While x86 probably won't 'die out' I wouldn't be surprised if ARM chips make decent inroads in more sectors than just laptops and tablets like they have in the server market.
I really hope AMD does something incredible with their GPUs during the next years, and force Nvidia to actually do better to make sure GPU competition isn't like CPU competition was before Ryzen
I think it's actually a different problem now. Actually making full use of a 4090 even in 4k gaming, you're typically CPU-bottlenecked unless you go out of your way to turn on silly Crysis-type settings. If the 5090 were twice as fast as a 4090, you're looking at a card that's only worth it for VR, at hardware resolutions that haven't even released yet.
@@pondracek Then the 5090 should be same performance as 4090, but that isnt the case, the 5090 seem to be 50% faster than 4090 which negate your reasoning above. The only reason NV is doing what they are to their RTX 5000s series is greed. Nothing else. Only those who pay big dollar will get the performance jump we wanted
@@Gattberserk ? Because of the 4090, there is now a new free 3dmark benchmark. The card itself is 90% more GPU with 50% more memory bandwidth than a 4080. Yet the reviewers average their reviews out to a 20% gain at 4k over a 4080, even less at any lower resolution. It's just very hard to actually find a GPU-bottlenecked test to actually throw at it. A 5090 at only 50% faster over a 4090 would be less than a 10% gain at 4k on those same benchmarks, with an asterisk when they turn on path tracing in Cyberpunk at 4k (and ray/path tracing itself has a separate CPU hit anyway). And for some reason they're talking about the non-cpu bottlenecked use case as the exception and not the rule. (Putting an asterisk, around frame gen is valid. Putting one around like-for-like ray tracing against a same-gen nVidia card is not.)
@@Gattberserk My reply was for some reason obliterated. The 4090 is already 90% more gpu and 50% more memory bandwidth than a 4080, with reviewers saying it has a ~20% average advantage at 4k on existing ultra benchmarks. If you can't see that that's bottlenecked, I don't know what to tell you.
@@Airwolf2030 it's easy to avoid Recall, just get Linux, it's actually worth the experience. some Linux distros even come with systemwide adblockers built-in, like Garuda Linux. Linux isn't targeted by a lot of hackers either, so it's really not worth to care that much about viruses, as it's very unlikely you manage to get any viruses at all anyway
@@TheHangarHobbit what about the 5900 XT which will be available i july? AM4 is still more budget friendly, even for "high end" desktop builds. just getting into AM5 at all is easily more expensive than building a decent "high end" AM4 build with comparable performance
@@lassebrustad Problem is that 7800x3d plays sooo much better than 5800x3d in UE5 games, like 15-20% better. Not sure about 5900, but I just do not see a future for it 5 years from now. My rigs last 5-7 years so I am due for a makeover. Soon be breaking out the popcorn and waiting for a good deal, maybe on a pre-built.
4070 Super is the best product of a ripoff generation. Fake pixels, fake frames, stuttering ray tracing at dog-slow fps ... Of course its low power its not really doing more than 720p upscaled most of the time! And yet NVidiots worship this company!?!?!
You need the hardware adoption to happen for the software development to take place. Hopefully MS will stick with ARM this time around but I'm much more interested in a nice all day battery life laptop with Linux on it.
**not including gaming.... the Snapdragon X's are only 40-50 percent faster when compared to the Intel Core Ultra's and AMD Ryzen HS's... those are laptop/mobile CPUs and while that is impressive for apps, editing, coding/programming, graphic design, video production, etc.... it's not anywhere near the power of even mid-range gaming CPU/GPU combinations in a full-sized PC
AMD just announced new x86 chips, including two new AM4 chips. No, it's not "the end of x86". I can see myself grabbing an ARM-based system for something like Plex, but I'm not sure what other use I'd (personally speaking) have for one at this time. Also, let's not act too surprised about the 50 series. Nvidia has gone all-in on AI, and it shows. Now, watch as the 50 series is just as overpriced, if not moreso, as the 40 series.
I wouldn't mind a low cost low power mini pc like arm device for tinkering with in my homelab. Thought about grabbing the Qualcomm dev kit machine but that's a bit more than I want to pay for a box to tinker with when I don't actually do software development.
