Tuesday, April 23rd 2024

NVIDIA DLSS 3 Coming To Gray Zone Warfare, Remnant II: The Forgotten Kingdom DLC and More Games

Over 500 games and applications feature RTX technologies, and barely a week goes by without an incredible new game integrating NVIDIA DLSS, NVIDIA Reflex and advanced ray-traced effects to deliver the definitive PC experience for GeForce RTX players. Gray Zone Warfare, a highly anticipated new tactical shooter, is coming soon with DLSS 3, and the DLSS 3-enhanced Remnant 2 receives a new DLC expansion this week. Plus, Manor Lords, the #1 most wishlisted title on Steam, launches with support for DLSS 2, the latest Mortal Kombat 1 content has dropped and can be enjoyed at a maxed out 60 FPS with DLSS 2, and Oddsparks: An Automation Adventure arrives with day-one DLSS 2 support. Learn more about each announcement below.

Gray Zone Warfare Coming Soon With Support For DLSS 3 & Reflex
MADFINGER Games' Gray Zone Warfare is an immersive tactical FPS with a maximum focus on realism, and is one of Steam's 20 most wishlisted games. Following a mysterious event that took place on an island in Southeast Asia, the entire area has been put under international quarantine. With almost all the civilian population evacuated by the United Nations, three PMCs are tasked by their enigmatic clients to explore the region, uncover its secrets, and bring back anything of value.
Join one of the PMCs and navigate a vast open setting alone or in a squad. Adapt to your surroundings, use tactics to your advantage, and fight against other human operators and AI-controlled enemies in a persistent world that keeps on living even when you're not playing.

Coming soon, Gray Zone Warfare will utilize the latest technologies to deliver an enhanced experience for GeForce RTX users. Starting day one, DLSS 3 will accelerate performance, Reflex will reduce system latency by up to 50% for more responsive gameplay, and hardware-accelerated, ray-traced Unreal Engine 5 Lumen lighting will increase the game's graphical fidelity even further.


Remnant II: The Forgotten Kingdom DLC Launches Today - Accelerate Performance With DLSS 3 & Reflex
Remnant II from Gunfire Games is the sequel to the best-selling game Remnant: From the Ashes that pits survivors of humanity against new deadly creatures and god-like bosses across terrifying worlds. Play solo or co-op with two other friends to explore the depths of the unknown to stop an evil from destroying reality itself.

The second of three announced Remnant II DLC expansions launches later today, further expanding this massive game that you can sink hundreds of hours into. The new expansion, The Forgotten Kingdom, takes players to a new part of Yaesha, where they'll unravel the mystery of a lost tribe in a brand-new storyline. Discover the secrets of the new zone, unlock the new Invoker Archetype, acquire new armor, items and weapons, and face off against new bosses and enemies.

GeForce RTX gamers can tailor their Remnant II experience with a full gamut of DLSS modes. DLSS 2 enables all GeForce RTX users to accelerate frame rates, DLSS 3 gives GeForce RTX 40 Series users the power of frame rate multiplication for the fastest performance possible, and DLAA uses a native resolution image to maximize image quality, rather than boosting performance. You can even combine DLAA and DLSS 3 Frame Generation for excellent image quality, and performance.

Using DLSS 3, GeForce RTX 40 Series gamers can multiply Remnant II frame rates by an average of 3.5X, enabling owners of the GeForce RTX 4070 and faster cards to enjoy the game at max settings at 4K, at over 60 FPS.

Manor Lords, The #1 Wishlisted Steam Game, Launches With DLSS 2 On April 26th
Created and coded by a solo developer, with publishing support from Hooded Horse, Manor Lords is a medieval strategy game featuring in-depth city building, large-scale tactical battles, and complex economic and social simulations. Grow your starting village into a bustling city, manage resources and production chains, and expand your lands through conquest in Steam's most wishlisted game, followed by over 3 million users.

When Manor Lords launches into Early Access on April 26th, GeForce RTX gamers can instantly increase frame rates thanks to day-one support for DLSS 2. At 4K, a 47% average boost to performance enables owners of the GeForce RTX 4070, and faster cards, to exceed 60 FPS, and enables owners of faster GPUs to increase performance.

