[SPON: New Customers get Creality Ender 3 V2 3D Printer for $99: micro.center/9e28b0 ] [SPON: FREE 128GB USB when you Sign Up for Early Access to Miami: micro.center/1lii ] [SPON: Shop Micro Center’s Top Deals: micro.center/e2f1a6 ]
@@benpedi6645it could be enough until Nvidia releases 6080 with 24GB of VRAM, the next generational gap is going to destroy many RTX 50 series durability, I wouldn't buy any of those card, it's going to be one of the worst generation ever
if the performance is better at comparable price points then who fucking cares if it's a 96-bit bus people get so fussy about it but ada got more performance with narrower bus widths than ampere did, if they can do that again and still provide better price/perf and ACTUALLY give more VRAM then sure, fuck it, that's awesome and before someone is like "you will have smaller bus and you will like it" or something stupid like that, the reality is that the specifics of the bus widths DO NOT MATTER what matters is the final performance you get from the card. in an idea world every GPU would have a bazillion terabytes of bandwidth but in reality each given tier only needs so much bandwidth and what matters is the final performance not the numbers on a spec sheet. one of these is how your game plays the other one is a number on a spreadsheet that you wank to.
@@richardhunter9779 right because ada GPUs clearly have miserable, irredeemable 1% lows in benchmarks and -- wait, what's that? they don't? hmm, curious
@@ElysaraCh People confuse where the bad performance of low memory bus comes into play. Gaming is not affected the most (with the more advanced cache technologies, it's still mostly fine), compute is. Low end Ada has been struggling A LOT compared to the comparable Ampere products in compute tasks, because of much lower memory bandwidth. 4070 and higher are fine, but they're not the huge improvements over 3070 and up we should've expected from a new gen. Only incentive to get Ada for compute is the much improved power efficiency.
With each module having 32 bit interfaces I wouldn't think that'd be possible but then this is nVidia and the GTX 970 comes to mind with it segmenting the last 512mb of VRAM to make it useless ;)
In 2025 anything under 16gb is a bad idea, in rdr2 with hwunboxed optimized settings my vram buffer get to 12gb after an hour of playing and that's a game from 2018! What most people don't realise is that vram gets filled when you play and a simple benchmark run won't show you the vram requirements of a game
@@MegaLoquendo2000 what is meant by 'soyface' is wide gaping mouth while smiling, or some other overly-exaggerated expression, like the O-face. it makes you look ridiculous.
The reason for the recent price increases for all PC components is because of Prime Day and Fantastech sales. The retailers always bump up the price before a sales event to "magnify" the savings. What I'm currently interested in is how the upscaling and frame generation features help with power efficiency. Would be nice if someone can do an in depth video on the settings and best cases for using upscaling and frame generation to improve frame per watt.
From my own personal testing if you lock the framerate you get a pretty decent ~15-25% power reduction. Like a 4090 goes from ~110w to ~85W. Depending on the scene it can be as much as 40% but that's super rare. Not sure on FSR version but it should theoretically be a little worse. Would be best to test it on something like a 1080ti but I don't have one anymore. Not a huge boost but definitely big enough to investigate implementing it for battery powered devices.
Yeah and presumably most people fall for it. I also see how they alternate between having the discount already applied as "X% off!" or they offer a "coupon" to lower the price to essentially the same level so you somehow feel like you're getting a good deal. Happens with the Minisforum PCs on Amazon, they never really go down in price.
Running BG3 at ultra at 2K without DLSS for a 4070 Ti Super + 5700X3D resulted in my UPS supply beeping nonstop due to the power draw exceeding 300W even in environments with relatively lower numbers of objects. With DLSS resulted in occasional beeps when environmental scenes had higher performance demands. Getting rid of the UPS fixed the beep though and also humbled me greatly as it took way too long to figure out what was causing the problem 😔
(XTX owner here) I expect AMD to launch the 8800XT at $700 or something stupid, and then reduce it to $550 once all the reviewers have ripped it completely appart.
If they're dumping 7900xt while raising xtx prices, that's all u need to know about top end RDNA 4 being between the xt and xtx in performance, (confirming earlier rumors)
I just bought an ASRock Taichi 7900xtx for $950 last month. If the prices are going up, then I might have got it at the right time. 😅 Probably won't buy another gpu for at least 4-5 years. (Took me 7 years to upgrade from my 1070)
Won't the top end rdna4 still beat the xtx overall since it's supposedly getting a big RT uplift? Doesn't make sense they would price it under the xtx in that case. Someone else was saying they raised it so they could "slash prices" for upcoming prime day and that makes more sense for me
As long as Lenovo keeps using unified chassis between intel and amd, amd is going to be a killer product thermally speaking, the amd lenovo I’ve had run crazy cool by comparison because intels are space-heaters which means the thermal solutions are crazy good for AMD processors.
It's seriously hilarious how people legitimately have -19000 understanding of how much ram and VRAM these modern programs and games actually use. Its "allocating" that much for a reason people. It's not just some arbitrary amount the developers chose throwing darts at a board. All those meshes, textures, and polygons moving around aren't nothing kiddies 😂
Yes and no, there are absolutely titles that allocate far more than they really need if the capacity is available. Game optimization went out the window a long time ago.
Winding down 7900 XTX seems weird IF RDNA 4 really has nothing that matches or exceeds it...unless it's to keep XTX prices up. Or your theory that they care more about using that fab capacity for something besides consumer graphics.
Early rumors were that the 7900XTX might get sold alongside RDNA4, possibly as a refresh, but if RDNA4 performs close enough to it while having a large enough RT uplift then keeping the 7900XTX at all could be considered counterproductive. Tom was speaking kinda fast as he speculated on it but remember the 7900XTX and 7900XT are being made on the same chip, Navi 31. He’s guessing they could be trying to get through as much leftover volume of Navi 31 as they can by putting them all in 7900XTs (and GREs). Then the number of 7900XTXs dwindles and that may be what’s driving XTX prices up while everything else is staying cheap. The big picture is AMD needs to sell as much remaining inventory of 7000-series cards as they can in time for RDNA4’s release. That they are doing this now suggests maybe we will see RDNA4 before the end of the year after all.
@@tringuyen7519 since we’re speculating anyway, another possibility is AMD may be hoarding top Navi 31 chips, saving them for that long-ago rumored 7900XTX refresh (aka 7950XTX?) after all. It would be another good reason to not bother keeping up regular 7900XTX inventory. It’s old news AMD hoped to be able to get them to run closer to the 4090 in raster than ended up happening, but some people allege to have been able to OC certain AIB 7900XTXs pretty close. To me a 7950XTX primarily makes sense if the anticipated 8800XT or whatever doesn’t have that big of an RT uplift. Which to be completely honest I don’t expect that much for RT this gen anyway. A lot of people got overly-optimistic about the RT uplift anticipated for RDNA3 as well, despite leaks pointing to fairly moderate uplift and what we got was in fact pretty incremental. I just think a lot of people would be best served keeping their expectations humble for RDNA4, and if something happens to exceed our expectations then hey, great.
Depends on gimick, they create new magic AI core units and you will have DLSS 4.0/5.0.... I wait for 5080 or 8800xt (depends which will come first and perf/price/power usage ratio...
I've said this for months... the export restrictions on 4090 class performance pretty much guarantees they'll kneecap the 5080 and make up for it with more frame gen nonsense and efficiency claims.
@@adamb.834 I use the same trick... i play on a 1070 Ti and wait for Navi 48 256 bit bus GPU so their max product or 5080 RTX ... if Nvida card for 1000-1200€ has only 16gb VRAM and Navi 48 has so much vram like 7900XT and cost max 650€ i am on AMD train than. Even if Navi 48 is 25% slower than 5080 RTX.
I’d think the best way to judge the switch 2 power is to look at the best games on switch like Mario kart 8, Zelda tofk, Metroid prime remake etc and imagine what can be done with a bit more perforce than a ps4. That’s quite a jump in power
@dante19890 His analysis about the Switch 2 and Steam Deck's CPUs also doesn't make sense. Sure, the Deck's cores may be faster - which would hurt Switch 2 in _single thread_ situations... ...but he's also saying Switch 2 will have more cores and threads, with two BIG differences he didn't mention being that it will be ARM instead of x86 and will be paired with faster RAM than the Deck. That becomes less of a _raw power_ issue and more of an _optimization_ issue. A skilled developer could offset the single thread weakness by shifting work to multithreaded + use Nvidia's features to have Switch 2 easily beat the Deck both visually & FPS wise.
@@mitchtellers2547exactly, the mere fact that cpu is octa core beats the steam deck, games would benefit much more from way better mutlithread/multicore perfomance than just slightly better single core perfomance (and even worse in some cases as steam deck cpu has variable clocks), i'm counting on his """"leak""" based on a 2ghz cpu, cuz a78 has a 3.3ghz version for tablets, which is on par with xbox series x cpu
@@Elchinodawn I'm thinking that in portable mode, the two CPUs will ultimately be on par with each other when all of my points are taken to account. Benchmarks will trade off of each other based on how well devs optimize their games. But Switch 2 will really have a leg up with its GPU. That'll be the difference maker.
a 5060 with 8Gb would be a disaster. Even for 5050 it would be too low. I honestly feel like it's about time we start with the 12Gb or at the very least 10Gb... for '50 series cards... 10 Gb 5050 actually seems perfectly fair. And 12 Gb 5060 . 14 Gb 5070 ,and 16gb 5080. That's a bare minimum in my opinion.
anything less the 16gb is a problem for the 5070 , 5060 - 12 gb 5070 - 16gb 5080 - 20-24 gb 5090 - 32gb . Vram costs a few $ per chip , and max 10-15$ for the buss size increase . There is not a single reason they should not have 12 gb for low end , and 16 gb for mid end .
@@snakeinabox7220 again. I would personally never accept a new card with less than 16gb. So that's a bare minimum for me. But I was talking more in terms of how much they can screw people and get away with, before it becomes an outrage and boycotting the company... If it was up to me I would be giving the maximum possible amount of vram since about mid range... Actually this issue is personal to me, because I was waiting to upgrade from my 1080 ti, for so many years now... and proving even 5 years ago, that I needed more than 11gb was difficult. Thank god people are starting to realize that. I'm glad that my 5 years of battle of proving to people that we need more vram had worked.
