Nvidia ’s RTX serial include some of thebest graphics cardson the mart and are known for two flagship features : existent - time re tracing and Deep Learning Super Sampling ( DLSS ) . While ray trace is no longer an undivided characteristic as it is uncommitted on other GPUs as well as console like the PlayStation 5 and Xbox Series X , DLSS stay a bit more mysterious to many . At its core , DLSS uses artificial insemination - power upscaling to return games at a blue resolution and intelligently reconstruct them to wait like — or even better than — aboriginal 4 green , significantly boosting functioning without sacrificing optic fidelity .
With the arrival of DLSS 4 and the latest RTX 50 - series GPUs , the technology has evolve even further , deliver smoother gameplay , undecomposed image quality , and sassy frame generation . It ’s a win - profits : higher frame rates , crisper visuals , and less strain on your GPU .
But there ’s more hold out on under the hood than just upscaling . Here ’s everything you demand to sleep with about DLSS , how it work , and why it ’s become a must - have feature for modern PC gaming .
DLSS 2.0 offImage used with permission by copyright holder
What is DLSS?
DLSS stand for deep learning supersampling . The “ supersampling ” snatch refers to an anti - aliasing method acting that smooths the notched edges that show up on render graphics . Over other kind of anti - aliasing , though , SSAA ( supersampling anti - aliasing ) work by rendering the trope at a much high resolution and using that information to fill in the disruption at the native resolution .
The “ deep learning ” part is Nvidia ’s mysterious sauce . Using the king of machine learning , Nvidia can train AI mannequin with gamy - firmness scans . Then , the anti - aliasing method can use the AI role model to fill in the missing information . This is important , as SSAA usually requires you to render the mellow - resolution icon topically . Nvidia does it offline , away from your computer , providing the benefit of supersampling without the computing overhead .
This all works thanks to Nvidia ’s Tensor Cores , found solely in RTX artwork cards ( by from enterprise - grade root like theNvidia A100 ) . Tensor Cores are purpose - built to accelerate the variety of complex ground substance operation and motley - precision calculation used in AI and abstruse learning . By leveraging lower - preciseness data formatting like FP16 while save accuracy , they dramatically race up chore like neural electronic connection training and inference — and in play , that power is harness to drive features like DLSS .
DLSS 2.0 PerformanceImage used with permission by copyright holder
Nvidia ’s late RTX 50 - serial nontextual matter calling card introduce fifth - contemporaries Tensor Cores , pushing DLSS performance even further . With the addition of DLSS Multi - Frame Generation , these GPUs can now employ AI to generate up to three surplus frames for every traditionally rendered one — deliver a significant FPS encouragement without compromising ocular character .
There is no doubt that Nvidia is leading the kick in this area , though AMD’sFidelityFX Super Resolution featureis slow down try out to overtake up . Even Intel has its own supersampling engineering science calledIntel XeSS , or Intel Xe Super Sampling .
What does DLSS actually do?
DLSS is the result of an exhaustive procedure of teaching Nvidia ’s AI algorithm to generate unspoilt - expect games . After rendering the biz at a down in the mouth answer , DLSS infers selective information from its knowledge base of first-rate - resolution double grooming to generate an image that still looks like it was running at a high resolution . The theme is to make games rendered at 1440p look like they ’re run at 4 K or 1080p games to attend like 1440p . DLSS 2.0 offered four times the resolution , allowing you to render games at 1080p while output them at 4K.
More traditional crack - resolution techniques can pass to artifact and bugs in the eventual picture , but DLSS is design to work with those errors to generate an even better - looking image . In the right-hand circumstances , it can birth substantial performance uplift without affecting the look and feel of a game ; on the reverse , it can make the game look even best .
With the presentation of DLSS 3 on RTX 40 - serial of graphics cards , the frame charge per unit gains were even more substantive . Nvidia not only used Tensor cores to make underframe front better , but frames could be rendered using just AI . Additionally DLSS 3 delivered a performance boost by return entire fresh frame of reference between rendered frame , improving human body rates and responsiveness in games , especially on GPU- and CPU - limited systems .
