RTX 5090 and 5080 leaks! RTX 50 Neural Rendering?!? Strix Halo| Nvidia App issues confirmed | More!!
Ringkasan
TLDRVideoen fokuserer på flere teknologinyheder, primært inden for gaming og hardware. Det inkluderer forventninger om Nvidia's kommende præsentation af RTX 50-serien ved CES, hvor neural rendering i videospil bliver et stort emne. Neural rendering vil muligvis revolutionere den måde, grafik behandles og vises på. Acer har ved en fejl offentliggjort en gaming-computer konfigureret med RTX 5090 og 5080 grafikkort, som bekræfter specifikationer som 32 GB VRAM for 5090 og 16 GB for 5080. Der er også diskussion af ydeevneproblemer med Nvidia's app, hvor brugere rapporterer reduceret spilpræstation. AMD's Strix Halo-serie og dens potentiale inden for integrerede grafikenheder bliver også nævnt, sammen med enkelte software-opdateringer og spilnyheder.
Takeaways
- 💻 RTX 50-serien afsløres muligvis ved CES med fokus på neural rendering.
- 📊 Acer har ved en fejl offentliggjort specifikationer for kommende gaming PC'er.
- 🕹 Nvidia undersøger ydeevneproblemer i deres app.
- ⚡ Ny GDDR7 VRAM forventes i RTX 5080 og 5090.
- 🔍 Nvidia avancerer med neural rendering og AI integration i grafik.
- 🚀 AMD's Strix Halo APU ser lovende ud for integreret grafik.
- 🎮 Softwareopdateringer forbedrer spiloplevelser og ydelse.
- 🔧 Intel Arc B570 review enheder er begyndt at blive sendt ud.
- 🎲 Witcher 4 projekt bekræfter inklusion af Gwent.
- 📈 Benchmark leak afslører potentielle evner af AMD's kommende chips.
Garis waktu
- 00:00:00 - 00:05:00
Videoen diskuterer forventede annonceringer ved CES i januar, herunder Nvidia RTX 50-serien med fokus på neural rendering i videospil. Andre emner inkluderer hardwarelækager fra Acer og Nvidia's undersøgelser af reduceret ydeevne ved brug af deres app. AMDs nye APU, Strix Halo, og Intels B570 omtales også.
- 00:05:00 - 00:10:00
Fortsætter med at diskutere neural rendering og avanceret DLSS teknologi. Indlæg fra Inno3D forventes at fokusere på revolutionerende grafiske processer, AI-accelereret grafik og forbedret AI i spil- og indholdsskabelsesarbejdsbyrder, herunder kraftværdi AI-assisterede opgaver.
- 00:10:00 - 00:15:00
Der diskuteres Nvidia's neural kompressionsteknik for materialeteksturer, der muliggør høj opløsning uden øget lager og hukommelsesefterspørgsel. Videoen refererer også til forskning på dette område fra Siggraph 2023 og 2024, der muliggør brug af filmkvalitetsvisuelle billeder i realtid applikationer.
- 00:15:00 - 00:20:00
Nye teknikker og systemer til neural rendering introduceres, herunder realtids neural appearance models, der muliggør hurtigere og mere effektive renderingmetoder. Potentialet i at integrere disse teknologier i videospil og den fremtidige anvendelse af CES præsenteres også.
- 00:20:00 - 00:25:00
Acer lækker information om RTX 5090 og 5080 med henholdsvis 32 GB og 16 GB VRAM. Bekymringer om VRAM-grænsen for de kommende GPU'er diskuteres, især med hensyn til fremtidssikring og prispunkt. Leaks om hukommelseshastigheder og specifikationer for den nye RTX-serie nævnes også.
- 00:25:00 - 00:31:47
Hardware unboxed tester Nvidias app-relaterede performance problemer og AMD's Strix Halo APU benchmarks. Tekniske detaljer om nye chip egenskaber og potentiale i gaming ydeevne samt en notits om opdateringer af et spil diskuteres, samtidigt med at videoen afsluttes med information om Witcher 4.
Peta Pikiran
Video Tanya Jawab
Hvad forventes der præsenteret ved CES?
Ved CES forventes der præsentation af RTX 50-serien samt teknologier relateret til neural rendering af videospil.
Hvilken information er lækket fra Acer?
Acer har utilsigtet listet en gaming-PC udstyret med RTX 5090 og 5080 grafikkort.
Hvad er neural rendering?
Neural rendering angår revolutionerende metoder til at processere og vise grafik ved brug af AI-teknologier.
Hvad er RTX 5080's forventede specifikationer?
RTX 5080 er forventet at have 16 GB GDDR7 VRAM.
Hvilke problemer er rapporteret med Nvidia's app?
Nvidia's app reducerer ydeevnen i spil og det er rapporteret, at de arbejder på en løsning.