GDDR7 isn't the massive upgrade it seems. The first GPUs with GDDR7 will run at 24GBps speeds which is just 14.3% faster than GDDR6X's 21GBps. That is why the 5090 is going 448-bit to get a 33% bandwidth increase. As for the rest of the lineup, bandwidth isn't as big of an issue as memory bus width and capacity, and with RTX4000 series Nvidia has set 128-bit as upper midrange, 192-bit as high end and 256-bit as extreme high end. It still blows my mind they were selling a 128-bit 16GB card for $500 and 192-bit 12GB card for $600. Cheapest 256-bit card was $1200 at launch - how tf do you even justify that...
I'm actually not that keen to upgrade my PC after getting the RX 7900XTX on launch. For normal gamers, this card is already beast and can perform well on 4K monitors for most games on the market. I can see myself sticking to my main PC for at least 6~7 more years before even considering another upgrade. Currently I'm more interested on AMD's APU scene though. Recently when I sent my main PC back for cleaning and regular maintainence, I did got myself into a situation where I'm stuck with completely losing a workstation to support me during commuting. So if AMD's APU can manage to chase the performance of a 3060 or 4060 series card and can be stuck into a mini PC case, it is something I can consider to build as a back up system for my main computer. I'll even consider to switch some of the work to that mini computer to save some electric bills it they can make it possible. I've recently realized most of the games I play can perform relatively well on 60 series cards if I'm willing to cut back the effects and if it is just on a subsitute system, I can just swallow it for a few days before the main system comes back.
@@herbertwalter8693 Believe me, if Nintendo managed to force gamers to play with a Tegra for over half a decade, the headroom for that 7900XTX is there for quite a long while. lol Just kidding, but after switching to PC gaming, I did realize that gamer does have a certain tolerance level as long as the game is good.
The whole Switch 2 info doesn't make a lick of sense. It is well known from the Nvidia leak that the T239 SoC has 12 Ampere SMs (1536 CUDA cores). To run those at 800Mhz means it's pushing 2.4 TFlops of GPU power. That is 50% more than what Steam Deck pushes to its max in terms of raw power. And that component is expected to consume 5W in portable mode and the system be fanless? Even if 5W is for 400Mhz, it makes even less sense because pushing to 800Mhz would mean at least 10W by that same notion, and there's no way a fanless setup would work. As it is, Samsung 7nm is inferior to TSMC 7nm used by devices like the Steam Deck, which has lesser performance but also requires a fan. You make mention that the CPU will be weak, but in various Linux commits that reference the T239 from the Nvidia leak say the CPU will have 8 cores on a single cluster. The only CPU that fits that description is the ARM Cortex A78C, which ARM themselves have stated to be used for high-performance laptops. Even if one were to say it would be clocked low (what is considered low anyways?), there is a situation that people aren't taking into account with Steam Deck. Anyone that has tested it with the GPU maxed out will also notice that the CPU gets throttled down to no higher than 2.4Ghz because it hits the 15W limit of the chip. It's a 4-core CPU, but does have 8 hardware threads. This is through SMT where 2 threads share the same resources, but is only a 25% boost because those resources are being fought over most of the time. When two threads want access to the same resource, one has to wait. So, a 4c/8t CPU clocked at 2.4Ghz is equivalent to around an 8c CPU clocked at 1.5Ghz when all 8 threads are heavily utilized. So, 800Mhz on the GPU for portable mode that is clocked higher than what Switch 1 docked mode set for its GPU, but what would be the expected clock of the CPU? Even at an equal clock, the A78C surpasses Zen2 in IPC. There is absolutely no reason why Nintendo with a custom chip would push the GPU so heavily and not balance it with clocking the CPU enough to keep up with it, especially when they are using a CPU designated for high-performance laptops.
This idiot thinks Switch 2 GPU works like Steam Deck's. That 400-800mhz are small while the 1,6ghz of the Steam Deck are good. But completely ignores that T239 has 1536 cores while the Steam Deck has just 512 cores. He is just sooo dumb.
i don't see why the switch 2 would be passively cooled in handheld mode, unless they are really focusing on battery life more than switch 1, i hope it atleast has a bigger battery
@@MooresLawIsDeadSHUT UP. You dont know anything about Switch 2. Stop shitting through your mouth. Do you even know how the Switch 2 GPU works??? Stop comparing it with the Steam Deck. Do you know with the clocks you gave it outclasses the Steam Deck because the GPU has 1536 cores while the Steam Deck has 512 fucking cores??? 400-800mhz may seem small for the Steam Deck(reason it is powered at 1,6ghz) but for the Switch 2 it isn't. With that big of a GPU, 8nm or 7nm and passive cooling is just out of the question for it. Unless you want a physically huge chip, a heater and poor battery life. Stop talking about Switch 2 alltogether if you don't even know basic known information about it.
@MooresLawIsDead SHUT UP. You dont know anything about Switch 2. Stop shitting through your mouth. Do you even know how the Switch 2 GPU works??? Stop comparing it with the Steam Deck. Do you know with the clocks you gave it outclasses the Steam Deck because the GPU has 1536 cores while the Steam Deck has 512 fucking cores??? 400-800mhz may seem small for the Steam Deck(reason it is powered at 1,6ghz) but for the Switch 2 it isn't. With that big of a GPU, 8nm or 7nm and passive cooling is just out of the question for it. Unless you want a physically huge chip, a heater and poor battery life. Stop talking about Switch 2 alltogether if you don't even know basic known information about it.
1:31:50 this actually bad analogy. Bronze swords are at par with high quality iron swords what comes to toughness and hardness and bronze swords are actually far superior agains corrosion. Why ironworks was revolutionary technology was not quality of items it was the volume. Production of iron weapons and tools used like one tenth of resources compared to bronze working.
Nintendo did NOT say "they will launch the Switch 2 before the end of this fiscal year". They said they will announce it before the end of the fiscal year.
@@dante19890 this is going to be an announcement, not even necessarily a reveal. That Switch 1 was publicly announced in March 2015, codename previewed at E3 June 2016, officially revealed on Oct 20th 2016, and launched 5 months later on March 3rd 2017. 2 years bro
@@--waffle- fair enough. I think Nintendo doesn't wanna hurt the switch sales by revealing it too early. It seems PS5 pro would benefit launching later this year if switch 2 launching q1 2025
@@dante19890 agreed. Nintendo are scared of another Wii U type console flop. Switch has sold so well that unfortunately, they will have to continue supporting it for many more years. They wanna sell games
I'm on a 6700XT which has been good so far, but I'm itching for an upgrade. I don't care if it's from AMD or Nvidia. All I care for is price to performance. Still reluctant to adopt Intel, but I'm open to it if they continue to improve.
I believe the way to fix this perception of arm being a cure all is for x86 laptops to have the same sleep behavior as android, ios, and arm macos. In my experience, i can never trust windows sleep because it randomly turns on, or have bad standby time.
I disagree on the "upscaling can not be improved algorithmically" part. Advancements in image processing and rendering methods happen all the time and will continue to do so. Integrating AI upscaling into FSR might ba good idea, but there's certainly room for algorithmic improvement too.
Most of these decisions make it sound like AI is gobbling all of the highest speeed GDDR6 and GDDR6X chips up, and their options are bid higher (driving market prices up further) or avoid using it. All of the RDNA3 Skus referenced other than the 7700XT are using 19.5Gbps and 20Gbps memory speeds. Notice the GRE is missing from the report. Could it remain in production, or just become the outlet for which they disseminate remaining N31 silicon? Could there be a flood of 7700XTs with the last bits of N32 silicon? Shutting down production of these SKUs supports the rumor that RDNA4 will not be high-end and will use 18Gbps memory.
Of course Nvidia will skimp on v ram and then they will do a refresh and make the card what they should of been to begin with. We have all read this book more than once before.
The only game that makes me upgrade my PC anymore is Minecraft, so VRAM matters quite a bit with Blackwell. My 4070Ti can just barely handle the textures and shaders I throw at it
@@alexander1055 ray tracing falls under the subset of shaders, and many “non raytracing” shaders still offer many RT driven effects like AO, volumetric clouds, soft shadows etc. I was using a fully path traced shader until I switched to distant horizons, for which currently the best shader available is Bliss (if your system can handle maxing the settings). At those settings it can tax my system as much if not even more than SEUS ptgi.
@@alexander1055 Keep in mind I’m running the game with nearly 60 mods at 3440x1440 with a render distance of 450 chunks (20 chunks full quality + distant horizons chunks). The bliss shaders are using RTAO and flood fill to simulate occlusion and global illumination. Shadows resolution is 4096p for 20 chunks, with the remaining 430 chunks also shaded at 1024p. Volumetric clouds and fog effects that use ray marching. My texture pack is exorbitant with lush bushy trees, lots of detailed bumpy surfaces and material effects. My world build height limit is 2 kilometres…you get the gist lol
@@mickaelsflow6774 The only game that evolves with every generation of PC hardware. The ceiling is insanely high for what can be achieved, I wouldn’t be surprised if we have the same conversation again in 10 years despite having 128 gig vram buffers, 3d stacked dies, internal resolution of 240p thanks to AI overlord upscaling and whatever else the alpha nerds throw at us.
Snapdragon X for general computing use cases has gotten pretty decent reviews so far. Yes, for gaming it sux but one can see where by the time the next iteration of the X comes out, the iGPU could be improved and by then there will probably games compiled for it natively. In the meantime, sales for productivity and web browsing users, plus Microsoft’s support for the platform could it alive. Basically, I would not rule out Windows on ARM in the long run, and neither should anyone want to. Competition is good.
@@sonicboy678 more competition will come out in the future... Well i think it was supposed to because Qualcomm and MS exclusive deal is ending. Mediatek probably will have theirs next year. Samsung probably will try the market as well.
Warning Level 1 Wendel made a video telling how horrible was Qualcomm launch. And what happened behind the scenes and it's a nightmare. MS has an agenda, and Qualcomm paid too much propaganda to buy tech outlets. Qualcomm experience is a mess, something not fixable now. I would stay away from arm for atleast a year so ppl don't waste their money.
From what I've seen it's only ever so slightly faster than my 7840U in CPU and GPU, but it costs a fair bit more, and it doesn't have that much better battery life. That's not a good spot to be in for "just general computing". If you're catering to the "I need a webbrowser and an electronic typewriter" market you need to be price competitive.
You hit the nail on the head here, Microsoft should stay out of everyone business and just get windows to work consistently without all of the bloat ware. I don’t use their cloud software and haven’t even signed into it, but I have had other problems.