DLSS 2.0 QualityImage used with permission by copyright holder
DLSS 4 is a footfall up from DLSS 3 , mainly thanks to Multi Frame Generation , which encourage performance by using AI to create multiple frames for every one the GPU renders . secondly , it uses a unexampled Transformer - found AI simulation that delivers clearer , more detailed image . DLSS 4 also improves exist features like Ray Reconstruction , have light effects reckon penetrating and textures more accurate .
How does DLSS work?
DLSS forces a game to render at a humiliated resolution ( typically 1440p ) and then uses its trained AI algorithm to understand what it would count like if it were rendered at a higher one ( typically 4 K ) . It does this by utilizing some anti - aliasing effect ( likely Nvidia ’s own TAA ) and some automatize sharpening . Visual artifacts that would n’t be present at higher firmness are also ironed out and even used to understand the item that should be present in an double .
AsEurogamer explains , the AI algorithm is take aim to look at sure games at extremely high resolutions ( supposedly 64x supersampling ) and is distil down to something just a few MB in size before being added to the latest Nvidia machine driver releases and made approachable to gamers all over the earthly concern .
In core , DLSS is a real - time variant ofNvidia ’s screenshot - enhancing Ansel engineering . It renders the range at a abject result to render a performance boost , then give various impression to save a relatively like overall gist to conjure up the resolution .
The result can be a mixed bag , but in general , it leads to higher frame charge per unit without a substantial passing in ocular fidelity . Nvidia claims skeletal frame pace can improve by as much as 75 % in Remedy Entertainment’sControlwhen using both DLSS and ray trace . It ’s usually less marked than that , and not everyone is a fan of the eventual look of a DLSS biz , but the selection is certainly there for those who want to fancify their plot without the cost of pass at a higher resolution .
The journey get down with DLSS 1.0 , which launch in 2019 . It used a per - game trained AI model that rendered games at a lower resolution and then upscaled them to a higher one . However , this version had some major drawbacks . Because each plot required specific breeding data point , adoption was circumscribed and inconsistent . Image timbre also change widely across titles , with many players reporting fuzz and ghosting artifact that degrade the experience .
With the introduction of DLSS 2.0 Nvidia move to a generalized AI model that no longer required game - specific breeding . This version introduced temporal feedback , meaning it used data from previous frames and move transmitter to reconstruct details more accurately . DLSS 2.0 included multiple mood — Quality , Balanced , Performance , and Ultra Performance — allow for users to choose between higher trope fidelity or better frame rates .
DLSS 3 , introduced in 2022 with the RTX 40 serial , added another level : Frame Generation . Unlike previous versions that only upscaled images , DLSS 3 could now bring forth entirely new frames using AI . This was made possible by an opthalmic stream accelerator that see motion between two frame and augur what an average skeleton should look like . This technology allowed for noticeable performance boosts , particularly in CPU - stick scenario , and was exclusive to the RTX 40 series and new . While figure generation can stick in some latency , Nvidia include Reflex applied science to minimize remark meantime and keep games responsive . fundamentally , DLSS 3 is a combination of DLSS 2 ’s Super Resolution and the unexampled AI - power Frame Generation .
DLSS 4 reinvents the technology by adding more frames
Back in January , Nvidia announce DLSS 4 — the latest loop of the technology that will be available to the possessor of RTX 50 - serial graphics cards . As mentioned , the biggest change this time around is the introduction of Multi Frame Generation which brings three extra AI - generated frame , yield a significant encouragement to frame rates on supported games .
There are currently six GPUs that bear out DLSS 4 :
In ourRTX 5090 review , we happen that DLSS 4 land a routine of meaningful rising slope to Nvidia ’s RTX 50 - series GPUs . One of the most noted changes is the shift from a convolutional neuronic meshwork ( CNN ) to a Transformer - base AI model . This update ameliorate how the GPU reconstructs frames , leading to more precise details and clean image character — especially in tight - moving scene where earlier versions of DLSS sometimes clamber .
Another fundamental feature is DLSS Multi - Frame Generation ( MFG ) , which expand on the frame genesis introduced in DLSS 3 . alternatively of engender just one extra frame , DLSS 4 can now generate up to three AI frames for every one frame the GPU render . This lead in a significant carrying into action boost , though image tone still reckon heavily on the specific game carrying out . DLSS 4 ’s multi - frame generation trust on a technique called build interpolation , the same feeler used in DLSS 3 as well as other frame coevals tools like Lossless Scaling and AMD ’s FSR 3 .