Lihat lebih banyak ringkasan video
CLEANING OUT MY CLOSET (zero effort but good enough)
Pharmacology II: Drugs for Epilepsy part 1
5 BRAND NEW Digital Marketing Strategies for 2025 (BIGGEST Change Yet!)
Elizabeth Warren Grills United Health CEO Months Before Luigi Mangione Murder
AWS re:Invent 2024 - Leverage Anthropic's Claude models for AI's evolving landscape (AIM123)
Unboxing YOUR Squishy Packages | Squishy Makeover Candidates
- 00:00:00all signs point to CES in January 6 not
- 00:00:03only showing us the RTX 50 Series for
- 00:00:06the first time but also the era of
- 00:00:09neural rendering of video games more on
- 00:00:13that and where am I getting this
- 00:00:15demonstration of it in just a minute uh
- 00:00:18but we have a bunch of other news to
- 00:00:19talk about today including Acer
- 00:00:22accidentally listing a 55090 and 80
- 00:00:25equipped gaming PC ahead of time
- 00:00:27confirming only 16 GB of vram
- 00:00:30on the RTX 580 although it's gddr 7 and
- 00:00:33how fast we have some leaks discussing
- 00:00:36that as well as we have Nvidia saying
- 00:00:39they are investigating the reduced
- 00:00:41performance uh by using their Nvidia app
- 00:00:45and there's some uh Hardware unbox
- 00:00:47testing discussing exactly what in the
- 00:00:50app is causing the issues and confirming
- 00:00:53those issues we also have some exciting
- 00:00:56uh leaks regarding stricks Halo this is
- 00:00:59an amazing amazing looking Apu from AMD
- 00:01:02and then we also have apparently Intel's
- 00:01:05b570 which if you're excited about the
- 00:01:07b580 with 12 GB hey 10 GB on the
- 00:01:11b570 Apparently review samples already
- 00:01:14out to at least one review Outlet uh for
- 00:01:17that
- 00:01:18b570 uh we also have a little bit more
- 00:01:21gaming information towards the end a
- 00:01:24whole bunch more about neural rendering
- 00:01:25in just a second but first unfortunately
- 00:01:28I have to admit that I've been spreading
- 00:01:30misinformation in the comment section or
- 00:01:32at least information I spread is now out
- 00:01:34of date Mr Daniel do you sleep no has
- 00:01:37been upgraded to yes because I am now
- 00:01:40proudly Pro naap honestly guys I moved
- 00:01:43closer to my uh job teaching cuz I also
- 00:01:46do that on top of running the YouTube
- 00:01:47channel fulltime people are like how are
- 00:01:48you doing this power naaps in the
- 00:01:51afternoon are now possible I get home
- 00:01:53from work quick get in a power nap
- 00:01:54before my kids get home and it's been
- 00:01:56amazing now in the past I couldn't nap I
- 00:02:00am now completely blacked out cannot see
- 00:02:03a bit of light and also I can kick on
- 00:02:06some white noise right here with
- 00:02:08completely flat Bluetooth speakers which
- 00:02:10is amazing uh because I'm a sid sleeper
- 00:02:13and headphones in the past have been
- 00:02:15incredibly painful making it difficult
- 00:02:18for me to nap this is also really nice
- 00:02:20when I go to sleep at night a lot of
- 00:02:21times I like to put on like an Audi book
- 00:02:22or podcast and H I can't tell where any
- 00:02:25of you are this is completely blacked
- 00:02:26out anyway I'm take that back off uh but
- 00:02:28the point is um that this has completely
- 00:02:31revolutionized both my actual sleep at
- 00:02:34night as well as my ability to actually
- 00:02:37get in some afternoon naps which is uh
- 00:02:40re really amazing really it has made me
- 00:02:42so much more alert and productive there
- 00:02:44is a wide variety available again this
- 00:02:47is the uh sound version uh which I think
- 00:02:50is really cool but they also have the uh
- 00:02:52pro version they have the silk mask uh
- 00:02:55Manta weighted mask we've got a cool
- 00:02:58mask a steam mask mask uh the original
- 00:03:01Manta sleep mask and a huge thank you to
- 00:03:03Manta sleep for sponsoring today's video
- 00:03:06and allowing me to be proudly prap and
- 00:03:08if you check out the link in the video
- 00:03:09description you can get some amazing
- 00:03:11deals right now on holiday gifts there's
- 00:03:14a 15% off sitewide sale plus a free
- 00:03:18mystery sleep kit uh for those who miss
- 00:03:21the Cyber weekend this is the perfect
- 00:03:22opportunity to grab a great deal on
- 00:03:24sleep accessories for yourself or gifts
- 00:03:26can't emphasize this enough these make
- 00:03:28amazing gifts
- 00:03:30uh and again a mystery sleep kit
- 00:03:32included with the 15% off sitewide sale
- 00:03:35if you click the link in the video
- 00:03:36description so what do I mean about
- 00:03:39neural rendering of video games where am
- 00:03:41I getting this footage and why do I
- 00:03:43think this is very likely to be a big
- 00:03:45Topic at CES with the 50 Series well
- 00:03:50let's dive into my journey down this
- 00:03:54pathway today uh first of all Hardware
- 00:03:56Lux has a article this is a transl from
- 00:04:00German talking about Ino 3D's
- 00:04:04announcement for what they will be
- 00:04:05covering at uh CES now within that post