I’d take a X Elite on sale for 1/2 off? Not because it’s better than anything else, but because it would be a good development, laptop for arm code. The funny thing about it is the best Windows on arm laptop available right now is probably a Mac using virtualization.
When I bought 7800x3d/X670E-F from Microcenter in VA, I asked them if they sold a lot of these bundles. They said they sell more AM4/Intel stuff then they do AM5. I was surprised because of all the misinfo online. The associate said that they have had a lot of returns of AM5 boards due to instability. I havent put mine together yet but I did get a X670E-F for a hefty open box discount. I hope it isnt unstable. I was able to choose from 5 different open box X670E-F they had in store and I picked the best looking one (it looks brand new and has all unopened accessories). They actually had 20+ open box B650E-F in store as well. Perhaps its an Asus issue but I was shocked by how many open box products they had while they had hardly any Z690/Z790 open box.
The AM5 instability is down to OEMS shoving waaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaay too much voltage into the CPU and it started with AM4. As a fellow AM5 owner, I highly suggest you use voltage suppression to cap the voltage at 1.3x volts ceiling and voltage floor of the lowest value. My 7950X can no longer boost to 5.7 Ghz reliably, at one point I was not even getting reliable 5.4 Ghz.
Nintendo said they will only announce their successor/next console by the end of this fiscal year, NOT that it will come out. So they have given the time frame of up until March 2025 to announce/reveal the system with it launching presumably some months after. So the successor could be announced as late as March 2025 next year with a release as late as fall next year. (Of course we don't know this will be the case) they could in theory just as well announce the successor now and release it this autumn; I just think it's important to quote them correctly, that all they said was that there will be an announcement by the end of the fiscal year; not a release and judging by the very big lineup and roadmap Nintendo has just revealed at it's Direct the other week with it's games for the OG Switch for the rest of this and beginning of next year, it doesn't look like we're getting the successor any time soon. Judging by this Direct and the amount of games still coming, I think we won't get this new system until at least in a year's time from now, so summer 2025. Unless sales of the current systems dramatically drop leading up to Christmas, I don't see them seeing a need to release or reveal anything sooner. (But of course I hope I'm wrong because I would love a new system sooner rather than later but even judging by Nintendo's own comments, 5 years in to the Switch that it's only 'half way in it's life cycle' ,shows that Nintendo is in no rush with how well the current system is still doing.)
I'm genuinely curious how VRAM allocation works in games. I mean just think about it. If 8 GB doesn't cut it for 1080p gaming anymore, how does 4k (driving 4x the number of pixels) get by with 16 GB? That doesn't make sense to me.
it's not really about the amount of pixels, it's the textures and stuff like that. if the game uses 8gb at 1080p and functions well, then there's not a lot of reason why it would consume 4 TIMES that at 4k. I mean think about it, it's not like one pixel consumes 10 megabytes and the entire memory gets overwritten each frame, that's not how it works.
Didn't see any mention in the live stream itself (and not caught up on the latest broken silicon), but anything regarding the alleged 6400mts sweetspot for zen 5? It's quite a bit off the 7000-8000 figure I remember you mentioned over the last year or so. I mean maybe it supports it just like zen 4 supports it, but support and sweetspot are very different things. Anyway, great stream as always, thanks c:
@@_Kaurus Okay, but I'm looking for Tom's take on that leak. Its source appears to be wccftech who claim sources with an engineering sample, and techpowerup basing their coverage on wccftech's article. Once the 6400 sweetspot leak dropped, the 7-8k support just confused me as to why it received any emphasis if ultimately it's a figure of only secondary importance. So the question remains - is that leak accurate? Any chance AMD'll manage to crank the fclk up to have higher memory speed sweetspot? Or is this likely to be the final figure
It's wild to me that we'll probably get a 5060 8gb in 2025 when Xbox will be pushing next gen games in 2026. That means 3rd parties will do the same and their games will probably be less optimized than Xbox's. There's no way games a couple years out like Gears of War E-Day will run even at 1080p medium on an 8gb gpu.
I'm really lost tbh. I wanted to go back in to PC gaming and so i looked up what is going on with CPU's and GPUs. And i have no freaking idea what to buy, to wait for what ever, or just don't care and buy what is on the market. Money does not matter, but i don't want to burn it 😅😂
@@puffyips isn't the 7950x3d the hot shit? And yes obviously the 4090 would be the way to go, but tbh I don't know if I want to switch to a 5090 in 5-7 months.😅 Probably i will test a 7900xtx and switch to the 5090 then, so I have also some mix and experience in the differences between AMD and NVIDIA.
@@ConsiderbeBasedyou said pc gaming so 7800X3D is the only chip you’d want rn, and I didn’t say 4090 cause honestly it’s not needed, 7900xt/ 4070ti super perf will hold you over even in 4k but if you’re going high highend then ig wait
@@puffyips yes i see where you go with your recommendations and thanks for that. The 7800 is insanely good for the price and probably the best gaming CPU to purchase right now. I think i will go with the 7950 so I have some power for rendering stuff extra and honestly in short there will be new CPUs so will see if I switch to a nother when the x3d versions come.
Is anybody surprised ARC is dead? They jumped in thinking they could ride the crypto bubble but it burst before Alchemist came out, the only ones I have seen sing any praises for Alchemist is streamers buying the A380 to transcode their streams while their Nvidia and AMD GPUs do the actual heavy lifting, and with its piss poor support for older versions of DirectX and OpenGL during a time that the Steam top 20 is filled with older titles was just the final nail in the coffin. It is a damn shame as I was hoping it would be a 3 horse race like in the old days with 3DFX, ATI and Nvidia but with Intel taking huge body blows from AMD in everything from desktop to server to laptop spending billions to try to open a new market at this point was just a bad call.
@@dante19890 I think it all depends on how much Switch stock they still have left. If it's a decent amount I could see them doing a small discount for Christmas sales of the Switch to get rid of as much of leftover stock before the Switch2 release. Otherwise I expect that they would want the Switch2 available to be sold in large numbers for Christmas time. But this is Nintendo we are talking about and they don't always do what one would expect them to do. lol
Anyone have insight on how large the RDNA 4 lineup will be? We are hearing that the highest will be a 7900XTX equivalent with better ray tracing. So will they have a 8800 XT, 8700 XT, 8600 XT, 8500 XT? Or are they just going to leave it at two cards? 8800 and 8600?
They typically do three chips, and the top one has been cancelled. So we are probably going to get 8800XT and 8700XT on the now-top die and 8600XT and 8600-non-XT on the small die. If AMD have any brain left they'd do 16GB on the 8800XT, 8700XT and 8600XT (clam-shell), and only put 8GB on the 8600-non-XT. Resisting the urge to put 12GB on the 8700XT could really do them a favour here. Alternatively they could make a 8700-non-XT with 12GB (and a few less CUs) to get rid of any dies that have defects in a memory controller.
Nvidia will probably launch 5000 series just like 4000 series(except 5070ti, that will be 16gb). Then when 3gb vram chips become available, Nvidia will probably launch 5000 super series with increased vram. 5080=16gb/5080super=24gb 5070=12gb/5070super=18gb 5060=8gb/5060super=12gb
This is what I'm expecting for mid-range: 5060 - 8 5060ti - 12 5070 - 16 Personally, I'm buying a 5090 which will be overkill for 1440p, but vram won't be an issue. Also, wondering if Valve will go with Strix for SD2 or wait for RDNA4. Should be pretty big efficiency improvements.
I don't have sources but I am 100% sure 5080 laptop will have 12GB and Nvidia calls the two 16GB mobile GPUs 5090 and 5080 Ti. They have zero competition in laptop so they can milk as hard as they want.
If AMD could launch the 12 core strix with 16 cu mode on desktop for like 330 or 350 dollar then many will buy it in south asia and south east asian region. Even the gamers who will throw away old pc in 2025 when win10 is eol.
I once had that onedrive problem and I swear that if I had had the guy responsable for storing all my files in the cloud by default in front of me he would probably be dead now.
There is no way a 5090 is acceptable at 24GB. I have a 4090 and people that spend that kind of money will not upgrade without a boost in RAM capacity. It needs to be 32GB, if you need more then you go to professional cards. Keep in mind that AI models are getting bigger by the day and 24GB is not enough today. They wouldn't erode the more expensive products because the gap is too great anyways
I've been waiting on the new CPU and MBs, nwo they saying the new CPU will launch before the new boards, well damn, there goes my buying a 7600 and a x870 board til the new 9000x3ds release . So I guess I will jsut keep on waiting since my 5600x 6750 xt are doing just fine, was also hoping to buy a +16gb gpu around the same time, but now they selling through and going up in price smdh AMD releases any new GPU's with less than 16gb and over 600 can shove it where the sun dont shine! Really thinking of grabbing 7900xt before they sell through
Glad I didn't buy 7000 series this year, I was about to pick RX 7800 XT or RX 7900 GRE, because they were coming out below MSRP. I think this is year of Monitors not GPU. Better get GPU's next year
To be fair I saw the 7800 XT for $440 and the 7900 XTX for $800 for a split second...I don't think that's a bad deal. However, neo we are probably a quarter away from the RDNA 4 launch, and I wouldn't pay a cent more than that therefore...
@@Valueshooter it also costs less. pricing is disappointing, but Nvidia is making huge leaps gen over gen, which Intel was not. it's not even comparable
@@zuckdaddy1596 1. It does not cost less, i bought my friend a 3060 $299 bestbuy, BEFORE 4090 came out. don't use crypto mining crazy price as a comparison. 2. #2 reason 4060 is a scam, it is a cut down die compare to 3060, basically nvidia is trying to pass off a 4050 as the new 4060, everyone knows that, so stop lying. so not only are you paying same price for a "4050" its marginally better and sometime not better than the 3060. it is very much comparable to what INTEL did for 10 years. 3. Ask anybody around if the gen over gen improvement vs the price is worth it. unless you never heard of the rtx 2060, 2070, 2080s, be honest the RTX 4000 is the repeat of what nvidia trying to pull over the gamers during the rtx 2000 generations. Its like you never watch any of Moore's law is dead videos, this point has been talked about over multiple videos, only a nvidia fanboy would defend them on RTX 4000 series performance and price anymore, other than the 4090, there is nothing good about 4000 series, and i would add 4090 is overpriced and not very well built. 4. you don't build goodwill by ripping off your fans/customers, by over charging them over and over, INTEL did that for 10 years, look at them now, nobody feel sorry for them.