A five years old CPU with (as example) 6 cores compared with today’s 6 cores CPU is working different and efficiently as the old one. So the same amount of cores are not really proving less power or performance brake. But sure for each customer it is our wish to have more cores. But the focus should be the smoothness game experience without matters of the cores.
Yeah, people get "sticky" expectations. I've got an 8 core. Even though a new 6 makes it look slow, I still kinda want a new 8 core minimum. If someone came out with a 100ghz single core that made all the current chips look slow, most people wouldn't want it.
@@myne00theres a balance between core count and speed. The problem is that 1 core can only process so much information at a time. Even if it's fast, it can only do a single thing at a time. Would you really want games to stutter all the time?
@@doge151. Technically a 1-core CPU at 100 GHz, assuming similar IPC to the current CPUs, would be twice as fast than an 8 core, in all workloads, as long as it's not memory bound aka waiting for data to arrive and doing nothing in the mean time. But that would require that programs (and games) to be coded a bit specifically for it, to not have excessive amount of threads. Kind of like having a program or game work well with multiple cores also needs a lot of code and architecture of the game engine restructuring, to properly benefit from having multiple cores. But if the OS is configured to have quick switches, not have too many background threads ... well threads in general, then the 100 GHz would blow existing CPUs out of the water. Because in some workoads, those that aren't well (or at all) paralelized (like, say, Starcraft 2) it could be as much as 16 times faster than current CPUs, no matter the core count. What I wanted to tell is that the current software might not fully benefit from it, but if coded for it (and single threaded coding is much easier than multithreading one. Also has less overhead) then the "can only process so much information at a time" quickly dismantles. It's really a bad argument.
I see Qualcomms new chip as their M1 moment. Massive performance on the table with the first gen, just like AMD did with Bulldozer to Zen and then will be followed by incremental updates just like the M series chips. x86 still has a lot of life left and will go on for quite some time.
So crazy how gunpowder was invented in China, went all the way to Portugal, then got introduced to Japan. Nearly made a full trip around the globe to get next door.
Yes, and the English build their Empire on that gun powder.... and them made a book with propaganda against the Portuguese who were the ones saving the Samurais selling then guns.... instead Netlix made a good Series called Shogun on that propaganda book were the good ones were the English who were atacking the Samurais and the Portuguese who were trading with them were made the bad guys somehow.....
X86 IS HERE FOREVER, NPUs cannot replace X86 since its only meant for laptops, and no Copilot cannot and does not work natively in Desktop CPUs since you need a minimum of 40 Tops for Copilot to work natively with no internet, thats why NPUs like the AMD AI 300 with 50 Tops and Snapdragon Elite X with 45 Tops where Introduced.
'Needing' 40 tops has already been disproven, people have gotten recall running without NPU's though some functions were obviously way slower without an NPU.
@@nadtz Recall is in the Copilot App, which does not need to run natively if connected to the internet, so no need in the NPU since its running from the Clouds NPU server.
Yeah just like I thought due to no competition and the main focus being AI and servers. Unless you want a flagship 90 class GPU there is no reason to buy new GPU’s.
BTW I built a SFF with an AMD K6 chip and an ATI videocard long long long ago. It was even one of the first to use heat pipes I believe. Very very tiny psu in it as well. I forget the name, my parents wanted one to fit in a tiny spot.
Snapdragon on desktop is a non-starter! AMD’s Strix Point will destroy it bc Strix doesn’t need emulation. RDNA3.5 & 50 TOPS NPU will annihilate Snapdragon on gaming.
the Snapdragon doesn't need emulation for everything tho, but anything that isn't optimized for ARM yet, just like it was when Apple made their M1 chip, except that Apple made their impressive emulation. I'm not an Apple fan, but they did something very impressive anyway, even tho it was far from perfect
as a consumer you should want Qualcomm to be successful here. competition drives innovation and reduces prices for consumers. Look how cheap things got when AMD got competitive compared to before. Now imagine another player get involved and if ARM takes off. even more ARM players will come on board
@@percy9228 According to some leaks I saw recently AMD is designing their ARM cpu so I think it will take off in next few years. Nvidia most likely too.