- 00:04:10if you actually follow up the actual
- 00:04:11in3d post uh they talk about their
- 00:04:14coolers and whatnot but then we get this
- 00:04:18this is where we get into neural
- 00:04:20rendering there's a lot going on here uh
- 00:04:23one of them is specifically neural
- 00:04:26rendering capabilities revolutionizing
- 00:04:29how Graphics are processed and displayed
- 00:04:33so that is one of the highlights for Ino
- 00:04:363D's C presentation talking about their
- 00:04:40products related to gpus now it's very
- 00:04:43likely that board partners that work
- 00:04:45closely with Nvidia are fully aware of
- 00:04:48what nvidia's big CES talking points are
- 00:04:51going to be and have presentations to
- 00:04:53follow up and expand on how their
- 00:04:55products relate to these so there's the
- 00:04:58neural rendering cap capabilities being
- 00:05:00referenced right here uh along with this
- 00:05:04we also have discussion of advanced dlss
- 00:05:07technology now it's possible that
- 00:05:10they're just talking about dlss
- 00:05:12technology that already exists and the
- 00:05:14fact that it is Advanced it's also
- 00:05:17possible that this is referring to a
- 00:05:20further advancement of dlss Technology
- 00:05:23although it's still just discussing uh
- 00:05:25even better image quality and higher
- 00:05:27frame rates so potentially this is um
- 00:05:30you know expansion of what's kind of
- 00:05:32already out there but it doesn't sound
- 00:05:33like an entirely new thing um that is
- 00:05:37like somehow different than the dlss
- 00:05:40that we already have at a fundamental
- 00:05:41level they also talk about enhanced rate
- 00:05:43tracing improved RT cores um AI
- 00:05:47accelerated Graphics now this is talking
- 00:05:49about improved performance in AI
- 00:05:50assisted tasks and better integration of
- 00:05:52AI in gaming and content creation
- 00:05:54workloads okay but then then we get into
- 00:05:56that neural rendering capabilities
- 00:05:58revolutionizing how Graphics are
- 00:06:00processed and displayed uh they're
- 00:06:02talking about AI enhanced power
- 00:06:03efficiency whether that's something that
- 00:06:05in3d is doing themselves or that'll be a
- 00:06:08wider scale thing for the 50 Series in
- 00:06:10general is unknown at this point uh then
- 00:06:12we have improved Aid driven upscaling
- 00:06:15however this is talking about Beyond
- 00:06:17gaming so this is more for Content
- 00:06:18creators and they also have generative
- 00:06:21AI acceleration so at this point the
- 00:06:24biggest one here that stands out as
- 00:06:26something that is potentially new is
- 00:06:29neural rendering
- 00:06:31capabilities so I know that first of all
- 00:06:34that rung a bell talking about some
- 00:06:36Nvidia white papers that I've reported
- 00:06:38on in the past uh from sigraph
- 00:06:41presentations so we've already talked
- 00:06:44about um uh neural texture compression
- 00:06:49in previous videos at sigraph
- 00:06:522023 Nvidia presented research on the
- 00:06:55neural compression of
- 00:06:57materials however I went in to see if
- 00:07:00they had any newer presentations related
- 00:07:03to neural rendering at the 2024 sigraph
- 00:07:08and they do we have realtime neural
- 00:07:11appearance models and that's the video
- 00:07:15that I was showing you here this is from
- 00:07:17their materials available uh for the
- 00:07:202024 sigraph presentation on neural
- 00:07:24materials you can see the diagonal bar
- 00:07:27across the screen here where where this
- 00:07:29is the ground truth reference model in
- 00:07:31the bottom right corner and the top left
- 00:07:34corner is the neural material pathway
- 00:07:37now this just switched to the 2x 32
- 00:07:40neurons Which is less accurate than the
- 00:07:422x 64 that we saw
- 00:07:45initially uh but you can see that even
- 00:07:48at this level there's uh quite a lot of
- 00:07:51accuracy here also it's important to
- 00:07:54point out that this does not mean I I'm
- 00:07:57sure you guys have seen videos talking
- 00:07:59about entirely AI generated games where
- 00:08:03you're literally playing a game that is
- 00:08:05basically being hallucinated by an AI
- 00:08:07and that is not what I think we're going
- 00:08:09to be getting from Nvidia when we look
- 00:08:12at neural rendering capabilities I think
- 00:08:14it's more along the lines of uh these
- 00:08:17neural compressions that we've seen in
- 00:08:20uh previous
- 00:08:21videos as well as potentially uh this
- 00:08:25neural appearance models so what is
- 00:08:27going on here I have fully read the the
- 00:08:30available information here and to be
- 00:08:32honest a lot of this is rather indepth
- 00:08:36regarding the actual implementation of
- 00:08:38how this works uh the abstracts
- 00:08:42themselves provide a bit more of a kind
- 00:08:45of summary of what is going on here that
- 00:08:49perhaps we can delve into a bit so again
- 00:08:51the big neural uh uh rendering stuff
- 00:08:54that I could find research on is these