Do we know that ARM64 emulation is the main cause of issues? Could many of the issues be caused by Adreno DX12 GPU drivers? Intel had a lot of DX12 driver issues too. Vista was considered a buggy mess at launch, and Microsoft revealed that Nvidia’s GPU driver was causing majority of the blue screens. Are we sure that ARM64 emulation is the main culprit? On Android side I have had to write workarounds for Adreno driver bugs. And they have shipped GLES/Vulkan driver for years. It should not surprise anybody that their first DX12 GPU driver has issues. On PC we also have compute centric workloads that are not common on mobile. I would expect improvements to both stability and performance once the GPU drivers mature. Assuming of course that they prioritize PC driver work heavily. Intel didn’t prioritize iGPU gaming drivers heavily in the past. Let’s hope Qualcomm understands the importance right now.
It's not surprising that the 5080 will be gimped considering they've already reached peak performance when it comes to being able to "export" to china and thats a quarter of their income. On top of that it sounds like Radeon is pretty much throwing in the towel on rdna 4 so what motivation do they have to make a whole separate sku for a 5080 ti or whatever with more vram? A 5080 will destroy any game on the market for years to come and anyone that truly NEEDS a 5090 (small market) will cough up the cash. On top of that they're focusing on AI bc their market valuation is through the roof. A couple of aussie dogs barking at the proverbial caravan wont change market dynamics.
As Tom and many others have stated: It will be shorter between Zen 5 and Zen 5 X3D than it was between Zen 4 and Zen 4 X3D. Other than that, nobody has anything solid yet.
It makes sense for Intel in the dedicated gpu department wait until it matures, before releasing a mid tier. So anyone hoping a new release unfortunately it will be igpu related. It's a smart move, launch when ready. Lovers or haters of Intel, as we speak Intel can't afford to launch another series of cards at a loss. The more L's, the more the house burns in fire. I'm hopefull maybe next year around fall, they could announce something, but not guarented.
@@Lucid_XP here are some of the games im gonna play on Ps5: Ghost of tsushima Rise of ronin Demon souls Elden ring + DLC FF 7 remake FF 7 rebirth FF 16 Post trauma Black myth wu kong Wo long Sekiro Lies of P Skydancers behemoth All Uncharted games All Yakuza games Nier automata Stellar blade Enotria The relic Clair obscure expedition 33 Star Wars outlaws This is just some of the games i want to play, the game i want to play the most is MH wilds.
@@webtiger1974PTG Not even close in my country. Just a rx 7600 alone is 500$ 1500$ will get me a 5700X3D And a RX 7700XT Pc. Its better than Ps5 for sure but with my post i meant ill use a ps5 until i can afford a decent Pc Btw a Ps5 disc version is 870$ 🤣
I must be a wizard or to smart for some no plan people ... I say it even 1 year many times 12 gb VRAM is not enough... for gamers. When you play at 1440 you keep your GPU what? 1 year? I buy and keep gpus for 4-5 years. Before buying some new. Even 16 gb VRAM on a 5080 RTX that have a lot of power is not enough for 4-5 years. Its Nvida trick to get paid for all 2 years for their new gpu. I don't need to be a rocket scientist to see that. Raytracing in more games, Upscaling from 1440 to 4k , better looking game better = higher settings and Hardware needed = More VRAM useing ^^. Some people dont know that Unreal 5 Engine dont even shows in a game true power and 16 gb for 1000€or $ gpu is way to expensive. If you upgrade all 2 years a new GPU from now on ok another topic ^^. We have a lot of Techno logic going on the Last 1 years, that looks better, but eats more of your system.
I have a 7900XT, had to RMA my original one because of hotspot issues.. Anyway I'm of two minds with the discontinuing of RDNA3. One side is saying it makes sense if they need the packaging for other things and if RDNA4 is going to come out at the end of the year. The other though knows that they will only have 1 generation of chiplet based GPUs so it will make driver development for the 7000 cards lower priority and they will probably have more issues as time goes on. Yes we will get our 5 years of support 100% but having the almost 10 years that something like Polaris got will probably not happen
The 7600 is not a chiplet design and they are going to continue to make them, possibly for a while. The PS5 Pro *might* be RDNA3 based, at least partially, so there could be more attention given to further optimizations if it doesn't break away too far in its design.
The past amd drivers such. I have 6800 xt. I get Flickr scenen when playing UA-cam and a video appear on Facebook same time. The last driver dint fix it. Had AMD past decades. Buying NVIDIA nexttime.
I had issues with one drive doing that same thing, storing certain game files on the cloud then I boot the game and wait 5 minutes while it individually downloads each file to my machine again, fucking hate one drive. It’s because I subbed to ms 365
might be that rdna4 might not be better than rdna3, so they don't want any rdna3 left to avoid lowering rdna4 sells, like rdna2 did with rdna3 by being really competitve and widly available.
Who knows, that's being very pessimistic lmao. If so Jensen Huang will become the leader of the UN and make 5090 the 196th country of the world. Realistically, all manufacturers want to clear left inventory before new launch.
[SPON: New Customers get Creality Ender 3 V2 3D Printer for $99: micro.center/9e28b0 ]
[SPON: FREE 128GB USB when you Sign Up for Early Access to Miami: micro.center/1lii ]
[SPON: Shop Micro Center’s Top Deals: micro.center/e2f1a6 ]
Guessing 2880x1600
If 5080 only have 16gb, I'm skipping the 5000 series as well.
I have a 3090, there's no way I'd buy a card that has less VRAM.
Get an AMD card
@@benpedi6645it could be enough until Nvidia releases 6080 with 24GB of VRAM, the next generational gap is going to destroy many RTX 50 series durability, I wouldn't buy any of those card, it's going to be one of the worst generation ever
But Nvidia says DLSS4 and Frame Gen. Nvidia says you dont need so much RAM if you enable DLSS
@@adarion2994every generation destroys the previous generation; some by more than others it's true.
@@formulaic78the 1080ti begs to differ
[12GB on 96bit] people: yay nvidia listens and stopped giving us 8GB cards! buy buy buy!
if the performance is better at comparable price points then who fucking cares if it's a 96-bit bus
people get so fussy about it but ada got more performance with narrower bus widths than ampere did, if they can do that again and still provide better price/perf and ACTUALLY give more VRAM then sure, fuck it, that's awesome
and before someone is like "you will have smaller bus and you will like it" or something stupid like that, the reality is that the specifics of the bus widths DO NOT MATTER what matters is the final performance you get from the card. in an idea world every GPU would have a bazillion terabytes of bandwidth but in reality each given tier only needs so much bandwidth and what matters is the final performance not the numbers on a spec sheet. one of these is how your game plays the other one is a number on a spreadsheet that you wank to.
@@ElysaraCh Enjoy fire explosions tanking your framerate, bud.
@@richardhunter9779 right because ada GPUs clearly have miserable, irredeemable 1% lows in benchmarks and --
wait, what's that?
they don't?
hmm, curious
@@ElysaraCh People confuse where the bad performance of low memory bus comes into play. Gaming is not affected the most (with the more advanced cache technologies, it's still mostly fine), compute is. Low end Ada has been struggling A LOT compared to the comparable Ampere products in compute tasks, because of much lower memory bandwidth. 4070 and higher are fine, but they're not the huge improvements over 3070 and up we should've expected from a new gen. Only incentive to get Ada for compute is the much improved power efficiency.
With each module having 32 bit interfaces I wouldn't think that'd be possible but then this is nVidia and the GTX 970 comes to mind with it segmenting the last 512mb of VRAM to make it useless ;)
Anything less than 12gb for 4k is a bad idea
In 2025 anything under 16gb is a bad idea, in rdr2 with hwunboxed optimized settings my vram buffer get to 12gb after an hour of playing and that's a game from 2018! What most people don't realise is that vram gets filled when you play and a simple benchmark run won't show you the vram requirements of a game
@@FLAXMS That is precisely WHY they will give you a bare minimum... So to me it is obvious that we are getting a 16 Gb 5080.
I play at 3440x1440 and 12 is just barely enough to scrape by on my 4070Ti
If it is $350, I would be fine with it.
16*
One thing I appreciate about this channel is the lack of soyface thumbnails. Thanks for having self respect.
This I agree with....but unfortunately he's a soyboy regardless....
What do you mean?
@@MegaLoquendo2000 what is meant by 'soyface' is wide gaping mouth while smiling, or some other overly-exaggerated expression, like the O-face. it makes you look ridiculous.
The reason for the recent price increases for all PC components is because of Prime Day and Fantastech sales. The retailers always bump up the price before a sales event to "magnify" the savings.
What I'm currently interested in is how the upscaling and frame generation features help with power efficiency. Would be nice if someone can do an in depth video on the settings and best cases for using upscaling and frame generation to improve frame per watt.
From my own personal testing if you lock the framerate you get a pretty decent ~15-25% power reduction. Like a 4090 goes from ~110w to ~85W. Depending on the scene it can be as much as 40% but that's super rare. Not sure on FSR version but it should theoretically be a little worse. Would be best to test it on something like a 1080ti but I don't have one anymore.
Not a huge boost but definitely big enough to investigate implementing it for battery powered devices.
Yeah and presumably most people fall for it. I also see how they alternate between having the discount already applied as "X% off!" or they offer a "coupon" to lower the price to essentially the same level so you somehow feel like you're getting a good deal. Happens with the Minisforum PCs on Amazon, they never really go down in price.
Running BG3 at ultra at 2K without DLSS for a 4070 Ti Super + 5700X3D resulted in my UPS supply beeping nonstop due to the power draw exceeding 300W even in environments with relatively lower numbers of objects. With DLSS resulted in occasional beeps when environmental scenes had higher performance demands. Getting rid of the UPS fixed the beep though and also humbled me greatly as it took way too long to figure out what was causing the problem 😔
My 1080 Ti: “I’m tired, grandpa.”
Also my 1080 Ti: “Well that’s too damn bad!”
You keep diggin! 😂
My 970: I AM TIRED.
@@DeadNoob451 RIP bro. I had my 970 since they came out and didn't stop using it till I got a 3080 at the beginning of this year.
I am expecting to upgrade to an AMD card that is 16GB for around 500 next year no matter what is released.
Me too 😂
(XTX owner here) I expect AMD to launch the 8800XT at $700 or something stupid, and then reduce it to $550 once all the reviewers have ripped it completely appart.
7900 gre for me
So an 8600xt with no other difference other than the vram capacity to the 8600💀
@@andersjjensenwhen will they learn?