@@jawohlbxb3534 Yep, keep seeing Nvidia and AMD come up in regards to possible ARM CPU's for desktop (probably servers also), looks like Intel are the only ones not doing so at this point.
I am really not excited at all of having ARM chips soldered to the motherboard, with soldered RAM (not expandable/interchangeable), soldered storage (also not expandable/interchangeable) to the same motherboard. Those things will just create more e-waste in the world. 😩
Probably a bit more, 10% IPC (that's what i believe they'll bring to the table) + higher clocks i believe, i think it'll be around 25-30%, but 20% is a realistic guess.
Essentially the 5080 will be like 10% faster than the 4080/super. The 4090 already has a huge lead with 20 to 30% more performance compared to the 4080/super
Well that's a weird conclusion to draw from AMD not launching high-end RDNA 4 GPUs, since Nvidia still appears to be making big gains in the highest tier card but then they're not doing much for the lower tier cards where AMD is actually going to compete, AMD is only abandoning the high-end, not the entire GPU market for RDNA 4 so it doesn't make sense that Nvidia is planning to nerf every card other than their highest tier one due to "a lack of competition from AMD".
Trying to stuff high performance GPU's in SFF cases is a fool's errand. Coupled with the 12VHPWR cable and tension from limited cable routing options you're asking for trouble. I have a Fractal Terra I use for video editing. I use a low profile Intel A310 video card. Temps never get above 54C and it's relatively quiet. Just because you can stuff a 4080 or an i9 in a SFF case doesn't mean you should.
So the 4090 is 420 watts on a 384-bit bus. What is the 5090 going to be with 48 more SMs on a 512-bit bus? I know one thing. That cooler is going to be MASSIVE.
Maybe and most probably they are gonna do this. GB202: RTX 5090/Ti,5080/Ti. GB203: RTX 5070Ti and might be RTX 5070. GB205: RTX 5070, RTX 5060 Ti. GB206: RTX 5060Ti,RTX 5060. GB207: RTX 5050.
By looking at the rumored specs, it feels like the 5080 will be around 1200usd and the 5090 will be around 2500usd, that leaves room for 5080Ti and 5080 Super.. my thoughts.
nvidia anouncing next gen already reminds me of intel when they hit their peak and began giving us only small 10% upgrades. It will get worse i promise you they already had their 6000 series rdy before they launched 5000.
I’m still on a 3060Ti and probably won’t be able to upgrade for at least 6 months. Maybe just like with the 9000 series CPUs they’re working on cutting down power consumption while increasing performance a little and maybe redesign architecture a bit to use AI processing more and optimize performance in other ways besides just adding cores (like faster RAM and larger cache or faster bus speeds ect) We’ll just have to wait and see
are we talking about NUC sized systems? If so, then heat would be a consideration. Whatever card you put in there, best have a way to exhaust its own hot air.
Why are you mentioning the Auto overclocking feature like it's a new addition. It LITERALLY says right in front of you while you speak about it that it is a returning feature and it's something that the Geforce Experience has been capable of for more than 2 years. I believe the auto overclocking features were added specifically for the advent of RTX GPU in particular.
I get you're in a rush to get this content out, but at least try and read the articles you're copying for content and reiterate the content with your own perspective before uploading a nothingburger statement.
On the nVidia part of talking about the next datacenter chip. Well you have to order your cards years in advance. With order forms for a chip that isn't out yet in some cases being common. Mostly for the cloud sector not the local to the data center compute. Yes there is a difference but in this it mostly is on how many cards are bought and used 24/7/365 at or close to 100% in custom designed for the programs it is to run computers. So most likely they already sold every blackwell chip for the data center they are planning on making in the first few years.
Qualcomm CPU has no where near the power required for serious desktop or general pc use. Notebooks and low power units they would be ok. The limited instruction set requires more instructions to get things done so they BS to people a bit with their instuctions per sec etc as it is irrrellevent. Also windows basically runsin emulation mode on these CPUs so will never be as fast (lol) as X86 . THey would need to rewrite vast chunks of the OS to truly take advantage.