- 00:08:56neural materials sorry neur neural
- 00:08:59appearance models and then neural
- 00:09:01compression so neural compression uh of
- 00:09:05material textures seems a little simpler
- 00:09:07to wrap your head around uh the idea
- 00:09:10here and and I've I've again I've shown
- 00:09:12this in uh videos you know back after
- 00:09:14the 2023
- 00:09:15sigraph uh is that the textures on the
- 00:09:18left here are uh neural textures on the
- 00:09:21right is BC uh bcx textures that's a
- 00:09:24more uh traditional method of doing this
- 00:09:27and you can see that these are similar
- 00:09:29texture size at 3.6 like storage size
- 00:09:313.6 megabytes on the left versus 3.33 on
- 00:09:34the right however I play the video uh
- 00:09:36you can see as it scans through here on
- 00:09:39the left you have massively more detail
- 00:09:42at just a slightly larger uh file size
- 00:09:45that's cool however there's a major
- 00:09:48downside to this which is the uh 1.15
- 00:09:52milliseconds to uh to do this on an RTX
- 00:09:554090 GPU at Native 4K whereas this one
- 00:09:58was 0. for 9 milliseconds in the video
- 00:10:01if you listen to it they describe uh you
- 00:10:03know rounding off as about uh 1
- 00:10:05millisecond of overhead now a
- 00:10:08millisecond doesn't sound like a lot but
- 00:10:09in terms of uh if you're rendering
- 00:10:11frames at a high frame rate uh every
- 00:10:14millisecond counts however they also
- 00:10:17describe in this process that in a if
- 00:10:20you're rendering a larger scene
- 00:10:22including using like Ray tracing and
- 00:10:24path tracing uh there there's you know
- 00:10:27the compute time that's happening for
- 00:10:28those other processes so during that the
- 00:10:32rest of the computation of the frame you
- 00:10:35know this extra 1 millisecond can kind
- 00:10:37of be hidden by happening in parallel to
- 00:10:40those if I'm understanding it correctly
- 00:10:42meaning it might not actually slow you
- 00:10:44down a millisecond for your frame times
- 00:10:47uh because other things would be taking
- 00:10:49at least a millisecond as well and this
- 00:10:51could kind of be computed at the same
- 00:10:53time as some other things like the ray
- 00:10:55tracing for example that's at least how
- 00:10:58I understand this to be clear I am not
- 00:11:01like like like I'm I'm a little out of
- 00:11:03my depth reading papers like this like I
- 00:11:05have a math degree but not specifically
- 00:11:07in like computer graphics and this style
- 00:11:09of stuff anyway uh so that's that so the
- 00:11:13abstract for neural uh textures uh
- 00:11:16neural neural compression of textures is
- 00:11:18the continuous advancement of photo
- 00:11:20realism and rendering is accompanied by
- 00:11:21a growth in texture data and
- 00:11:24consequently increasing storage and
- 00:11:25memory demands so the idea here is to
- 00:11:28have higher resolutions without
- 00:11:30increasing storage and memory demands so
- 00:11:33to address this issue we propose a novel
- 00:11:35neural compression technique
- 00:11:36specifically designed for material
- 00:11:38textures we unlock two more levels of
- 00:11:40detail IE 16 times more texal using low
- 00:11:43bit rate compression with image quality
- 00:11:46that is better than Advanced image
- 00:11:47compression techniques such as avif and
- 00:11:49jpeg XL at the same time our method
- 00:11:52allows on demand realtime decompression
- 00:11:55with random access similar to block
- 00:11:58texture compression on gpus enabling
- 00:12:01compression on disk and memory the key
- 00:12:03idea behind our approach is compressing
- 00:12:05multiple material textures and their M
- 00:12:08map chains together and using a small
- 00:12:10neural network that is optimized for
- 00:12:12each material to decompress them finally
- 00:12:15we use a custom Training implementation
- 00:12:17to achieve practical compression speeds
- 00:12:20whose performance surpasses that of
- 00:12:22General Frame Works like Pi torch and uh
- 00:12:25by an order of magnitude so that's a
- 00:12:28general idea all my sources will be
- 00:12:30linked in the video description so those
- 00:12:31of you more technically minded uh could
- 00:12:34dive into some of the additional details
- 00:12:37about that now again that's from
- 00:12:402023 so uh not only has it given more
- 00:12:43time to potentially get this into at
- 00:12:45least one video game potentially uh
- 00:12:48there's also like I said this update for
- 00:12:502024 I've seen real-time neural
- 00:12:52appearance models so this seems a bit
- 00:12:54more advanced uh so this is where I was
- 00:12:57showing you that that you know uh
- 00:12:59diagonal uh side by side at the
- 00:13:01beginning here this is the neural
- 00:13:02material idea so let's go ahead and read
- 00:13:06at least the abstract of what's going on
- 00:13:09here so they're saying we present a
- 00:13:10complete system for realtime rendering
- 00:13:13of scenes with complex appearance
- 00:13:16previously reserved for offline use
