After a lot of leaks,
speculation, and teasing announcements that appeared on us every now and then
over the past period, today, Tuesday, the graphics giant "NVIDIA"
finally revealed the new GeForce RTX 4000 series of graphics cards, nearly two
years after the RTX 3000 series. About the new generation During the GTC 2022
event that was broadcast online a while ago, Nvidia showed us more details
about the RTX 40 series, which currently consists of two cards, and represents
the greatest generational leap in the company's history in the words of the
company's CEO, Mr. Jensen-Hwang.
As expected, this
generation is also witnessing the first appearance of the Ada Lovelace
architecture, which succeeds Ampere and is the first architecture in RTX cards
with a precision of 4 nm from TSMC. Without any exaggeration at all, there are
powerful updates and technologies like never before, breaking all the current
performance barriers and enough to take the level of graphic performance in
gaming to a whole new “revolutionary” level, everything you know about the
performance of any card in the RTX 30 generation has doubled by up to up to 4
times per new generation. There is a lot to talk about, so let us explain to
you below all the important information and details that were announced during
the conference.
Launching Ada Lovelace Architecture
The GeForce RTX 40
series is based on a new architecture that Nvidia named Ada Lovelace after the
world's first computer programmer. The architecture will provide a huge leap in
performance, efficiency, and capabilities compared to previous generation GPUs.
Initially built with 4nm TSMC's custom manufacturing precision, this
architecture contains up to 76 billion transistors (compared to the previous
generation's 28 billion) and up to 18,000 CUDA cores — enough to make it the world's
most powerful graphical architecture, with no competition! We're talking about
more than twice the performance with almost the same power consumption compared
to the Ampere architecture, with the comfortable ability to reach a level that
exceeds the minimum when overclocking with increased power consumption.
Ada Lovelace
architecture introduces a lot of completely new and improved technologies such
as Shader Execution Reordering technology that contributes to improving the
performance of ray tracing by better reorganizing the shading work within
games, where previously ray tracing technology required calculating the effect
of millions of rays that collided with many Different elements are located
throughout the in-game scene, creating a series of disparate and inefficient
shadowing workloads, but thanks to SER technology, Ada Lovelace will be able to
deliver up to 3x better Ray-Tracing performance and better gaming performance
at up to frame rates 25%.
As for technologies
improved with the new architecture, NVIDIA has announced the third generation
of ray tracing cores dedicated to handling performance-intensive ray tracing
workloads in games. Now the 3rd generation ray tracing cores in the Ada
architecture delivers up to 191 teraflops of performance - more than twice the
performance of ray tracing cores compared to the previous generation (78
teraflops) enabling the delivery of immersive ray tracing worlds, incredibly
smooth and fast.
The Ada Lovelace
architecture carries 4th generation Tensor Cores, these cores are simply the
part responsible for accelerating and processing artificial intelligence
technologies including NVIDIA DLSS and NVIDIA Broadcast. The fourth generation
of these cores translates into an impressive speed in processing any AI-based
tasks. With the all-new Tensor engine, you'll get a performance of up to 1.32
petaflops, increasing productivity up to 5 times compared to the previous
generation.
NVIDIA is also trying to
provide new and better ways to deliver immersive experiences at high-performance
frame rates. So, keen to harness the power of the massive Ada architecture and
take advantage of new features that greatly speed up advanced ray-tracing
workloads and deliver faster frame rates in your favorite games, the company
has released the third version of NVIDIA DLSS technology.
The GeForce RTX 40
Series supports DLSS 3 technology, which generates brand new, high-quality
frames, rather than pixels. DLSS 3 builds on the company's leadership in
AI-accelerated Ultra HD technologies to deliver the superb image quality and up
to 4x frame rates over native resolution, plus responsive responsiveness that
defines the overall gaming experience. That is, the third version of DLSS takes
advantage of the fourth generation of Tensor cores and the new Ada Optical Flow
accelerator, which allows DLSS technology to predict what will happen in the
next moment within the game and then speeds up the processing time of the frame
itself to increase the FPS rate significantly.
.webp "DLSS 3 technology")
The creativity doesn't
stop here, as the AI capabilities of the Ada architecture will allow
processing two game frames at the same time, one generated by DLSS and one
created by DLSS super-resolution frames technology integrated with DLSS 3 that
renders seven-eighths of the displayed pixels using AI to ensure Ensure that
every frame in the game is displayed correctly without errors or interruptions
in the scene, and even enhances the game performance by up to 4 times compared
to running the same game without DLSS technology enabled. You can find out more
details about how to do this in NVIDIA's article on DLSS 3.
It is also worth noting
that DLSS version 3 may end the era of CPU-intensive games such as Microsoft
Flight Simulator or end the "bottleneck", which limits the use of
faster external graphics and high-resolution technologies. Full on the GPU
processor, to bypass any obstacles caused by the central processor without
affecting the performance of frame rates. In other words, this version makes
the new NVIDIA cards able to provide comfortable performance in games that
require intensive effort from the processor, and playing on 4K resolution will
be The first choice for many players.