The ARM Snapdragon means nothing to a PC desktop user until or unless it can do all the PC apps and games. I am hoping that will be the case in the future. It would be really cool with a new form of CPU as in ARM for PC gaming/desktop. I mean those suckers if they actually make use of being able to push more power for the performance. It could be beastly in performance. It's not like they need to sit with some power budget of a laptop. Of course that doesn't mean we want some heat and power demon.
I've seen this whole song and dance with X86 is dead! ARM and RISC-V are gunning for X86, but all X86 needs to do to compete is drop a ton of legacy ISA. I think Intel is doing something like X86S and I wouldn't be surprised if AMD isn't doing something similar.
Ah yes how is Arm supposed to beat x86? Unless x elite beats x3d gaming at i3 prices, or i9 productivity at i3 efficiency barely anyone is going to buying it if they value their money.
It depends on what you use it for. If you want to max out all settings in all games and play on a 1440p+ panel, then yes, the 4090 is a good option and future-proof for a minimum of 3-4 years. It's also good for video editing and rendering. Compared to the 2080 Ti, it's a considerable jump. Previously, I had a 2070 and upgraded to the 4090. I feel the difference massively.
Only company capable of ditching everything and starting fresh, is Apple. And unlike other companies that only claim they can too, they’ve actually proven to be able to do it.
I'm enthusiastic but for me it means quality. I expect that graphic card gives good performance below 200W. Price is not a problem. Problem is heat, noise level, unstability or short lived components.
They can't keep on upping core count, since it affects power draw, cooling requirements and physical size. There will be an upper practical limit, and after that improvements will come in other ways, such as optimized logic and such things. The 4000 series was already hitting that practical limit. You can't just keep on making GPUs bigger and bigger, cases won't take them. Having the same number of cores does NOT mean performance will be the same. The new cards are going to smaller, meaning they are going to be more efficient and more practical to use than the last gen, which had a lot of issues relating to power draw and size.
RDNA 4 and Battlemage should launch later this year and are said to be competitive in the midrange segment. I don't think drivers would be nearly as big of a problem for Battlemage as they were for Alchemist. After RDNA 4 and Battlemage, RDNA 5 and Celestial should be expected to launch after that, hopefully not too far after. AMD is expected to return to the high end and enthusiast class segment with RDNA 5, and Celestial shouldn't be too far away from launching then as well. They reckon Intel will compete in the high end and enthusiast class with Celestial, which should bring the heat to Nvidia and AMD, so hopefully there will be strong competition in the gaming GPU market within a few years. There is still hope.
First off I have been following you for a really long time and have wanted to say I am ok with what ever form of video you put out even if it is a talking head. You always do a great job putting together the info you are sharing with us.
Since the GB202 is such a massive difference to the GB203 I have a feeling the 5080 will share that with the 5090. It wouldn't make sense otherwise. They would want and need something for yield. All the failed full or almost full 5090 could be made into 5080s. The only thing speaking differently about it might be the 512-bit bus It would depend on memory configurations. What would it be? 16GB/20GB? for 5080. Does that even work? Think about it there would be a bigger gap between 5090 and 5080 by a LARGE margin more than 4090 vs 4080 that was already close to 30%. I mean it would be 50%. The 4090 would still be faster than a 5080 or could be considering still 84SM vs 144SM.
In one way I love that hardware slows down. It means that devs will have to rely less in future hardware to run their games well and start giving a damn about performance
That Nvidia 50 series news is great! When 50 series releases, 40 series will go down in price. And since the hardware is essentially the same, and the only difference will probably be software-locked AI nonsense like DLSS4. You will essentially be able to buy a 50 series card for half the price.
Well the new architecture and faster ram with hopefully lower power/lower cost might be good? Sadly it will probably be the same ram on all cards other than the 5090.
I have a 4090 and I do not like what Nvidia is doing, if rumors are true. Them releasing the 5080 to drive sales for the 4090 is something I wish they wouldnt do but it seems to be the case. They better not release the 5080 first. I would be so pissed.