- 00:13:19meaning you couldn't do this in like
- 00:13:20real-time rendering uh this is achieved
- 00:13:23with a combination of algorithmic and
- 00:13:26system level Innovations our appearance
- 00:13:29model utilizes learned hierarchical
- 00:13:31textures that are interpreted using
- 00:13:34neural decoders which produce
- 00:13:36reflectance values and important sample
- 00:13:38directions to best utilize the modeling
- 00:13:41capacity of the decoders we equip the
- 00:13:43decoders with two Graphics priors the
- 00:13:46first prior transformation of directions
- 00:13:48into learned shading frames facilitates
- 00:13:51accurate reconstruction of mesoscale
- 00:13:53effects the second prior a microfacet
- 00:13:56sampling distribution allows the neural
- 00:13:58decoder to perform important sampling
- 00:14:01efficiently the resulting appearance
- 00:14:03model supports an I don't know how to
- 00:14:05pronounce this word guys anisotropic
- 00:14:07sampling I know I see it in graphic
- 00:14:09settings menus but I don't always I I'm
- 00:14:11not confident how to pronounce it anyway
- 00:14:13and level of detail rendering and allows
- 00:14:16baking deeply layered material graphs
- 00:14:18into a compact unified neural
- 00:14:22representation by exposing Hardware
- 00:14:24accelerated tensor operations to Ray
- 00:14:27tracing shaders we show that it is
- 00:14:29possible to inline and execute the
- 00:14:32neural decoders efficiently inside a
- 00:14:35real-time path Tracer we analyze
- 00:14:38scalability with increasing number of
- 00:14:40neural materials and propose to improve
- 00:14:43performance using Code optimized for Co
- 00:14:46for coherent and Divergent execution our
- 00:14:49neural material shaders can be over an
- 00:14:52order of magnitude faster than non
- 00:14:54neural layered materials this opens up
- 00:14:57the door for using film qual quality
- 00:14:59visuals in real-time applications such
- 00:15:01as games and live previews so this if if
- 00:15:05if a lot of this your eyes were glazing
- 00:15:07over um I think in the end you're like
- 00:15:09okay so what is the point of any of that
- 00:15:11what am I supposed to take away so the
- 00:15:12idea here is that the neural material
- 00:15:15shaders can be can be right can be not
- 00:15:17is always right over an order of
- 00:15:19magnitude that's a lot faster than non
- 00:15:23neural layered materials and that this
- 00:15:26would open the door for using film
- 00:15:28quality visuals in realtime applications
- 00:15:30such as games and live previews um okay
- 00:15:35so that's that now the idea here again
- 00:15:38they they talk about the neural B brdf
- 00:15:40model I don't think we need to try to
- 00:15:42make sense out of this diagram because
- 00:15:43you guys are probably not the ones
- 00:15:45programming this stuff um but here's
- 00:15:48kind of the idea here so so if we look
- 00:15:51at this diagram we see a scene with four
- 00:15:54materials that we approximate using the
- 00:15:57proposed neural BRD F the first three
- 00:16:00columns correspond to different
- 00:16:02configurations of the brdf decoder from
- 00:16:04fastest to the most accurate so it's uh
- 00:16:08uh fastest on the left getting more
- 00:16:10accurate right from going from 16 uh 16
- 00:16:13neurons to 32 neurons to 64 neurons but
- 00:16:17this time also going up to three layers
- 00:16:19instead of two layers on the first two
- 00:16:21then they have the ground truth
- 00:16:23reference on the right hand side
- 00:16:26um so uh then they have flip error
- 00:16:30images in the corners timings quantify
- 00:16:34the cost of rendering a one spp image of
- 00:16:36the scene we show images with 8,192
- 00:16:40samples to suppress path tracing Noise
- 00:16:44Okay so in other words what does it mean
- 00:16:45about the the the flip image so you guys
- 00:16:47see in the in the corners here it might
- 00:16:49be a little hard to see in this video by
- 00:16:50the way if you're watching on a phone
- 00:16:51you can pinch to zoom on the YouTube app
- 00:16:53anyway uh you can see that so this is
- 00:16:56basically like a a a map where the more
- 00:16:59stuff is showing up here I think is
- 00:17:00being uh showing up as error when
- 00:17:03compared to the ground truth image so
- 00:17:05you can see that the um the faster ones
- 00:17:08have more error and there's less error
- 00:17:11showing up on the more accurate ones you
- 00:17:14can see the cost in milliseconds of
- 00:17:17rendering time so here we have 3.15
- 00:17:20milliseconds and then you get a lot more
- 00:17:22accuracy or in other words a lot less
- 00:17:24error uh going up to 32 neurons and only
- 00:17:27going up to 3.71 Mill seconds compared
- 00:17:29to
- 00:17:303.15 uh but then jumping up to 64
- 00:17:32neurons with three layers increases all
- 00:17:34the way up to 6.31 milliseconds however
- 00:17:37their ground truth reference uh was a
- 00:17:40total of 13.25
- 00:17:42milliseconds so all of that sounds like
- 00:17:45a lot of milliseconds to me so then the
- 00:17:48idea is so so how are you able to get