At 2560x1440 and 1920x1080, DLSS 2 allows all GeForce RTX 40 Series owners to comfortably surpass 60 FPS with max settings, enabling you to watch over your bustling settlement with the highest possible frame rates, and zoom in to battles, issuing orders in real-time without slowdown.
For the absolute best experience, be sure to download and install our Manor Lords GeForce Game Ready Driver before you play.


Mortal Kombat 1 Season 5: Storms Available Now - Max Out The Game With DLSS 2
Liu Kang has restarted history, crafting a New Era reflecting his vision of peace. Meanwhile Mileena, twin sister of Kitana and rightful heir to Outworld's throne, has been infected by a dreaded and lethal disease. Lin Kuei warriors and brothers Scorpion and Sub-Zero fight for the future of the clan, while Earthrealm champions Raiden and Kung Lao fight for family and honor. Johnny Cage, who is more concerned with staying relevant in the eyes of his fans, fights for his own vanity. Fight your way through the new Universe of Mortal Kombat 1, ushering in a new era of the iconic franchise with a new fighting system, game modes, and new, spectacular fatalities.

Since the launch of Mortal Kombat 1, DLSS 2 has been available to help GeForce RTX gamers hit 60 frames per second, with every setting maxed out. Now, Season 5: Storms is available to play. This all-new season of content is themed around the arrival of Raiden as the main boss in Invasions mode. The new season features fresh challenges, story elements, boss battles, and rewards, including character skins - Gaffer Johnny Cage, Sudden Overload Scorpion, Voltaic Deity Liu Kang, Electric Soul Liu Kang, Voltaic Warrior General Shao, Charged Frost Sub-Zero, Magnetic Courage Kung Lao, Glistening Mage Rain, Tainted by Evil Raiden, and more - all earnable through Invasions and Kombat League.

Additionally, Ermac has now joined the Mortal Kombat 1 roster as the latest downloadable content character. Ermac is a collection of souls bound together by Quan Chi's dark magic that are intended to function as a group mind. But that spell is temporarily undone when Ermac is defeated by Mileena and the mind of Emperor Jerrod, deceased ruler of Outworld and one of the souls in Ermac's collection, takes control and rejoins the royal family, aiding Empress Mileena's reign. Eventually, the amalgam of souls reestablishes control, and Ermac retreats to Outworld's shadows. No longer Quan Chi's slave, nor bound to the royal family, he must find a future worth fighting for.

If you haven't played Mortal Kombat 1 before now, be sure to enable DLSS and crank everything to the max when you load in.


Oddsparks: An Automation Adventure Available April 24th With DLSS 2 & DLAA
Massive Miniteam and HandyGames' Oddsparks: An Automation Adventure is a single-player and co-op blend of automation, exploration and real-time strategy, tasking you with building production lines with the help of your odd and adorable Sparks. These Sparks will accompany you on your journeys across procedurally generated worlds filled with enemies, resources, and ancient secrets.

Launching into Early Access on April 24th, Oddsparks: An Automation Adventure will feature day-one support for DLSS 2, enabling all GeForce RTX gamers to accelerate performance when building their mega-sized workshops across Oddsparks' world. Alternatively, enable DLAA to take image quality to 11. DLAA is an AI-based anti-aliasing mode that utilizes the same technology developed for DLSS, but instead uses a native resolution image to maximize image quality, rather than boosting performance, for an even richer experience.

Source: NVIDIA
Add your own comment

17 Comments on NVIDIA DLSS 3 Coming To Gray Zone Warfare, Remnant II: The Forgotten Kingdom DLC and More Games

#1
dgianstefani
TPU Proofreader
Gray Zone looks like an actual Tarkov competitor with an actual good game engine.