We can clearly see just how much Nvidia has been holding back from "gamers" with their industry focused AI accelerators. Why don't we move to the SXM socket type? SXM5 can deliver upto 700w, has more than 10x the bandwidth, and it has no need for power cables from the PSU. It can deliver its full 700w right through the board. This makes for much more efficient, and compact cooling solutions. At scale, the cost difference would be negligible.
Sounds like Nvidia scrambling because of MI300 and MI350. They may have to trash a generation because AMDs chiplet design is leaps more powerful and interchangable then their dedicated motherboard GPU design.
The only new feature I want from nvidia is for my 3090 to wake up after my pc goes to sleep. The rest of my pc wakes up, but no display. Its not my monitor cuz my laptop soesnt have this problem when its hooked up to it. & yes, Ive tried updating the driver... several times.
Not good for X86 should have their warning when M2 came out. Seriously, do both think they can rest on their laurels or they are the only competitors to the other? Apple gave the roadmap to Microosft.
I often say that the choice of an operating system is the applications that it runs. I saw the snapdragon presentation. It looks like if you only run Microsoft office apps, snapdragon is for you. Will other developers release applications, maybe, but this is not the first time Microsoft has tried to replace the x86.
Honestly, people shit on Nvidia, and for good reason... monopolies are annoying, but there's a clear reason they have a stranglehold on the market. If you have the money for it, it's great
Really Nvidia, you stole EVGAS program. Also the only new thing on Nvidias video cards is GDDR7(not even x), how is this going to be better? I really hope AMD is pulling a fast one on all of and mostly Ngreedia
Mmmm, I'm not surprised Nvidia hasn't kept up with the rumoured specs. They have no need to, particularly in the lower end. However, the fact that GB202 doesn't seem to have been nerfed also seems to be the same play they did for 40 series, make the flagship sooooo much further ahead in performance that it makes it better value than the next SKU down (5090 -> 5080).
@@tstager1978 then it will be a lot of electrical trahs in future, what is not acceptable and it is not a system which I would like to have. But thanks for your reply!
If your new to the current state of the computing market, then you would be surprised at Nvidias tactics. In a nutshell the tactic is to introduce superficial upgrades while maintaining or increasing the price on said hardware.. IE. 5090, 5090 Super, 5090 TI.. basically a rebadge and resell.
This is a really weird way for Nvidia to say that poor people "aren't enthusiastic" 😂😂
Also Nvidia trying to make my 3080ti look like a 1080ti and I'm here for it. 😂😂
fr
Like BSG: True believers… 😁
It's not like the competition is any different 😅
You are misrepresenting what enthusiast means.
They have been saying x86 is done for 30 years… Not going to happen.
I'm starting to think it's sponsored propaganda
While x86 probably won't 'die out' I wouldn't be surprised if ARM chips make decent inroads in more sectors than just laptops and tablets like they have in the server market.
30 years ago was the early 90's.
@@matthewelisha8797 x86 came out in the 70's and ARM in the 90's so it's almost like x86 has been dying ever since arm dropped their first chips =DD
Unless you live under a rock, you should know that it's actually happening this time.
I really hope AMD does something incredible with their GPUs during the next years, and force Nvidia to actually do better to make sure GPU competition isn't like CPU competition was before Ryzen
I think it's actually a different problem now.
Actually making full use of a 4090 even in 4k gaming, you're typically CPU-bottlenecked unless you go out of your way to turn on silly Crysis-type settings.
If the 5090 were twice as fast as a 4090, you're looking at a card that's only worth it for VR, at hardware resolutions that haven't even released yet.
So Nvidia releases the best gpu 😂
@@pondracek Then the 5090 should be same performance as 4090, but that isnt the case, the 5090 seem to be 50% faster than 4090 which negate your reasoning above.
The only reason NV is doing what they are to their RTX 5000s series is greed. Nothing else. Only those who pay big dollar will get the performance jump we wanted
@@Gattberserk ?
Because of the 4090, there is now a new free 3dmark benchmark.
The card itself is 90% more GPU with 50% more memory bandwidth than a 4080. Yet the reviewers average their reviews out to a 20% gain at 4k over a 4080, even less at any lower resolution. It's just very hard to actually find a GPU-bottlenecked test to actually throw at it.