- 00:17:51this running uh you know you know fast
- 00:17:53enough for a video game or live preview
- 00:17:56which is what they mentioned uh so again
- 00:17:58again if we go back to the abstract
- 00:18:00where a lot of this might not have made
- 00:18:02sense um the idea is is of how they're
- 00:18:07getting it running is talking talked
- 00:18:08about right here so by exposing Hardware
- 00:18:12accelerated tensor operations to rracing
- 00:18:15shaders we show that it is possible to
- 00:18:19inline and execute the neural decoders
- 00:18:22efficiently inside a realtime path
- 00:18:26Tracer so again while a lot of this is
- 00:18:29also over my head so you're not the only
- 00:18:31one a little confused here as I'm trying
- 00:18:33to read that what I'm kind of hearing is
- 00:18:37that this can happen so if you're doing
- 00:18:38this inside of a path tracing situation
- 00:18:41again the path tracing takes time to
- 00:18:43compute as well so in other words you
- 00:18:46wouldn't necessarily by running it in
- 00:18:48line uh right to show that it is
- 00:18:50possible to in line and execute the
- 00:18:52neural uh uh decoders efficiently inside
- 00:18:55the real-time path Tracer I'm reading
- 00:18:57that as you don't
- 00:18:59add these these frame time costs to the
- 00:19:03uh the rest of the frame just as a sum
- 00:19:05right if some of this is happening in
- 00:19:07line I think that means like at the same
- 00:19:09time so the path tracing itself has a
- 00:19:12frame time cost and perhaps and this has
- 00:19:15a frame time cost but the cost of doing
- 00:19:17both together isn't necessarily their
- 00:19:20sum if it's happening at the same time
- 00:19:23and not necessarily um competing for the
- 00:19:26same resources on the GPU if if that
- 00:19:29makes any sense so in other words uh
- 00:19:33whether or not this is actually what
- 00:19:35in3d is referencing uh on their CES
- 00:19:39presentation here this is definitely
- 00:19:42research that Nvidia is doing so I don't
- 00:19:46know if this is coming to games in 2025
- 00:19:48but I have a feeling we are going to see
- 00:19:50discussion of these types of neural
- 00:19:52rendering techniques at CES
- 00:19:552025 and that's why it's showing up here
- 00:19:57as neural rendering capabilities
- 00:19:59revolutionizing how Graphics are
- 00:20:01processed and displayed on uh on this
- 00:20:04post here okay now let's go ahead and
- 00:20:09move along to some additional leaks so
- 00:20:12videoc
- 00:20:13cards.com caught Acer posting their
- 00:20:17Predator origin 7,000 series gaming PCs
- 00:20:21with their 5090 and 80 listed so if
- 00:20:25there was any doubt now that the 5090
- 00:20:28would have 30 2 GB of gddr 7 memory Acer
- 00:20:31should have now just completely removed
- 00:20:34any doubt also if there was any doubt
- 00:20:37that the 580 would have only 16 GB of
- 00:20:41vram and it is gddr 7 again Acer has
- 00:20:45removed any doubt so it's basically a I
- 00:20:49would consider it pretty much 100%
- 00:20:51confirmation maybe we'll call it 99% I
- 00:20:54guess Acer could have just put up some
- 00:20:56placeholder text here but that doesn't
- 00:20:58seem like what this is right um they're
- 00:21:00getting their products ready to unveil
- 00:21:02for CES it's on their website goodjob
- 00:21:04video cards.com for catching them all
- 00:21:07the sources will be in the video
- 00:21:08description so a 80 with 16 gbt what do
- 00:21:11you guys think about that like I I've
- 00:21:13seen a lot of comments on my channel uh
- 00:21:16pretty upset about the idea of a 580
- 00:21:18with 16 GB personally I'm more upset
- 00:21:20that the 470 is rumored to have 12 and
- 00:21:24that the 4060 is rumored to have eight
- 00:21:27because eight is definitely a problem
- 00:21:29for rendering maximum texture settings
- 00:21:32in a lot of games right now today and 12
- 00:21:35is fairly borderline and there are some
- 00:21:38examples of games spilling over that now
- 00:21:4216 GB on the other hand in realistic
- 00:21:46gaming situations that don't involve a
- 00:21:48whole bunch of modding and a b or other
- 00:21:50crazy stuff I haven't seen 16 GB being
- 00:21:53an issue however a 580 is a very
- 00:21:58powerful graphics card that could
- 00:21:59theoretically be used many years into
- 00:22:02the future in which case uh I could see
- 00:22:06it still being a viable GPU when we get
- 00:22:09you know the next generation of consoles
- 00:22:11even uh at which case if those consoles
- 00:22:14come with an increased uh you know Ram
- 00:22:15capacity compared to current gen you
- 00:22:17know it's like vram usage tends to go up
- 00:22:20over time and 16 gbes is still quite a
- 00:22:23bit uh it's not immediately presenting
- 00:22:26massive concerns to me however um the
- 00:22:30longevity of a card this powerful uh
- 00:22:32being potentially impacted by having 16
- 00:22:35rather than something like 24 or 32 is a
- 00:22:39bit frustrating especially when you when
- 00:22:41you throw price into this right because
- 00:22:44if the 5080 is at least