But we'll see.
Posted on Reply
#2
cmguigamf
Announcing DLSS3 for the DLC of a game whose base version already had DLSS3 is bonkers.
Posted on Reply
#3
apoklyps3
cards still expensive as hell and no bundle to sweeten the deal in 2024.
last year they had bundles every two months or so
Posted on Reply
#4
Upgrayedd
Could someone explain why older RTX cards couldn't get a different version of DLSS FG? besides selling newer cards.
Instead of doubling why can't we just get 20% or 25%? Would the latency be too high? Too much ghosting? would older cards even be capable of that or would they be resolution limited?
Posted on Reply
#5
dgianstefani
TPU Proofreader
UpgrayeddCould someone explain why older RTX cards couldn't get a different version of DLSS FG? besides selling newer cards.
Instead of doubling why can't we just get 20% or 25%? Would the latency be too high? Too much ghosting? would older cards even be capable of that or would they be resolution limited?
No optical flow accelerator hardware unit
Posted on Reply
#6
evernessince
UpgrayeddCould someone explain why older RTX cards couldn't get a different version of DLSS FG? besides selling newer cards.
Instead of doubling why can't we just get 20% or 25%? Would the latency be too high? Too much ghosting? would older cards even be capable of that or would they be resolution limited?
It's 100% to sell cards and nothing else. Latency of cards prior to 4000 series would be worse with FG enabled but that can be fixed by including multiple quality levels where lower quality levels would result in lower latency. In fact that would be a nice option to have even for 4000 series owners as then users can choose a setting with latency that they find acceptable. Nvidia's NVOFA (Nvidia optical flow accelerator) actually has different quality presets accessible via API, it's just that Nvidia controls the implementation in sponsored games and doesn't allow multiple quality presets for the FG itself when the functionality is already available and documented by itself.

It's a very Apple-like approach but similar to Apple, if the market doesn't punish them for those tactics they will only continue using them. I'd wager that Nvidia has at least as strong mindshare as Apple had at it's peak.
dgianstefaniNo optical flow accelerator hardware unit
This is factually incorrect. Both the 2000 and 3000 series have it: docs.nvidia.com/video-technologies/optical-flow-sdk/nvofa-application-note/

Nvidia's documentation lists the 3000 series being fully feature equivalent to the 4000 series as well. The 2000 series lacks some of the latest feature set but is still fully capable and equipped with an optical flow accelerator.
Posted on Reply
#7
dgianstefani
TPU Proofreader
evernessinceIt's 100% to sell cards and nothing else. Latency of cards prior to 4000 series would be worse with FG enabled but that can be fixed by including multiple quality levels where lower quality levels would result in lower latency. In fact that would be a nice option to have even for 4000 series owners as then users can choose a setting with latency that they find acceptable. Nvidia's NVOFA (Nvidia optical flow accelerator) actually has different quality presets accessible via API, it's just that Nvidia controls the implementation in sponsored games and doesn't allow multiple quality presets for the FG itself when the functionality is already available and documented by itself.

It's a very Apple-like approach but similar to Apple, if the market doesn't punish them for those tactics they will only continue using them. I'd wager that Nvidia has at least as strong mindshare as Apple had at it's peak.



This is factually incorrect. Both the 2000 and 3000 series have it: docs.nvidia.com/video-technologies/optical-flow-sdk/nvofa-application-note/

Nvidia's documentation lists the 3000 series being fully feature equivalent to the 4000 series as well. The 2000 series lacks some of the latest feature set but is still fully capable and equipped with an optical flow accelerator.
They have an early version of it that is not as powerful or as refined. It's not really actively used outside of development. DLSS3 FG also requires fourth gen Tensor cores
A combination of OFA and the 4th Gen Tensor core is why the Ada architecture is required to use DLSS 3, and why it won't work on older architectures.
You can try to call it artificial segmentation if you want, but NVIDIA is competing with itself at this point, so it's not surprising even if that were true, which strictly it isn't. Hardware does change gen to gen, and it's not unreasonable to see how something that works as well as DLSS FG (compared to FSR FG) took a while to perfect.