A 5090 at only 50% faster over a 4090 would be less than a 10% gain at 4k on those same benchmarks, with an asterisk when they turn on path tracing in Cyberpunk at 4k (and ray/path tracing itself has a separate CPU hit anyway).
And for some reason they're talking about the non-cpu bottlenecked use case as the exception and not the rule. (Putting an asterisk, around frame gen is valid. Putting one around like-for-like ray tracing against a same-gen nVidia card is not.)
@@Gattberserk My reply was for some reason obliterated.
The 4090 is already 90% more gpu and 50% more memory bandwidth than a 4080, with reviewers saying it has a ~20% average advantage at 4k on existing ultra benchmarks.
If you can't see that that's bottlenecked, I don't know what to tell you.
Nvidia has zero reason to make the 50 series any better than the 40 series. Until AMD and Intel can give them competition they will not care.
True
Well it’s gotta have a decent bump in order sell them 🤷🏻♂️
sadly true
Intel used to think that way
yes "new" 50 series its a pure refresh of only the name other than the top tier...... paper lunch....
I need an AI that gets rid of all AI
Agreed, AI for Recall/spyware/adware blocker/remover
Good luck getting the terminator on this
@@Airwolf2030 it's easy to avoid Recall, just get Linux, it's actually worth the experience. some Linux distros even come with systemwide adblockers built-in, like Garuda Linux. Linux isn't targeted by a lot of hackers either, so it's really not worth to care that much about viruses, as it's very unlikely you manage to get any viruses at all anyway
This makes me think about an antivirus that also counts its self as a virus :-p.
To me, it's called Linux
I'm just glad to see AMD is still supporting AM4 through 2024 which makes it still a great budget baller for 1080p and 1440p gaming. 😁
AMD 4k ❤ I have a 5950x I do gaming and content creation... I have a 6950xt as well...and I game in 4k
@@mwk1939 Nice but I wouldn't exactly call a 5950x a "budget" CPU, especially when you can get the 5600 for $120 and the 5700x 3D for $199.
@@TheHangarHobbit what about the 5900 XT which will be available i july? AM4 is still more budget friendly, even for "high end" desktop builds. just getting into AM5 at all is easily more expensive than building a decent "high end" AM4 build with comparable performance
@@lassebrustad Problem is that 7800x3d plays sooo much better than 5800x3d in UE5 games, like 15-20% better. Not sure about 5900, but I just do not see a future for it 5 years from now. My rigs last 5-7 years so I am due for a makeover. Soon be breaking out the popcorn and waiting for a good deal, maybe on a pre-built.
"5000 series are getting nerfed"
Is this Nvidia's response to AMD not making high end gpus next gen?
Nvidia will never get a place in my PC..
Horrible company..
Exactly......................its a true sh.t company like Apple!
I love it and my 4090
If you can afford a 4090 then of course you love Nvidia@@dieglhix
4070 Super is the best product of a ripoff generation. Fake pixels, fake frames, stuttering ray tracing at dog-slow fps ... Of course its low power its not really doing more than 720p upscaled most of the time! And yet NVidiots worship this company!?!?!
Then your pc won’t be the best
i cant believe that in 2024 a triple slot 305mm gpu is considered sff
I can fit it in my cooler master nr200 mini itx case…
Though, I know my case is just barely itx, it’s the case that convinced many people to downsize…
i feel Geforce 256 old atm
fury nano is the best
Not until they fix native arm support for software, right now there are many issues for games especially...
You need the hardware adoption to happen for the software development to take place. Hopefully MS will stick with ARM this time around but I'm much more interested in a nice all day battery life laptop with Linux on it.
Win 12 will have native ARM support.
@@shanemitchell477 Yes but the apps do not...
**not including gaming....
the Snapdragon X's are only 40-50 percent faster when compared to the Intel Core Ultra's and AMD Ryzen HS's... those are laptop/mobile CPUs and while that is impressive for apps, editing, coding/programming, graphic design, video production, etc.... it's not anywhere near the power of even mid-range gaming CPU/GPU combinations in a full-sized PC
THIS is SAD news for the 50 series
nvidia bullshitting us, waited this long for the 5080 and its looking like i might as well cut my losses and get a used 4080 smh
just wait a bit longer and see the benchmarks for yourself. Im sure a 5080 is gonna match a 4090. A lot of these videos are make up, click bait shit.