- 00:22:46$11,000 or maybe even uh maybe uh even
- 00:22:50$1,200 $1,300 may we don't know the
- 00:22:53price right but depending on the price
- 00:22:57you know it starts to feel like penny
- 00:22:58pinching cuz could you just do a uh a
- 00:23:01clamshell design and get 32 GB on here
- 00:23:04for the sake of um you know longevity uh
- 00:23:08and just uh okay if you're spending over
- 00:23:10$1,000 here how much more does 16 gab of
- 00:23:13vram cost uh in the grand scheme of
- 00:23:15things like it feels like it could be on
- 00:23:17there if Nvidia wanted it to be right
- 00:23:21but they don't want it to be and this is
- 00:23:22confirmation that the 80 only has 16 it
- 00:23:26is gdr 7 though and how fast is it well
- 00:23:29this is I would say a less solid leak uh
- 00:23:33but this leak is pointing to 30 GB per
- 00:23:35second sorry not gigabytes gigabits per
- 00:23:38second there's a difference there times8
- 00:23:40big difference okay um and this is again
- 00:23:43coming from video cards.com but they're
- 00:23:45getting the information sourced from
- 00:23:47bench life the translation is according
- 00:23:51to our sources that being bench life
- 00:23:52sources at this stage in addition to the
- 00:23:55GeForce RTX 5080 using 30 GB per second
- 00:23:57ddgr 7 memory models including the
- 00:24:00GeForce RTX 5060 series are also
- 00:24:02configured with 28 GB per second gdr 7
- 00:24:05memory so now video cards.com has this
- 00:24:08helpful table kind of summarizing the
- 00:24:10information we've had leaked so far uh
- 00:24:13regarding these gpus and their memory
- 00:24:15speeds uh and just other specs that
- 00:24:17we've had leaked previously so it's
- 00:24:19really starting to look like this is
- 00:24:20where we're at with things again my
- 00:24:22biggest issue with vram is here on that
- 00:24:255070 um this 5070 TI could poten be
- 00:24:28interesting if it's uh at the right
- 00:24:31price but if it's super expensive
- 00:24:33that'll put a damper on that and the
- 00:24:355090 does look like an absolute monster
- 00:24:38especially compared to the rest of the
- 00:24:39product stack uh since the 5080 looks to
- 00:24:41be like half of it now does that mean
- 00:24:43it's half as fast especially if it has
- 00:24:45slightly faster memory speed compared to
- 00:24:46the 90 um you know we'll have to see
- 00:24:49what what actually Bears out in actual
- 00:24:51gaming
- 00:24:52performance now uh in yesterday's video
- 00:24:55I discussed the fact that there were
- 00:24:57reports that uh nvidia's new app was
- 00:25:02reducing gaming performance and there
- 00:25:04were charts available testing this from
- 00:25:06Tom's Hardware however as I mentioned in
- 00:25:08that video uh they had the app open but
- 00:25:11they didn't try out like which specific
- 00:25:13settings in the app or what if the app's
- 00:25:15closed what if you turn off overlays all
- 00:25:17of that kind of stuff since then uh
- 00:25:20we've had a couple of things one is
- 00:25:22we've had Nvidia themselves state that
- 00:25:25they are aware of a reported performance
- 00:25:27issue related to to game filters and are
- 00:25:29actively looking into it you can turn
- 00:25:31off game filters from the Nvidia app
- 00:25:33settings features overlay game filters
- 00:25:35in photo mode and then relaunch your
- 00:25:37game and we've also had Hardware unboxed
- 00:25:40uh do a video testing this out and again
- 00:25:43they have confirm that it's not the
- 00:25:45entire app that's causing issues it
- 00:25:48appears to just be the photo and filters
- 00:25:52uh needing to be set to off in some
- 00:25:54games like cyberpunk there's a small but
- 00:25:57measurable performance difference uh in
- 00:26:00some other games the performance
- 00:26:01differential is much much higher uh for
- 00:26:05example in Hogwarts Legacy uh you can
- 00:26:08see a a much more significant
- 00:26:10performance uh uh uh differential with
- 00:26:14the with the filters turned on anyway uh
- 00:26:18so that confirms this and again the
- 00:26:20performance hit uh changes based on the
- 00:26:22game you could look at Hardware unbox
- 00:26:24video which will be linked in the video
- 00:26:25description if you want to see all of
- 00:26:27their testing
- 00:26:28uh basically what you should take away
- 00:26:30from this is if you have the Nvidia app
- 00:26:31uh installed you should probably turn
- 00:26:33off photos mode SL filters uh unless
- 00:26:37you're specifically using that feature
- 00:26:39at that time you also require a full
- 00:26:41game restart in order for the
- 00:26:43performance to go back to normal now
- 00:26:45we've seen a benchmark leak for AMD
- 00:26:48ryzen Max Plus 395 stricks Halo now this
- 00:26:52is interesting because stricks Halo is
- 00:26:54supposed to be a ridiculously massive
- 00:26:57APU U when it comes to the GPU inside of
- 00:27:00it it's looking like a um a 40 compute
- 00:27:04unit uh GPU according to the the leaks
- 00:27:08which means you could have some
- 00:27:09significant gaming performance on an