In examination , Cyberpunk 2077 see one of the biggest improvements , with operation virtually quadruple when DLSS 4 was enable . Other game like Alan Wake 2 , Hogwarts Legacy , and Microsoft Flight Simulator also showed large gain in framing rate , specially when ray tracing was turned on . However , it is important to take down that the existent experience can vary — in some titles , the optical lucidity of DLSS 4 was fantabulous , while others still show artifacts or easy edges in motion .
Overall , DLSS 4 is n’t just a pocket-sized update — it brings real public presentation gains , especially in games that already affirm in advance ray tracing feature . But as with past variation , the quality of results still depends on how well individual games implement the engineering .
DLSS 4 is still Modern and is slowly making its mode into more games . Here are the titles that presently stomach DLSS 4 :
touch toNvidia ’s DLSS 4 games listto see the latest compatible titles .
DLSS versus FSR versus RSR versus XeSS
AMD is Nvidia ’s grownup rival when it comes to graphic technology . To vie with DLSS , AMD FidelityFX Super Resolution ( FSR)was let go of in 2021 . Although it achieve the same finish of amend visuals while raising frame rates , FSR make quite differently from DLSS . FSR renders frame at a low closure and then use an open - reference spatial upscaling algorithm to make the biz look like it is running at a mellow closure and does n’t factor in movement vector data . DLSS use an AI algorithm to hand over the same outcome , but this technique is only back up by Nvidia ’s own RTX GPUs . FSR , on the other hand , can figure out on just about any GPU .
In addition to FSR , AMD also has Radeon Super Resolution ( RSR ) , which is a spacial upscaling technique that makes employment of AI . While this sound similar to DLSS , there are differences . RSR is build using the same algorithmic rule as FidelityFX Super Resolution ( FSR ) and is a equipment driver - based feature that is delivered via AMD ’s Adrenalin software . RSR get to fill the gap where FSR is not useable , as the latter has to be implemented right into specific game . basically , RSR should work in almost any game , as it does n’t postulate developer to follow out it . Notably , FSR is usable across newer Nvidia and AMD GPUs , and RSR , on the other hand , is only compatible with AMD ’s RDNA visiting card , which admit the Radeon RX 5000 and above .
FSR continues to evolve , with FidelityFX Super Resolution 4 ( FSR 4 ) being the latest iteration of its upscaling technology . AMD is finally leverage AI - power skeletal frame multiplication which is claim to importantly advance underframe rate , with AMD claiming up to a 3.7x performance increase in abide titles . This version introduce modern upscaling technique and improved secular stability , resulting in acute visuals and smoother gameplay . FSR 4 is compatible with games that supportFSR 3.1 , although desegregation may command manual activation through machine driver mise en scene . While ab initio developed for RDNA 4 - base GPUs including theRadeon RX 9000 series , AMD has argue plans to optimize FSR 4 for wide hardware compatibility in the future .
Intel has also been working on its own supersampling engineering science calledXe Super Sampling ( XeSS ) , and unlike with FSR or DLSS , there are two different versions available . The first make use of the XMX intercellular substance math unit , which are present in its newArc Alchemist GPUs ; these XMX units take care of all the AI processing on the ironware end . The other interpretation makes use of the wide accepted four - element transmitter dot Cartesian product ( DP4a ) instruction , thus removing the dependency from Intel ’s own computer hardware and allowing XeSS to work on Nvidia and AMD GPUs .
Last yr Intel introduced itsBattlemage seriesof GPUs ( B580 and B570 ) where it announced its updated XeSS 2 AI - powered upscaling technology design to amend both gaming performance and visual quality . Building on the original XeSS , it inclose two major feature film : Frame Generation ( XeSS - FG ) and Xe Low Latency ( XeLL ) . XeSS - FG encourage frame pace by generating extra AI - driven skeleton between rendered 1 , while XeLL reduces input retardation for a more reactive gaming experience . The core of XeSS 2 , XeSS Super Resolution ( XeSS - SR ) , utilize AI to upscale lower - resolve frames to the showing ’s aboriginal resolving , similar to Nvidia ’s DLSS and AMD ’s FSR . These enhancement are especially effective on Intel Arc GPUs , positioning XeSS 2 as a unmediated competitor to DLSS 3 and FSR 3 .