If they're dumping 7900xt while raising xtx prices, that's all u need to know about top end RDNA 4 being between the xt and xtx in performance, (confirming earlier rumors)
Prices can just be going up on xtx if supply is limited. Doesn’t necessarily mean what you’re implying.
Probably just slowing down xtx production due to lack in demand?
I just bought an ASRock Taichi 7900xtx for $950 last month. If the prices are going up, then I might have got it at the right time. 😅 Probably won't buy another gpu for at least 4-5 years. (Took me 7 years to upgrade from my 1070)
Won't the top end rdna4 still beat the xtx overall since it's supposedly getting a big RT uplift? Doesn't make sense they would price it under the xtx in that case. Someone else was saying they raised it so they could "slash prices" for upcoming prime day and that makes more sense for me
I'd buy an "RX 8800XTX" with 24GB of GDDR6X for $699 or GDDR6 for $599-$625 if it matches the 7900XTX in raster but doubles ray-tracing performance.
As long as Lenovo keeps using unified chassis between intel and amd, amd is going to be a killer product thermally speaking, the amd lenovo I’ve had run crazy cool by comparison because intels are space-heaters which means the thermal solutions are crazy good for AMD processors.
Excited for rdna4... might try nvidia again also
everybody is mainly focussing on capacity or bandwidth but not price jump from previous generation.
FSR 3.1 is in some cases worse due to the more visible ghosting. DLSS is still in a league of its own in terms of image quality and performance.
I just upgraded to a RX 6800 XT from GTX 1080...8GB to 16GB jump is HUGE!
If you're running that thing at 1440p and mostly play on Ultra settings, then yeah, you just remembered what "smooth gameplay" used to look like :P
I'm still rocking the gtx 1070, with FSR enabled, the performance in bearable but not enjoyable for me, I will wait for rdna 4 I guess.
Don't think the vram did the jump😂
@@TheFirestorm240RX 8800XT with 16G & FSR 3.1, RDNA4 will be a great upgrade form a GTX 1070.
@@TheFirestorm240 keep n eye out for a good 6800xt deal
It's seriously hilarious how people legitimately have -19000 understanding of how much ram and VRAM these modern programs and games actually use.
Its "allocating" that much for a reason people. It's not just some arbitrary amount the developers chose throwing darts at a board. All those meshes, textures, and polygons moving around aren't nothing kiddies 😂
Yes
.....but really no..,.
Yeah no, the thing is overblown.
Yes and no, there are absolutely titles that allocate far more than they really need if the capacity is available. Game optimization went out the window a long time ago.
FSR 3.1 with the new upgraded DLL files for future versions should help it improve faster.
Winding down 7900 XTX seems weird IF RDNA 4 really has nothing that matches or exceeds it...unless it's to keep XTX prices up. Or your theory that they care more about using that fab capacity for something besides consumer graphics.
It makes sense when you factor in that no one is buying them
Early rumors were that the 7900XTX might get sold alongside RDNA4, possibly as a refresh, but if RDNA4 performs close enough to it while having a large enough RT uplift then keeping the 7900XTX at all could be considered counterproductive.
Tom was speaking kinda fast as he speculated on it but remember the 7900XTX and 7900XT are being made on the same chip, Navi 31. He’s guessing they could be trying to get through as much leftover volume of Navi 31 as they can by putting them all in 7900XTs (and GREs). Then the number of 7900XTXs dwindles and that may be what’s driving XTX prices up while everything else is staying cheap.
The big picture is AMD needs to sell as much remaining inventory of 7000-series cards as they can in time for RDNA4’s release. That they are doing this now suggests maybe we will see RDNA4 before the end of the year after all.
@@kenshirogenjuro873Think AMD has something under its sleeve for RDNA4 to match 7900XTX performance.
Or RDNA3 was in truth a failed architecture and they're abandoning it ASAP, as per tinycorp exposed.
@@tringuyen7519 since we’re speculating anyway, another possibility is AMD may be hoarding top Navi 31 chips, saving them for that long-ago rumored 7900XTX refresh (aka 7950XTX?) after all. It would be another good reason to not bother keeping up regular 7900XTX inventory.
It’s old news AMD hoped to be able to get them to run closer to the 4090 in raster than ended up happening, but some people allege to have been able to OC certain AIB 7900XTXs pretty close.
To me a 7950XTX primarily makes sense if the anticipated 8800XT or whatever doesn’t have that big of an RT uplift. Which to be completely honest I don’t expect that much for RT this gen anyway. A lot of people got overly-optimistic about the RT uplift anticipated for RDNA3 as well, despite leaks pointing to fairly moderate uplift and what we got was in fact pretty incremental.
I just think a lot of people would be best served keeping their expectations humble for RDNA4, and if something happens to exceed our expectations then hey, great.
Its looking like the 4090 will be better then the 5080, more VRAM and maybe a bit more performance in the end.
Depends on gimick, they create new magic AI core units and you will have DLSS 4.0/5.0....
I wait for 5080 or 8800xt (depends which will come first and perf/price/power usage ratio...
I suspect the RT will be better in the 5080
I've said this for months... the export restrictions on 4090 class performance pretty much guarantees they'll kneecap the 5080 and make up for it with more frame gen nonsense and efficiency claims.
@@adamb.834 I use the same trick... i play on a 1070 Ti and wait for Navi 48 256 bit bus GPU so their max product or 5080 RTX ... if Nvida card for 1000-1200€ has only 16gb VRAM and Navi 48 has so much vram like 7900XT and cost max 650€ i am on AMD train than. Even if Navi 48 is 25% slower than 5080 RTX.
I’d think the best way to judge the switch 2 power is to look at the best games on switch like Mario kart 8, Zelda tofk, Metroid prime remake etc and imagine what can be done with a bit more perforce than a ps4. That’s quite a jump in power
Problem is Tom think switch 2 is gonna be a handheld pc
@@dante19890 yeah it’s going to be a hand held Nintendo console which is obviously very different
@dante19890 His analysis about the Switch 2 and Steam Deck's CPUs also doesn't make sense.
Sure, the Deck's cores may be faster - which would hurt Switch 2 in _single thread_ situations...
...but he's also saying Switch 2 will have more cores and threads, with two BIG differences he didn't mention being that it will be ARM instead of x86 and will be paired with faster RAM than the Deck.
That becomes less of a _raw power_ issue and more of an _optimization_ issue. A skilled developer could offset the single thread weakness by shifting work to multithreaded + use Nvidia's features to have Switch 2 easily beat the Deck both visually & FPS wise.
@@mitchtellers2547exactly, the mere fact that cpu is octa core beats the steam deck, games would benefit much more from way better mutlithread/multicore perfomance than just slightly better single core perfomance (and even worse in some cases as steam deck cpu has variable clocks), i'm counting on his """"leak""" based on a 2ghz cpu, cuz a78 has a 3.3ghz version for tablets, which is on par with xbox series x cpu
@@Elchinodawn I'm thinking that in portable mode, the two CPUs will ultimately be on par with each other when all of my points are taken to account. Benchmarks will trade off of each other based on how well devs optimize their games.
But Switch 2 will really have a leg up with its GPU. That'll be the difference maker.
a 5060 with 8Gb would be a disaster. Even for 5050 it would be too low. I honestly feel like it's about time we start with the 12Gb or at the very least 10Gb... for '50 series cards...
10 Gb 5050 actually seems perfectly fair. And 12 Gb 5060 . 14 Gb 5070 ,and 16gb 5080. That's a bare minimum in my opinion.
What you said except 16gb 5070 and 18gb+ 5080
@@waynejr.6763 I was talking in terms of bare minimum. As I wouldn't buy anything less. Imo I would prefer a 24Gb on my 5080
anything less the 16gb is a problem for the 5070 ,
5060 - 12 gb
5070 - 16gb
5080 - 20-24 gb
5090 - 32gb .
Vram costs a few $ per chip , and max 10-15$ for the buss size increase . There is not a single reason they should not have 12 gb for low end , and 16 gb for mid end .
where did you take that 14gb?, 5060ti and 5070 should have 256bit bus and 16GB ram. 5080 should have 24 and 5090 32.
@@snakeinabox7220 again. I would personally never accept a new card with less than 16gb. So that's a bare minimum for me. But I was talking more in terms of how much they can screw people and get away with, before it becomes an outrage and boycotting the company... If it was up to me I would be giving the maximum possible amount of vram since about mid range...
Actually this issue is personal to me, because I was waiting to upgrade from my 1080 ti, for so many years now... and proving even 5 years ago, that I needed more than 11gb was difficult. Thank god people are starting to realize that. I'm glad that my 5 years of battle of proving to people that we need more vram had worked.
Higher density GDDR7 (4GB) would get you 12GB on a 96 bit bus. On a 128 bit bus that would be 16GB.
You're right, but those shouldn't be available for a couple years, maybe when Ruebin/RDNA5 are out.
3GB modules would make more sense.
You can also achieve that with double stacking which is how the 7600XT has 16GB on a 128b bus.
They wont come out
you can double capacity that via a clamshell implementation, provided you eat the extra cost ofc
Thank you for the switch info!
Don't use him as a source for Switch 2 info
Omg the camera swing to the laptop screen has me dieing 😂
bud forgor how to use OBS scenes lol
GDDR7 isn't the massive upgrade it seems. The first GPUs with GDDR7 will run at 24GBps speeds which is just 14.3% faster than GDDR6X's 21GBps. That is why the 5090 is going 448-bit to get a 33% bandwidth increase.
As for the rest of the lineup, bandwidth isn't as big of an issue as memory bus width and capacity, and with RTX4000 series Nvidia has set 128-bit as upper midrange, 192-bit as high end and 256-bit as extreme high end. It still blows my mind they were selling a 128-bit 16GB card for $500 and 192-bit 12GB card for $600. Cheapest 256-bit card was $1200 at launch - how tf do you even justify that...
All the major vram vendors will have 28Gbps on launch…
They can justify it because AMD isn't trying
They set dogshit. 128bit isn't even good enough for entry level.
I'm actually not that keen to upgrade my PC after getting the RX 7900XTX on launch.
For normal gamers, this card is already beast and can perform well on 4K monitors for most games on the market.
I can see myself sticking to my main PC for at least 6~7 more years before even considering another upgrade.
Currently I'm more interested on AMD's APU scene though.
Recently when I sent my main PC back for cleaning and regular maintainence, I did got myself into a situation where I'm stuck with completely losing a workstation to support me during commuting.