- 00:27:12integrated uh Graphics based laptop mini
- 00:27:15PC that kind of thing now uh it's
- 00:27:18looking like Asus will be launching a
- 00:27:21laptop with this chip and it's looking
- 00:27:26like we have seen this Rog flow z13 show
- 00:27:30up on uh a benchmark here so if we take
- 00:27:33a look at this Benchmark we get single
- 00:27:35core and multicore performance in other
- 00:27:37words it's a CPU Benchmark oh no I
- 00:27:40wanted to see what the GPU could do
- 00:27:43however that being said it does have
- 00:27:45some confirmation of the part here we
- 00:27:48can see the uh AMD
- 00:27:50ryzen um uh AI Max uh plus 395 with
- 00:27:55radon 860s look at that radon 860s is
- 00:27:59apparently what they're calling the GPU
- 00:28:01in this thing uh which is uh you know a
- 00:28:03new naming scheme is this is the low end
- 00:28:06of the 8,000 series integrated Graphics
- 00:28:08you know interesting thing to look at
- 00:28:10there uh but then again we are also
- 00:28:13seeing the um uh the CPU that's inside
- 00:28:16of this is a 16 core
- 00:28:19CPU anyway so we can see some of the
- 00:28:21specs there and all of that and the
- 00:28:23overall Benchmark result is a good
- 00:28:26result for the CPU
- 00:28:28uh apparently coming out significantly
- 00:28:31faster than a um 7945 HX which scores
- 00:28:372736 single core and
- 00:28:3915896 multicore in the same uh geekbench
- 00:28:42test uh which is
- 00:28:45interesting um but anyway and it also
- 00:28:47looks like it beats the 3D vcash version
- 00:28:49of the part which scores 16393 so it
- 00:28:52looks like the CPU is going to be uh
- 00:28:54very effective but then also again I'm
- 00:28:57really interested what this GPU is going
- 00:28:59to be able to do with 40 compute units
- 00:29:01that is absolutely massive uh compared
- 00:29:05to what we've seen in previous uh mobile
- 00:29:08chips we've seen 16 compute units as
- 00:29:10kind of the top end there so far now a
- 00:29:13lot of times uh integrated Graphics seem
- 00:29:15constrained by system memory so uh I
- 00:29:19don't know I'm just curious how that
- 00:29:20ends up all playing out also the Intel
- 00:29:23Arc b570 has apparently shipped out
- 00:29:26review units to at least one reviewer
- 00:29:29this being Andreas sheiling saying uh
- 00:29:32what have we got here the Intel Graphics
- 00:29:35Arc b570 from ASRock arrived earlier
- 00:29:38than expected now again that doesn't
- 00:29:40mean that the uh benchmarks will be uh
- 00:29:43allowed to be posted anytime soon um but
- 00:29:46maybe it's indicating that U Intel could
- 00:29:49be thinking about getting this
- 00:29:51information ahead of any big competitor
- 00:29:54announcements at CES so maybe uh despite
- 00:29:57the card's release date being a while
- 00:29:59out maybe they're trying to get
- 00:30:00information out there sooner than um uh
- 00:30:03than the Nvidia new GPU announcements
- 00:30:06now uh I've tested out Indiana Jones in
- 00:30:08the great circle on this channel I want
- 00:30:10to mention that update 2 is out bringing
- 00:30:12support for gpus that have less than 8
- 00:30:14gbes of vram because previously they
- 00:30:16were unsupported this game is very vram
- 00:30:18hungry but apparently a little bit less
- 00:30:21so now with less than 8 GB of
- 00:30:23vram um and Global illumination
- 00:30:27improvements to the console versions if
- 00:30:29you actually read the full system p uh
- 00:30:32uh patch notes uh you get PC specific
- 00:30:34fixes including solving some dlss
- 00:30:37related issues uh specifically also uh
- 00:30:40Nvidia frame generation had issues with
- 00:30:43HDR support and there's still a known
- 00:30:47issue that Nvidia frame generation may
- 00:30:49be temporarily disabled if HDR is
- 00:30:51activated for the first time to work
- 00:30:54around this issue you would disable and
- 00:30:55then reenable dlss and that only needs
- 00:30:58to be done once and then that should be
- 00:31:00solved there's some other issues uh kind
- 00:31:02of solved here as well and they also
- 00:31:05fixed an issue where Global illumination
- 00:31:07may be completely disabled when playing
- 00:31:09with below minimum vram graphics cards
- 00:31:11so previously if you had below 8 gabes
- 00:31:14of vram the global illumination would
- 00:31:16just be disabled so you wouldn't get any
- 00:31:18real lighting since all the lighting in
- 00:31:20the game was based on the ray tracing uh
- 00:31:23last thing I want to mention is that The
- 00:31:24Witcher 4 devs have confirmed gwent so
- 00:31:27if you uh concerned that the new gwent
- 00:31:30expansion might not actually have gwent
- 00:31:32they have confirmed gwent so there's
- 00:31:35that I hope you guys enjoyed today's
- 00:31:37video and uh uh I hope that you have an
- 00:31:41excellent day that's what I was thinking
- 00:31:43uh don't know why I struggled to say it
- 00:31:45hope all of you have an excellent day
- CES
- RTX 50 Series
- Neural Rendering
- Acer
- Nvidia
- AMD Strix Halo
- VRAM
- GDDR7
- Intel Arc