Besides, most of, if not all of the other DLSS features have been backported to RTX2/3 Series cards, so I don't think it's fair to say that they're fully feature equivalent hardware wise, because they're not.
DLSS 3.0 makes use of a new generation Optical Flow Accelerator (OFA) included in Ada Lovelace generation RTX GPUs. The new OFA is faster and more accurate than the OFA already available in previous Turing and Ampere RTX GPUs.[27] This results in DLSS 3.0 being exclusive for the RTX 40 Series.
Posted on Reply
#8
Vayra86
dgianstefaniThey have an early version of it that is not as powerful or as refined. It's not really actively used outside of development. DLSS3 FG also requires fourth gen Tensor cores



You can try to call it artificial segmentation if you want, but NVIDIA is competing with itself at this point, so it's not surprising even if that were true, which strictly it isn't. Hardware does change gen to gen, and it's not unreasonable to see how something that works as well as DLSS FG (compared to FSR FG) took a while to perfect.

Besides, most of, if not all of the other DLSS features have been backported to RTX2/3 Series cards, so I don't think it's fair to say that they're fully feature equivalent hardware wise, because they're not.
It is artificial segmentation in the very same way Intel 'needs' a new socket every other gen.

Its not needed to maintain leadership or advantages. Intel has done a fine job proving that. This serves a singular purpose: profit maximization and business strategy. There is no technological hurdle here. Stop fooling yourself. Nvidia's approach is always like this. A supposed requirement eventually turns out to be not quite so necessary. Cue Gsync. Cue RTX. Cue DLSS. Cue PhysX. The list is endless.

Its all business, saying otherwise is naive, and devoid of historical context.
Posted on Reply
#9
dgianstefani
TPU Proofreader
Vayra86It is artificial segmentation in the very same way Intel 'needs' a new socket every other gen.

Its not needed to maintain leadership or advantages. Intel has done a fine job proving that. This serves a singular purpose: profit maximization and business strategy. There is no technological hurdle here. Stop fooling yourself. Nvidia's approach is always like this. A supposed requirement eventually turns out to be not quite so necessary. Cue Gsync. Cue RTX. Cue DLSS. Cue PhysX. The list is endless.

Its all business, saying otherwise is naive, and devoid of historical context.
Meaningless to 95%+ of Intel's customers. Despite what enthusiast forums might have you believe, the vast majority of people do not upgrade their CPU or would even know how to.

OK, NVIDIA is a business? Still doesn't mean the three generations of RTX have the same hardware. It's related, but it's not the same.

You can call me naive if you want, but the RTX/DLSS feature set has consistently grown, for free, for all cards of the three generations since release. Even things like RTX Video super resolution, which was delayed for first gen RTX cards for a few months, still made its way even to the RTX 2060. A singular feature that many people whine about anyway, "fake frames", is exclusive, and suddenly it's a conspiracy from NVIDIA to artifically segment their cards.

Personally I prefer the DLSS approach such as its been, than how AMD have handled FSR, which might finally be good with the upcoming AI trained version announced recently, but I guess we'll see.

Try turning on FSR fluid motion frames and see if you enjoy it. I didn't. That's the only competition at the moment unless Intel pulls something out of their hat, which wouldn't surprise me.

Your example of GSYNC module being a "supposed" requirement is also disingenuous. There are significant differences in the hardware implementations between "ultimate" and "compatible" monitors, and not just testing/validation etc.

Same for the RTX thing.
Posted on Reply
#10
evernessince
dgianstefaniThey have an early version of it that is not as powerful or as refined. It's not really actively used outside of development. DLSS3 FG also requires fourth gen Tensor cores



You can try to call it artificial segmentation if you want, but NVIDIA is competing with itself at this point, so it's not surprising even if that were true, which strictly it isn't. Hardware does change gen to gen, and it's not unreasonable to see how something that works as well as DLSS FG (compared to FSR FG) took a while to perfect.