The revolutionary power
of DLSS 3 is a boon for game developers who want to express their artistic
vision. The technology has been supported in the world's most famous game
engines, such as Unity and Unreal Engine, and DLSS 3 has already received
support from many of the world's leading game developers, with more than 35
games and apps now announcing their support such as Cyberpunk 2077, Microsoft
Flight Simulator, HITMAN 3 and Hogwarts Legacy Now you can see the differences
for yourself by watching this video highlighting the graphic difference
in Cyberpunk 2077 before and after DLSS 3.
However, there are a lot
of other improvements in the Ada Lovelace architecture that we won't be able to
mention in this thread such as NVIDIA Reflex technology that now offers as low
as 10ms latency in major game titles, and GeForce Experience can now record
games at up to 8K 60fps with HDR enabled. Dual AV1 Encoders technology
specifically geared towards content creators now provides up to twice as low
rendering time as the previous generation. In general, we are facing an
architecture that is much stronger than we expected and provides the best
possible performance in a graphics card, whether for content makers or gamers.
New RTX 4090 and RTX 4080 cards revealed
The ultra-efficient Ada
Lovelace architecture delivers a quantum leap in the performance and
capabilities of any supported GPU, which translates to GeForce RTX 40 series
cards that will be able to provide users with a complete next-generation ray
tracing gaming experience and complete their creative work in no time.
Well, for the same cards
that Nvidia revealed today, I started with the flagship card in the series so
far, the RTX 4090, which includes 16,384 CUDA cores at a boosted speed of 2.52
GHz. It also has a massive 24GB GDDR6X memory space that will work with the new
Ada Lovelace architecture, and the power consumption is 450W. These numbers
translate to the RTX 4090 offering up to 4 times better performance than the
RTX 3090 Ti (the most powerful card in the RTX30 series) by up to 4 times. In
gaming, this card will allow any game, whether currently or in the coming years,
to run at the highest graphics settings at 4K resolutions at high frame rates.
And the ability to fully play ray tracing games with the highest levels of
accuracy, realism, and immersion.
As you can see in the
graph, the GeForce RTX 4090 in Microsoft Flight Simulator outperformed the
GeForce RTX 3090 Ti by nearly two times, in Warhammer 40,000: Darktide by more
than two times, and in Cyberpunk 2077 (updated to support the new generation of
ray tracing). ) more than 4 times. Of course, this jump in performance is
amazing thanks, all credit, to the new Ada architecture and its adoption of
TSMC's 4N manufacturing technology.
Next, NVIDIA introduced
the GeForce RTX 4080 graphics card, which will be available in two versions,
the first includes 16GB of graphics memory supported by 9728 CUDA cores at a
speed of up to 2.51GHz, making it twice as fast as the GeForce 3080 Ti card
with 10% lower power consumption Approximately as the card needs 320W of power.
This version is aimed more at content creators and creators, as the VRAM
capacity helps when viewing larger scenes, editing RAW 12K video, and working
with multiple applications at the same time.
As for the second
version, or we can say the smaller version, it comes with a memory capacity of
12 GB supported by 7680 CUDA cores with a speed starting from 2.31 and up to
2.61 GHz, and power consumption of 285 watts. Although it seems the weakest in
the series, according to the Benchmarks that Nvidia presented at the
conference, this card provides better performance than the GeForce RTX 3090 Ti
by two times in some games, and of course, is characterized by being supportive
of all the new innovations in the Ada architecture.
As for pricing and
availability, the RTX 4090 will be available on October 12 starting at $1,599,
and the GeForce RTX 4080 16GB and 12GB versions will be available in November,
starting at $1,199 and $899, respectively. And let us remind you that these are
the prices of Founders Edition cards (the raw version of the RTX 40 cards) that
are designed and manufactured by Nvidia, while versions made by Nvidia partners
such as MSI and ASUS will cost more, of course. As for the Founders Edition
cards, they come with the same design as the Dual-Axial Flow-Through used in
the previous generation RTX 30, but with internal improvements to enhance heat
dissipation efficiencies, such as increasing fan sizes and fins by 10%.
It is worth noting that
all Founders Edition cards from the GeForce RTX 40 series adopt the 16-pin
system in the fifth generation PCIe cables, thus supporting ATX 3.0 power
supplies. Kaplin as in the previous generation cards. However, you may not have
to update your power supply if you're using PowerPlay to power an RTX 30 card
as Nvidia says it will include an adapter cable with the new cards.
Introducing Portal with RTX Thanks to NVIDIA RTX Remix
Portal is one of the
highest-rated games ever, and NVIDIA, represented by NVIDIA Lightspeed Studios,
has considered making a new, improved version of Portal in collaboration with
Valve to support ray tracing technologies. Advanced graphics features such as
full ray tracing and DLSS 3 give the game a stunning new look and feel as every
aspect of the game's graphics has been upgraded and improved with attention to
detail. Portal will be released with RTX for free next November, just in time
to celebrate Portal's 15th anniversary!
The enhanced version of
the game was developed using the new NVIDIA RTX Remix mod platform built on the
NVIDIA Omniverse which was also revealed during today's GeForce Beyond conference.
Modding is new in-game content that is created by players rather than
developers. The NVIDIA RTX Remix Modding platform gives you all the tools to
easily create a mod of any of your classic games powered by RTX technologies!
The platform learns the basic elements of the game and then improves the
graphics quality automatically using artificial intelligence tools.
This is the most
important thing that was announced during the NVIDIA conference today, if you
need to know more details you can read this detailed report from NVIDIA.