So if AMD's APU can manage to chase the performance of a 3060 or 4060 series card and can be stuck into a mini PC case, it is something I can consider to build as a back up system for my main computer. I'll even consider to switch some of the work to that mini computer to save some electric bills it they can make it possible.
I've recently realized most of the games I play can perform relatively well on 60 series cards if I'm willing to cut back the effects and if it is just on a subsitute system, I can just swallow it for a few days before the main system comes back.
i also play with a xtx in 4k and it feels like it has enough headroom to last a couple of years
@@herbertwalter8693
Believe me, if Nintendo managed to force gamers to play with a Tegra for over half a decade, the headroom for that 7900XTX is there for quite a long while. lol
Just kidding, but after switching to PC gaming, I did realize that gamer does have a certain tolerance level as long as the game is good.
I dont see an apu being that strong tbh. Not yet atleast, but hey, id love to be wrong.
The whole Switch 2 info doesn't make a lick of sense. It is well known from the Nvidia leak that the T239 SoC has 12 Ampere SMs (1536 CUDA cores). To run those at 800Mhz means it's pushing 2.4 TFlops of GPU power. That is 50% more than what Steam Deck pushes to its max in terms of raw power. And that component is expected to consume 5W in portable mode and the system be fanless? Even if 5W is for 400Mhz, it makes even less sense because pushing to 800Mhz would mean at least 10W by that same notion, and there's no way a fanless setup would work. As it is, Samsung 7nm is inferior to TSMC 7nm used by devices like the Steam Deck, which has lesser performance but also requires a fan.
You make mention that the CPU will be weak, but in various Linux commits that reference the T239 from the Nvidia leak say the CPU will have 8 cores on a single cluster. The only CPU that fits that description is the ARM Cortex A78C, which ARM themselves have stated to be used for high-performance laptops. Even if one were to say it would be clocked low (what is considered low anyways?), there is a situation that people aren't taking into account with Steam Deck. Anyone that has tested it with the GPU maxed out will also notice that the CPU gets throttled down to no higher than 2.4Ghz because it hits the 15W limit of the chip. It's a 4-core CPU, but does have 8 hardware threads. This is through SMT where 2 threads share the same resources, but is only a 25% boost because those resources are being fought over most of the time. When two threads want access to the same resource, one has to wait. So, a 4c/8t CPU clocked at 2.4Ghz is equivalent to around an 8c CPU clocked at 1.5Ghz when all 8 threads are heavily utilized. So, 800Mhz on the GPU for portable mode that is clocked higher than what Switch 1 docked mode set for its GPU, but what would be the expected clock of the CPU? Even at an equal clock, the A78C surpasses Zen2 in IPC. There is absolutely no reason why Nintendo with a custom chip would push the GPU so heavily and not balance it with clocking the CPU enough to keep up with it, especially when they are using a CPU designated for high-performance laptops.
This idiot thinks Switch 2 GPU works like Steam Deck's. That 400-800mhz are small while the 1,6ghz of the Steam Deck are good. But completely ignores that T239 has 1536 cores while the Steam Deck has just 512 cores. He is just sooo dumb.
Bro...this UA-camr doesn't know what his talking about.
i don't see why the switch 2 would be passively cooled in handheld mode, unless they are really focusing on battery life more than switch 1, i hope it atleast has a bigger battery
They are focusing on battery life more
@@MooresLawIsDeadSHUT UP. You dont know anything about Switch 2. Stop shitting through your mouth. Do you even know how the Switch 2 GPU works??? Stop comparing it with the Steam Deck. Do you know with the clocks you gave it outclasses the Steam Deck because the GPU has 1536 cores while the Steam Deck has 512 fucking cores??? 400-800mhz may seem small for the Steam Deck(reason it is powered at 1,6ghz) but for the Switch 2 it isn't. With that big of a GPU, 8nm or 7nm and passive cooling is just out of the question for it. Unless you want a physically huge chip, a heater and poor battery life.
Stop talking about Switch 2 alltogether if you don't even know basic known information about it.
@MooresLawIsDead SHUT UP. You dont know anything about Switch 2. Stop shitting through your mouth. Do you even know how the Switch 2 GPU works??? Stop comparing it with the Steam Deck. Do you know with the clocks you gave it outclasses the Steam Deck because the GPU has 1536 cores while the Steam Deck has 512 fucking cores??? 400-800mhz may seem small for the Steam Deck(reason it is powered at 1,6ghz) but for the Switch 2 it isn't. With that big of a GPU, 8nm or 7nm and passive cooling is just out of the question for it. Unless you want a physically huge chip, a heater and poor battery life.
Stop talking about Switch 2 alltogether if you don't even know basic known information about it.
@@MooresLawIsDeadbullshit just use 8 Samsung prehistoric node. Greedy company
The only reliable channel for switch 2 leaks
He is full of shit when it comes to Switch 2 though
1:31:50 this actually bad analogy. Bronze swords are at par with high quality iron swords what comes to toughness and hardness and bronze swords are actually far superior agains corrosion. Why ironworks was revolutionary technology was not quality of items it was the volume. Production of iron weapons and tools used like one tenth of resources compared to bronze working.
Nvidia has 0 incentive to future proof anything, since they have monopolistic mindshare for now.
They are still pushing tech further even with no competition
@@dante19890dlss is the only worthwhile thing they did. All they are pushing are prices of GPU
Nintendo did NOT say "they will launch the Switch 2 before the end of this fiscal year". They said they will announce it before the end of the fiscal year.
Nintendo usually launch their console very close after the reveal
@@dante19890 this is going to be an announcement, not even necessarily a reveal. That Switch 1 was publicly announced in March 2015, codename previewed at E3 June 2016, officially revealed on Oct 20th 2016, and launched 5 months later on March 3rd 2017. 2 years bro
@@--waffle- fair enough. I think Nintendo doesn't wanna hurt the switch sales by revealing it too early.
It seems PS5 pro would benefit launching later this year if switch 2 launching q1 2025
@@dante19890 agreed. Nintendo are scared of another Wii U type console flop. Switch has sold so well that unfortunately, they will have to continue supporting it for many more years. They wanna sell games
@@--waffle- ye revealing the switch early had no impact on Wii u sales cuz it was dead anyway.
This time they have too much too lose
I'm on a 6700XT which has been good so far, but I'm itching for an upgrade. I don't care if it's from AMD or Nvidia. All I care for is price to performance. Still reluctant to adopt Intel, but I'm open to it if they continue to improve.
Some good intel in this vid for sure. People are still waiting for battlemage, lol.
I have a raptor lake cpu. I turn off the E-cores because it causes too much problems with games.
Nice glasses, good look!
like a 50s secretary
Not as good buffgamingtech😂
I believe the way to fix this perception of arm being a cure all is for x86 laptops to have the same sleep behavior as android, ios, and arm macos. In my experience, i can never trust windows sleep because it randomly turns on, or have bad standby time.
Can't wait for the tests.
The lowest SKU (5060, etc) should have no less than 12GB of frame buffer.
We all are aging with the leaks of Arrow lake
Intel aren't able to take that damned arrow out from their knee 🤣.
I used to be a lake like you, but then I took an Arrow to the shoreline.....
@@johnnyringo35 damn that's so bad which adds to it 🤣🤣.
just grabbed a new 7800xt for $400. ill hold onto this for a while
I disagree on the "upscaling can not be improved algorithmically" part. Advancements in image processing and rendering methods happen all the time and will continue to do so. Integrating AI upscaling into FSR might ba good idea, but there's certainly room for algorithmic improvement too.
Image reconstruction could actually add more detail than what was actually in the original image
Miami micrometer? Let's gooo
Most of these decisions make it sound like AI is gobbling all of the highest speeed GDDR6 and GDDR6X chips up, and their options are bid higher (driving market prices up further) or avoid using it. All of the RDNA3 Skus referenced other than the 7700XT are using 19.5Gbps and 20Gbps memory speeds. Notice the GRE is missing from the report. Could it remain in production, or just become the outlet for which they disseminate remaining N31 silicon? Could there be a flood of 7700XTs with the last bits of N32 silicon? Shutting down production of these SKUs supports the rumor that RDNA4 will not be high-end and will use 18Gbps memory.
I'm thinking Intel will have 25% cpu market share in 2026. They just seem to be giving up completely on competing ..
Of course Nvidia will skimp on v ram and then they will do a refresh and make the card what they should of been to begin with. We have all read this book more than once before.
The only game that makes me upgrade my PC anymore is Minecraft, so VRAM matters quite a bit with Blackwell. My 4070Ti can just barely handle the textures and shaders I throw at it
That's an insane statement, but I love to see it
With an RTX Card you should be able to play with Raytracing instead of shaders.
I remember doing that with my 3060 but it struggled to stay at 30 fps.
@@alexander1055 ray tracing falls under the subset of shaders, and many “non raytracing” shaders still offer many RT driven effects like AO, volumetric clouds, soft shadows etc. I was using a fully path traced shader until I switched to distant horizons, for which currently the best shader available is Bliss (if your system can handle maxing the settings). At those settings it can tax my system as much if not even more than SEUS ptgi.
@@alexander1055 Keep in mind I’m running the game with nearly 60 mods at 3440x1440 with a render distance of 450 chunks (20 chunks full quality + distant horizons chunks). The bliss shaders are using RTAO and flood fill to simulate occlusion and global illumination. Shadows resolution is 4096p for 20 chunks, with the remaining 430 chunks also shaded at 1024p. Volumetric clouds and fog effects that use ray marching. My texture pack is exorbitant with lush bushy trees, lots of detailed bumpy surfaces and material effects. My world build height limit is 2 kilometres…you get the gist lol
@@mickaelsflow6774 The only game that evolves with every generation of PC hardware. The ceiling is insanely high for what can be achieved, I wouldn’t be surprised if we have the same conversation again in 10 years despite having 128 gig vram buffers, 3d stacked dies, internal resolution of 240p thanks to AI overlord upscaling and whatever else the alpha nerds throw at us.
12Gb is bare minimum for 2025. Some games where already useing more than 12GB at 1080p. And nvidia wants to sell 16gb for over 1000$ 😅
Most newer UE5 engine games are actually very memory efficient.
The problem games where pretty much all cross gen titles
Snapdragon X for general computing use cases has gotten pretty decent reviews so far. Yes, for gaming it sux but one can see where by the time the next iteration of the X comes out, the iGPU could be improved and by then there will probably games compiled for it natively. In the meantime, sales for productivity and web browsing users, plus Microsoft’s support for the platform could it alive. Basically, I would not rule out Windows on ARM in the long run, and neither should anyone want to. Competition is good.