Besides, most of, if not all of the other DLSS features have been backported to RTX2/3 Series cards, so I don't think it's fair to say that they're fully feature equivalent hardware wise, because they're not.
You said they had no optical flow accelerator at all, which was false. The rest of your comment is just changing the goal posts to try and gloss over that fact.
Posted on Reply
#11
dgianstefani
TPU Proofreader
evernessinceYou said they had no optical flow accelerator at all, which was false. The rest of your comment is just changing the goal posts to try and gloss over that fact.
You got me. Congrats. It's not a lack of the unit altogether, although it's not the same unit, but a lack of one that can effectively be used by the technology. To be clear, it's an older version, just because it has the same name does not mean they have the same capabilities, which they do not.
DLSS 3.0 makes use of a new generation Optical Flow Accelerator (OFA) included in Ada Lovelace generation RTX GPUs. The new OFA is faster and more accurate than the OFA already available in previous Turing and Ampere RTX GPUs.[27] This results in DLSS 3.0 being exclusive for the RTX 40 Series.
You're also continuing to pretend that the spec doesn't clearly state that 4th gen Tensor cores are also required. While asserting that all RTX cards are fully feature equivalent, which seems at odds with the fact that there is a single technology which NVIDIA states is incompatible with previous generations. I'd assume this is something along the lines of how RTX 3000 can technically do path traced native processing in real time, but you'll have to enjoy 20 FPS, even with a 3090 Ti, still technically "feature compatible" but I don't see that being used by anyone. Speculating that you could make DLSS3 FG work on older generations is just that, speculation. We don't know that.
A combination of OFA and the 4th Gen Tensor core is why the Ada architecture is required to use DLSS 3, and why it won't work on older architectures.
The rest of my comments are true, and are not moving the goalposts.

I'd assume the 4th gen Tensor cores are required to do the complex new algorithms in real time.

Early TPU reviews show the tech working well.
www.techpowerup.com/review/palit-geforce-rtx-4070-ti-gamingpro-oc/35.html

UpgrayeddCould someone explain why older RTX cards couldn't get a different version of DLSS FG? besides selling newer cards.
Instead of doubling why can't we just get 20% or 25%? Would the latency be too high? Too much ghosting? would older cards even be capable of that or would they be resolution limited?
The original question was also "why can't we just get 25% more performance".

This likely stems from a fundamental misunderstanding of frame generation. It's not like DLSS super resolution, where a lower internal resolution is combined with AI upscaling to offer a relatively significant performance increase, that can vary in %. You can use FG with or without upscaling. You could for instance use it with DLAA.

Frame generation is the calculation of an intermediate frame between two native/upscaled/DLAA frames. Therefore the only way the tech can work is by offering doubling of frame rates. You can't add a generated frame once every four native frames for instance, to get a 20-25% increase in performance.

The quality of the generated frames relies on having a decent "before" FPS, so essentially 60 FPS or better, optimally. This is why DLSS upscaling is generally used in tandem, but you don't have to use both technologies concurrently.
Posted on Reply
#12
Random_User
dgianstefaniGray Zone looks like an actual Tarkov competitor with an actual good game engine.

But we'll see.
By the sole look, it reminds the Joint Operations: Typhoon Rising.
Posted on Reply
#13
dgianstefani
TPU Proofreader
Random_UserBy the sole look, this might be finally a successor to Joint Operations: Typhoon Rising.
Mmm, looks like an interesting game for sure.

Saw a little early gameplay from content creators, looked promising but poor performance/audio etc. Normal for game in that stage of development.
Posted on Reply
#14
evernessince
dgianstefaniYou're also continuing to pretend that the spec doesn't clearly state that 4th gen Tensor cores are also required. While asserting that all RTX cards are fully feature equivalent, which seems at odds with the fact that there is a single technology which NVIDIA states is incompatible with previous generations. I'd assume this is something along the lines of how RTX 3000 can technically do path traced native processing in real time, but you'll have to enjoy 20 FPS, even with a 3090 Ti, still technically "feature compatible" but I don't see that being used by anyone. Speculating that you could make DLSS3 FG work on older generations is just that, speculation. We don't know that.

I'd assume the 4th gen Tensor cores are required to do the complex new algorithms in real time.
Nvidia stated that DLSS 1.0 - 1.9 required tensor cores when in fact in run perfectly fine on CUDA cores. You are making the assumption that it's a hard requirement and not an Nvidia fabricated one.

The fact of the matter is Nvidia is on record stating that the OFA on prior gen cards is not performant enough to run DLSS FG. Nothing about the tensor cores not being performant enough or due a lack of hardware level feature support.