AMD just announced new x86 chips, including two new AM4 chips. No, it's not "the end of x86". I can see myself grabbing an ARM-based system for something like Plex, but I'm not sure what other use I'd (personally speaking) have for one at this time.
Also, let's not act too surprised about the 50 series. Nvidia has gone all-in on AI, and it shows. Now, watch as the 50 series is just as overpriced, if not moreso, as the 40 series.
I wouldn't mind a low cost low power mini pc like arm device for tinkering with in my homelab. Thought about grabbing the Qualcomm dev kit machine but that's a bit more than I want to pay for a box to tinker with when I don't actually do software development.
I mean when you realized that gaming has become more and more worthless over the years it doenst really matter much too.
That's bullshit. I'm sure they will cost significantly more
The RTX 5080 is just a 4080 Super Super.
A five years old CPU with (as example) 6 cores compared with today’s 6 cores CPU is working different and efficiently as the old one. So the same amount of cores are not really proving less power or performance brake. But sure for each customer it is our wish to have more cores. But the focus should be the smoothness game experience without matters of the cores.
Yeah, people get "sticky" expectations.
I've got an 8 core.
Even though a new 6 makes it look slow, I still kinda want a new 8 core minimum.
If someone came out with a 100ghz single core that made all the current chips look slow, most people wouldn't want it.
@@myne00 If that 100ghz single core draws less than 200w then it'd be an instant buy for me.
@@myne00theres a balance between core count and speed. The problem is that 1 core can only process so much information at a time. Even if it's fast, it can only do a single thing at a time. Would you really want games to stutter all the time?
@@doge151. Technically a 1-core CPU at 100 GHz, assuming similar IPC to the current CPUs, would be twice as fast than an 8 core, in all workloads, as long as it's not memory bound aka waiting for data to arrive and doing nothing in the mean time. But that would require that programs (and games) to be coded a bit specifically for it, to not have excessive amount of threads.
Kind of like having a program or game work well with multiple cores also needs a lot of code and architecture of the game engine restructuring, to properly benefit from having multiple cores. But if the OS is configured to have quick switches, not have too many background threads ... well threads in general, then the 100 GHz would blow existing CPUs out of the water. Because in some workoads, those that aren't well (or at all) paralelized (like, say, Starcraft 2) it could be as much as 16 times faster than current CPUs, no matter the core count.
What I wanted to tell is that the current software might not fully benefit from it, but if coded for it (and single threaded coding is much easier than multithreading one. Also has less overhead) then the "can only process so much information at a time" quickly dismantles. It's really a bad argument.
@@doge151. Lol spoken like someone born this century
I see Qualcomms new chip as their M1 moment. Massive performance on the table with the first gen, just like AMD did with Bulldozer to Zen and then will be followed by incremental updates just like the M series chips. x86 still has a lot of life left and will go on for quite some time.
So crazy how gunpowder was invented in China, went all the way to Portugal, then got introduced to Japan. Nearly made a full trip around the globe to get next door.
I guess they didn’t care about pollution!
Yes, and the English build their Empire on that gun powder.... and them made a book with propaganda against the Portuguese who were the ones saving the Samurais selling then guns.... instead Netlix made a good Series called Shogun on that propaganda book were the good ones were the English who were atacking the Samurais and the Portuguese who were trading with them were made the bad guys somehow.....
@@thepro08 It's literally fiction but okay.
Huh, bot comment from first encounters video
What dat got to do with computers
X86 IS HERE FOREVER, NPUs cannot replace X86 since its only meant for laptops, and no Copilot cannot and does not work natively in Desktop CPUs since you need a minimum of 40 Tops for Copilot to work natively with no internet, thats why NPUs like the AMD AI 300 with 50 Tops and Snapdragon Elite X with 45 Tops where Introduced.
Theoretically aren't desktop GPU better at handling AI tasks?
'Needing' 40 tops has already been disproven, people have gotten recall running without NPU's though some functions were obviously way slower without an NPU.
