming PC that features a GPU from the RTX 40 Series, you are going to be among the first experience history being made in games and game development. Congratulations.