The problem here is that as we can see from the Nvidia documention, multiple quality levels are available to enable higher performance so there's really no excuse as to why it can't be done on older cards. In addition, it's entirely possible to have a fallback akin to Intel's use of DP4a.

No, it's not an assumption to say DLSS FG could have worked on older RTX cards. They had all these tools in their toolbox and choose to use non of them on purpose. Anyone who's spent any time with AI knows that models can be pruned and reduced in size to reduce computational overhead often with little quality loss as Nvidia's own OFA medium quality setting demonstrated provided by Nvidia themselves.
dgianstefaniYou got me. Congrats. It's not a lack of the unit altogether, although it's not the same unit, but a lack of one that can effectively be used by the technology. To be clear, it's an older version, just because it has the same name does not mean they have the same capabilities, which they do not.
As I pointed out in my last comment and per Nvidia documentation the 3000 series and 4000 series are in fact feature equivalent:
evernessinceNvidia's documentation lists the 3000 series being fully feature equivalent to the 4000 series as well. The 2000 series lacks some of the latest feature set but is still fully capable and equipped with an optical flow accelerator.
Posted on Reply
#15
dgianstefani
TPU Proofreader
evernessinceNvidia stated that DLSS 1.0 - 1.9 required tensor cores when in fact in run perfectly fine on CUDA cores. You are making the assumption that it's a hard requirement and not an Nvidia fabricated one.

The fact of the matter is Nvidia is on record stating that the OFA on prior gen cards is not performant enough to run DLSS FG. Nothing about the tensor cores not being performant enough or due a lack of hardware level feature support.

The problem here is that as we can see from the Nvidia documention, multiple quality levels are available to enable higher performance so there's really no excuse as to why it can't be done on older cards. In addition, it's entirely possible to have a fallback akin to Intel's use of DP4a.

No, it's not an assumption to say DLSS FG could have worked on older RTX cards. They had all these tools in their toolbox and choose to use non of them on purpose.



As I pointed out in my last comment and per Nvidia documentation the 3000 series and 4000 series are in fact feature equivalent:
That's fair enough.

I guess they took the approach - we want quality to be as indistinguishable from native frames as possible, and they've certainly come close to achieving that. Without pixel peeking at still frames, or being an esports gamer (why would you turn it on in competitive twitch shooters anyway, those are generally very easy to run graphically), the FG implementation has had almost universal positive acclaim from reviewers and users. 120 FPS FG latency isn't as good as 120 native FPS, but it's essentially the same as 60 FPS with FG turned off, which is the alternative.

Would the theoretical DLSS implementation running on CUDA cores (that are also running the game) work to give the same performance/latency as Tensor cores? I doubt it.

Yes, I am making the assumption that it is a hard requirement, at least for the quality level NVIDIA is offering.

Perhaps as you say there is a way to get something working on older, slower cards with previous generation tensor and OFA hardware, but what would the result look like?
Posted on Reply
#16
evernessince
dgianstefaniThat's fair enough.

I guess they took the approach - we want quality to be as indistinguishable from native frames as possible, and they've certainly come close to achieving that.

Would the theoretical DLSS implementation running on CUDA cores (that are also running the game) work to give the same performance/latency as Tensor cores? I doubt it.
Their AI model should be able to run on both CUDA and Tensor cores (stable diffusion for example runs on CUDA primarily and will utilize Tensor if available and your implementation supports it).

Both Tensor and CUDA are utilized through the CUDA API. When it comes to a gaming application CUDA cores should only be used when necessary as their use deducts from gaming performance. You'd want to adjust quality of the OFA first and only use CUDA cores as a last resort. Looking at Nvidia's documention, Medium quality doubles performance of 2000 series which puts it almost on par with the 4000 series running at high quality so I'm not even sure having a DLSS FG implementation that utilizes CUDA cores would even be required (although it would give the user more options if they for example want higher quality and don't mind a small hit to FPS). It's more power to the end user as they can tailor quality, fps, and features to their liking.
Posted on Reply
#17
FiRe
Why does the top image show the old game name of Lords of the Manor? lol
Posted on Reply
Add your own comment
May 4th, 2024 04:50 EDT change timezone

New Forum Posts

Popular Reviews

Controversial News Posts