Even then, it's rather expensive for general use, which is a major issue with Apple in general.
@@sonicboy678 more competition will come out in the future... Well i think it was supposed to because Qualcomm and MS exclusive deal is ending. Mediatek probably will have theirs next year. Samsung probably will try the market as well.
Warning Level 1 Wendel made a video telling how horrible was Qualcomm launch. And what happened behind the scenes and it's a nightmare.
MS has an agenda, and Qualcomm paid too much propaganda to buy tech outlets. Qualcomm experience is a mess, something not fixable now.
I would stay away from arm for atleast a year so ppl don't waste their money.
@@arenzricodexd4409 If Samsung does jump in, well, I hope the SoC is good.
Let's see if a certain company is willing to jump into the ring.
From what I've seen it's only ever so slightly faster than my 7840U in CPU and GPU, but it costs a fair bit more, and it doesn't have that much better battery life. That's not a good spot to be in for "just general computing". If you're catering to the "I need a webbrowser and an electronic typewriter" market you need to be price competitive.
You hit the nail on the head here, Microsoft should stay out of everyone business and just get windows to work consistently without all of the bloat ware. I don’t use their cloud software and haven’t even signed into it, but I have had other problems.
I’d take a X Elite on sale for 1/2 off?
Not because it’s better than anything else, but because it would be a good development, laptop for arm code.
The funny thing about it is the best Windows on arm laptop available right now is probably a Mac using virtualization.
Best Windows on arm is a Mac using virtualization… So true!!!
When I bought 7800x3d/X670E-F from Microcenter in VA, I asked them if they sold a lot of these bundles. They said they sell more AM4/Intel stuff then they do AM5. I was surprised because of all the misinfo online. The associate said that they have had a lot of returns of AM5 boards due to instability. I havent put mine together yet but I did get a X670E-F for a hefty open box discount. I hope it isnt unstable. I was able to choose from 5 different open box X670E-F they had in store and I picked the best looking one (it looks brand new and has all unopened accessories). They actually had 20+ open box B650E-F in store as well. Perhaps its an Asus issue but I was shocked by how many open box products they had while they had hardly any Z690/Z790 open box.
The AM5 instability is down to OEMS shoving waaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaay too much voltage into the CPU and it started with AM4. As a fellow AM5 owner, I highly suggest you use voltage suppression to cap the voltage at 1.3x volts ceiling and voltage floor of the lowest value. My 7950X can no longer boost to 5.7 Ghz reliably, at one point I was not even getting reliable 5.4 Ghz.
@@earthtaurus5515 I have also seen many users report instability at stock in AVX2 workloads as well as SHA3 issues with any negative CO.
You just went on a whole tangent with no grounds about anything regarding the Switch 2 lol
I'm glad that Tom has his color leds working, miss those days.
@@djchristian82 eye damage
8GB is unacceptabe. Even entry gpus like a 5050 should have 10-12GB Vram. 60 tier cards need 12, just like 70 tier needs 16.
Man it would be awesome to see what we can do with it running Linux/Android. PS2/GC/Wii/PSVita emulation.
Nintendo said they will only announce their successor/next console by the end of this fiscal year, NOT that it will come out. So they have given the time frame of up until March 2025 to announce/reveal the system with it launching presumably some months after. So the successor could be announced as late as March 2025 next year with a release as late as fall next year. (Of course we don't know this will be the case) they could in theory just as well announce the successor now and release it this autumn; I just think it's important to quote them correctly, that all they said was that there will be an announcement by the end of the fiscal year; not a release and judging by the very big lineup and roadmap Nintendo has just revealed at it's Direct the other week with it's games for the OG Switch for the rest of this and beginning of next year, it doesn't look like we're getting the successor any time soon. Judging by this Direct and the amount of games still coming, I think we won't get this new system until at least in a year's time from now, so summer 2025. Unless sales of the current systems dramatically drop leading up to Christmas, I don't see them seeing a need to release or reveal anything sooner. (But of course I hope I'm wrong because I would love a new system sooner rather than later but even judging by Nintendo's own comments, 5 years in to the Switch that it's only 'half way in it's life cycle' ,shows that Nintendo is in no rush with how well the current system is still doing.)
It's 100% releasing in early 2025.
I'm genuinely curious how VRAM allocation works in games. I mean just think about it. If 8 GB doesn't cut it for 1080p gaming anymore, how does 4k (driving 4x the number of pixels) get by with 16 GB? That doesn't make sense to me.
it's not really about the amount of pixels, it's the textures and stuff like that. if the game uses 8gb at 1080p and functions well, then there's not a lot of reason why it would consume 4 TIMES that at 4k. I mean think about it, it's not like one pixel consumes 10 megabytes and the entire memory gets overwritten each frame, that's not how it works.
Didn't see any mention in the live stream itself (and not caught up on the latest broken silicon), but anything regarding the alleged 6400mts sweetspot for zen 5? It's quite a bit off the 7000-8000 figure I remember you mentioned over the last year or so. I mean maybe it supports it just like zen 4 supports it, but support and sweetspot are very different things.
Anyway, great stream as always, thanks c:
i don't think anyone said 7-8k sweeet spot, just support for memory in the 8k+ range. we can already do 8k as is now.
@@_Kaurus Okay, but I'm looking for Tom's take on that leak. Its source appears to be wccftech who claim sources with an engineering sample, and techpowerup basing their coverage on wccftech's article.
Once the 6400 sweetspot leak dropped, the 7-8k support just confused me as to why it received any emphasis if ultimately it's a figure of only secondary importance.
So the question remains - is that leak accurate? Any chance AMD'll manage to crank the fclk up to have higher memory speed sweetspot? Or is this likely to be the final figure
They don't want the top of their new stack having a smaller vram number than their last gen on the shelf. They need Navi 31 stock gone.
The switch 2 will easily run any ps5 game at native 960-1080p and use dlss to get 4k.
It's wild to me that we'll probably get a 5060 8gb in 2025 when Xbox will be pushing next gen games in 2026. That means 3rd parties will do the same and their games will probably be less optimized than Xbox's. There's no way games a couple years out like Gears of War E-Day will run even at 1080p medium on an 8gb gpu.
When will 5000 series from nvidia be officially announced? Next expo?
I'm really lost tbh. I wanted to go back in to PC gaming and so i looked up what is going on with CPU's and GPUs. And i have no freaking idea what to buy, to wait for what ever, or just don't care and buy what is on the market. Money does not matter, but i don't want to burn it 😅😂
It’s rather simple, atm 7800X3D and the most expensive gpu you can afford.
@@puffyips isn't the 7950x3d the hot shit? And yes obviously the 4090 would be the way to go, but tbh I don't know if I want to switch to a 5090 in 5-7 months.😅
Probably i will test a 7900xtx and switch to the 5090 then, so I have also some mix and experience in the differences between AMD and NVIDIA.
@@ConsiderbeBasedyou said pc gaming so 7800X3D is the only chip you’d want rn, and I didn’t say 4090 cause honestly it’s not needed, 7900xt/ 4070ti super perf will hold you over even in 4k but if you’re going high highend then ig wait
@@puffyips yes i see where you go with your recommendations and thanks for that. The 7800 is insanely good for the price and probably the best gaming CPU to purchase right now. I think i will go with the 7950 so I have some power for rendering stuff extra and honestly in short there will be new CPUs so will see if I switch to a nother when the x3d versions come.
After the AMD 9000 debacle I think you should find some new sources for your "leaks".
7900 is one of the most notorious unlucky numbers. Its not quite 666 or 73, but why risk it?? 😢
Is anybody surprised ARC is dead? They jumped in thinking they could ride the crypto bubble but it burst before Alchemist came out, the only ones I have seen sing any praises for Alchemist is streamers buying the A380 to transcode their streams while their Nvidia and AMD GPUs do the actual heavy lifting, and with its piss poor support for older versions of DirectX and OpenGL during a time that the Steam top 20 is filled with older titles was just the final nail in the coffin.
It is a damn shame as I was hoping it would be a 3 horse race like in the old days with 3DFX, ATI and Nvidia but with Intel taking huge body blows from AMD in everything from desktop to server to laptop spending billions to try to open a new market at this point was just a bad call.
Tom, I to would be surprised if Nintendo didn't release the Switch2 in time for Christmas sales.
I think they wait to q1 next year
@@dante19890 I think it all depends on how much Switch stock they still have left. If it's a decent amount I could see them doing a small discount for Christmas sales of the Switch to get rid of as much of leftover stock before the Switch2 release. Otherwise I expect that they would want the Switch2 available to be sold in large numbers for Christmas time. But this is Nintendo we are talking about and they don't always do what one would expect them to do. lol
Anyone have insight on how large the RDNA 4 lineup will be? We are hearing that the highest will be a 7900XTX equivalent with better ray tracing. So will they have a 8800 XT, 8700 XT, 8600 XT, 8500 XT? Or are they just going to leave it at two cards? 8800 and 8600?
They typically do three chips, and the top one has been cancelled. So we are probably going to get 8800XT and 8700XT on the now-top die and 8600XT and 8600-non-XT on the small die. If AMD have any brain left they'd do 16GB on the 8800XT, 8700XT and 8600XT (clam-shell), and only put 8GB on the 8600-non-XT. Resisting the urge to put 12GB on the 8700XT could really do them a favour here. Alternatively they could make a 8700-non-XT with 12GB (and a few less CUs) to get rid of any dies that have defects in a memory controller.
Nvidia will probably launch 5000 series just like 4000 series(except 5070ti, that will be 16gb). Then when 3gb vram chips become available, Nvidia will probably launch 5000 super series with increased vram.
5080=16gb/5080super=24gb
5070=12gb/5070super=18gb
5060=8gb/5060super=12gb
So are the super cards just the real product at this point
I expect 5060 to have 10gb vram
That’s being way too optimistic..
This is what I'm expecting for mid-range:
5060 - 8
5060ti - 12
5070 - 16
Personally, I'm buying a 5090 which will be overkill for 1440p, but vram won't be an issue.
Also, wondering if Valve will go with Strix for SD2 or wait for RDNA4. Should be pretty big efficiency improvements.
Please tell me you don't have a 4090.
@@zoonbk steam deck I reckon will be zen6 based.
If AMD can utilize PiSSeR for RDNA 4, then I would actually think about switching. Of course, RDNA 4 will have to be priced right too.
Think it's Sony proprietary tech
When we get on rdna 10 I change my 6700xt.
I don't have sources but I am 100% sure 5080 laptop will have 12GB and Nvidia calls the two 16GB mobile GPUs 5090 and 5080 Ti. They have zero competition in laptop so they can milk as hard as they want.
If AMD could launch the 12 core strix with 16 cu mode on desktop for like 330 or 350 dollar then many will buy it in south asia and south east asian region. Even the gamers who will throw away old pc in 2025 when win10 is eol.
Makes you wonder why they bothered releasing the overpriced desktop 8xxx so late.
Lunar lake being single thread was probably on the design board when there were all the mitigations were being carried out for multithread
I once had that onedrive problem and I swear that if I had had the guy responsable for storing all my files in the cloud by default in front of me he would probably be dead now.
There is no way a 5090 is acceptable at 24GB. I have a 4090 and people that spend that kind of money will not upgrade without a boost in RAM capacity. It needs to be 32GB, if you need more then you go to professional cards. Keep in mind that AI models are getting bigger by the day and 24GB is not enough today. They wouldn't erode the more expensive products because the gap is too great anyways
😅 i ll buy the rx 8500 xt or the rx 8400.
I've been waiting on the new CPU and MBs, nwo they saying the new CPU will launch before the new boards, well damn, there goes my buying a 7600 and a x870 board til the new 9000x3ds release . So I guess I will jsut keep on waiting since my 5600x 6750 xt are doing just fine, was also hoping to buy a +16gb gpu around the same time, but now they selling through and going up in price smdh AMD releases any new GPU's with less than 16gb and over 600 can shove it where the sun dont shine! Really thinking of grabbing 7900xt before they sell through
Do you expect ps5 pro will be revealed beside Wolverine gameplay? PS4 pro got revealed in September so it this possible this year as well for PS5?
Glad I didn't buy 7000 series this year, I was about to pick RX 7800 XT or RX 7900 GRE, because they were coming out below MSRP. I think this is year of Monitors not GPU. Better get GPU's next year
i agree
To be fair I saw the 7800 XT for $440 and the 7900 XTX for $800 for a split second...I don't think that's a bad deal. However, neo we are probably a quarter away from the RDNA 4 launch, and I wouldn't pay a cent more than that therefore...
nvidia is the new intel from 2010. we will see what happens in 10 years.
except that they get 30-60% gen over gen improvements instead of 5-7% ones
@@zuckdaddy1596 30-60% improvement? what drug are you on i would like some.
@@zuckdaddy1596 tell that to the 4060 owners. In fact sometimes 4060 is worse.
@@Valueshooter it also costs less. pricing is disappointing, but Nvidia is making huge leaps gen over gen, which Intel was not. it's not even comparable
@@zuckdaddy1596 1. It does not cost less, i bought my friend a 3060 $299 bestbuy, BEFORE 4090 came out. don't use crypto mining crazy price as a comparison.
2. #2 reason 4060 is a scam, it is a cut down die compare to 3060, basically nvidia is trying to pass off a 4050 as the new 4060, everyone knows that, so stop lying.
so not only are you paying same price for a "4050" its marginally better and sometime not better than the 3060. it is very much comparable to what INTEL did for 10 years.
3. Ask anybody around if the gen over gen improvement vs the price is worth it. unless you never heard of the rtx 2060, 2070, 2080s, be honest the RTX 4000 is the repeat of what nvidia trying to pull over the gamers during the rtx 2000 generations. Its like you never watch any of Moore's law is dead videos, this point has been talked about over multiple videos, only a nvidia fanboy would defend them on RTX 4000 series performance and price anymore, other than the 4090, there is nothing good about 4000 series, and i would add 4090 is overpriced and not very well built.
4. you don't build goodwill by ripping off your fans/customers, by over charging them over and over, INTEL did that for 10 years, look at them now, nobody feel sorry for them.
16gb is ok for a 5080.
You gonna use DLSS quality at 4k anyway
With RT even 1440p will reach 16G, in 2024. Someone buying a 5080 likely want to keep it until at least 2027.
@@Violet-ui RT doesnt really put more pressure on the vram its more added cpu calculations.
Do we know that ARM64 emulation is the main cause of issues? Could many of the issues be caused by Adreno DX12 GPU drivers?
Intel had a lot of DX12 driver issues too. Vista was considered a buggy mess at launch, and Microsoft revealed that Nvidia’s GPU driver was causing majority of the blue screens.
Are we sure that ARM64 emulation is the main culprit?
On Android side I have had to write workarounds for Adreno driver bugs. And they have shipped GLES/Vulkan driver for years. It should not surprise anybody that their first DX12 GPU driver has issues. On PC we also have compute centric workloads that are not common on mobile.
I would expect improvements to both stability and performance once the GPU drivers mature. Assuming of course that they prioritize PC driver work heavily. Intel didn’t prioritize iGPU gaming drivers heavily in the past. Let’s hope Qualcomm understands the importance right now.
It's not surprising that the 5080 will be gimped considering they've already reached peak performance when it comes to being able to "export" to china and thats a quarter of their income. On top of that it sounds like Radeon is pretty much throwing in the towel on rdna 4 so what motivation do they have to make a whole separate sku for a 5080 ti or whatever with more vram? A 5080 will destroy any game on the market for years to come and anyone that truly NEEDS a 5090 (small market) will cough up the cash. On top of that they're focusing on AI bc their market valuation is through the roof. A couple of aussie dogs barking at the proverbial caravan wont change market dynamics.
Bro would love some news on ZEN 5 X3D!
As Tom and many others have stated: It will be shorter between Zen 5 and Zen 5 X3D than it was between Zen 4 and Zen 4 X3D. Other than that, nobody has anything solid yet.
@@andersjjensen I know bro but there can always be more leaks & information. The whole genre is drooling for those X3D news and you know it.
@@normative1058Dual chiplet vcache, higher boost clocks than Zen4.
It makes sense for Intel in the dedicated gpu department wait until it matures, before releasing a mid tier. So anyone hoping a new release unfortunately it will be igpu related. It's a smart move, launch when ready. Lovers or haters of Intel, as we speak Intel can't afford to launch another series of cards at a loss. The more L's, the more the house burns in fire. I'm hopefull maybe next year around fall, they could announce something, but not guarented.
Ill get a ps5 until i can afford a 1500$ Pc
And play what games lol. Get an Xbox off anything
@@Lucid_XP there is 41 games i want to play on the ps5.
Xbox is so stupid, then id rather get a pc
@@Lucid_XP here are some of the games im gonna play on Ps5:
Ghost of tsushima
Rise of ronin
Demon souls
Elden ring + DLC
FF 7 remake
FF 7 rebirth
FF 16
Post trauma
Black myth wu kong
Wo long
Sekiro
Lies of P
Skydancers behemoth
All Uncharted games
All Yakuza games
Nier automata
Stellar blade
Enotria
The relic
Clair obscure expedition 33
Star Wars outlaws
This is just some of the games i want to play, the game i want to play the most is MH wilds.
You know you can build a pc for 600$ equivalent to a ps5?
@@webtiger1974PTG Not even close in my country. Just a rx 7600 alone is 500$
1500$ will get me a 5700X3D
And a RX 7700XT Pc. Its better than Ps5 for sure but with my post i meant ill use a ps5 until i can afford a decent Pc
Btw a Ps5 disc version is 870$ 🤣
Sucks for intel, competition is good uwu
xx80 tier people getting screwed again D:
I must be a wizard or to smart for some no plan people ... I say it even 1 year many times 12 gb VRAM is not enough... for gamers. When you play at 1440 you keep your GPU what? 1 year? I buy and keep gpus for 4-5 years. Before buying some new. Even 16 gb VRAM on a 5080 RTX that have a lot of power is not enough for 4-5 years. Its Nvida trick to get paid for all 2 years for their new gpu. I don't need to be a rocket scientist to see that.
Raytracing in more games, Upscaling from 1440 to 4k , better looking game better = higher settings and Hardware needed = More VRAM useing ^^. Some people dont know that Unreal 5 Engine dont even shows in a game true power and 16 gb for 1000€or $ gpu is way to expensive. If you upgrade all 2 years a new GPU from now on ok another topic ^^.
We have a lot of Techno logic going on the Last 1 years, that looks better, but eats more of your system.
I have a 7900XT, had to RMA my original one because of hotspot issues.. Anyway I'm of two minds with the discontinuing of RDNA3. One side is saying it makes sense if they need the packaging for other things and if RDNA4 is going to come out at the end of the year. The other though knows that they will only have 1 generation of chiplet based GPUs so it will make driver development for the 7000 cards lower priority and they will probably have more issues as time goes on.
Yes we will get our 5 years of support 100% but having the almost 10 years that something like Polaris got will probably not happen
The 7600 is not a chiplet design and they are going to continue to make them, possibly for a while. The PS5 Pro *might* be RDNA3 based, at least partially, so there could be more attention given to further optimizations if it doesn't break away too far in its design.
@@douglasmurphy3266 Yes they should keep making the one card out of all of the RDNA3 GPUs that reviews dislike. Genuis
@@douglasmurphy3266 I think it'll likely be RDNA4 based. I remember X1X and PS4 Pro being the latest architecture.
The past amd drivers such. I have 6800 xt. I get Flickr scenen when playing UA-cam and a video appear on Facebook same time. The last driver dint fix it. Had AMD past decades. Buying NVIDIA nexttime.
I'm curious to know if Nvidia DLSS 4 will be exclusive to the RTX 5000 series. Can anyone shed some light on this?
Knowing nvidia, they will very likely lock any new features to newer cards. Previous generations are enough evidence to back up my assumption.
I bet 5000 is gona be last rtx then 6000 is like 2000 😂 lol
@@Oliver-sn4be PTX next
I had issues with one drive doing that same thing, storing certain game files on the cloud then I boot the game and wait 5 minutes while it individually downloads each file to my machine again, fucking hate one drive. It’s because I subbed to ms 365
In before the timestamps
When anyone says DELL, I remember a bunny suit...
might be that rdna4 might not be better than rdna3, so they don't want any rdna3 left to avoid lowering rdna4 sells, like rdna2 did with rdna3 by being really competitve and widly available.
Who knows, that's being very pessimistic lmao. If so Jensen Huang will become the leader of the UN and make 5090 the 196th country of the world.
Realistically, all manufacturers want to clear left inventory before new launch.
Are you even a insider?
You're very bi-polar when it comes with this information. UA-camrs armchair developers
1:02:30 damm ):