Copilot also works with GPUs.
@@nadtz Recall is in the Copilot App, which does not need to run natively if connected to the internet, so no need in the NPU since its running from the Clouds NPU server.
@@tstager1978 It could but does not, cause Microsoft wants it that way.
Despite the same core count, isn't the 50 series is supposed to have a whole new architecture?
In that case im keeping my 4090 until 60 series's.
same lol 5080 will probably only be 10% faster than a 4080. Remember the 4090 has a huge lead over the 4080/super like 20 to 30% in performance
please do yourself and don't buy VR because you will want more power
Yeah just like I thought due to no competition and the main focus being AI and servers. Unless you want a flagship 90 class GPU there is no reason to buy new GPU’s.
Navidia Is That Expensive Bottle Of Wine That Doesn't impress me, Amd Wine is great for much a affordable price
BTW I built a SFF with an AMD K6 chip and an ATI videocard long long long ago. It was even one of the first to use heat pipes I believe. Very very tiny psu in it as well. I forget the name, my parents wanted one to fit in a tiny spot.
why up the core count when you can simply up the price instead
Snapdragon on desktop is a non-starter! AMD’s Strix Point will destroy it bc Strix doesn’t need emulation. RDNA3.5 & 50 TOPS NPU will annihilate Snapdragon on gaming.
the Snapdragon doesn't need emulation for everything tho, but anything that isn't optimized for ARM yet, just like it was when Apple made their M1 chip, except that Apple made their impressive emulation. I'm not an Apple fan, but they did something very impressive anyway, even tho it was far from perfect
as a consumer you should want Qualcomm to be successful here. competition drives innovation and reduces prices for consumers. Look how cheap things got when AMD got competitive compared to before. Now imagine another player get involved and if ARM takes off. even more ARM players will come on board
It is a start and hopefully it will get better.
@@percy9228 According to some leaks I saw recently AMD is designing their ARM cpu so I think it will take off in next few years. Nvidia most likely too.
@@jawohlbxb3534 Yep, keep seeing Nvidia and AMD come up in regards to possible ARM CPU's for desktop (probably servers also), looks like Intel are the only ones not doing so at this point.
Does anyone else think of the half life sciencist that says "OH MY GOD, WE'RE DOOMED"
I am really not excited at all of having ARM chips soldered to the motherboard, with soldered RAM (not expandable/interchangeable), soldered storage (also not expandable/interchangeable) to the same motherboard.
Those things will just create more e-waste in the world. 😩
been a sub since 5k! so proud to see you grow every day!
So will the 5000 series be a worse release than the 4000 series?
What should we expect from the RTX 5000 with the same core count? +20% ish perf. vs. RTX 4000?
Probably a bit more, 10% IPC (that's what i believe they'll bring to the table) + higher clocks i believe, i think it'll be around 25-30%, but 20% is a realistic guess.
Introducing the RTX 5080 now with 40% more Frame Generation performance than the RTX 4080!
Now on sale for $1,199
Essentially the 5080 will be like 10% faster than the 4080/super. The 4090 already has a huge lead with 20 to 30% more performance compared to the 4080/super
SFF is a great idea. Its why the 4070 is so popular right now (2x form factor and less than 300mm long, also beautiful doesnt hurt!)
My 4070 Ti Super is 330mm long xD
Well that's a weird conclusion to draw from AMD not launching high-end RDNA 4 GPUs, since Nvidia still appears to be making big gains in the highest tier card but then they're not doing much for the lower tier cards where AMD is actually going to compete, AMD is only abandoning the high-end, not the entire GPU market for RDNA 4 so it doesn't make sense that Nvidia is planning to nerf every card other than their highest tier one due to "a lack of competition from AMD".
Trying to stuff high performance GPU's in SFF cases is a fool's errand. Coupled with the 12VHPWR cable and tension from limited cable routing options you're asking for trouble. I have a Fractal Terra I use for video editing. I use a low profile Intel A310 video card. Temps never get above 54C and it's relatively quiet. Just because you can stuff a 4080 or an i9 in a SFF case doesn't mean you should.
that overclock feature was already available on g force experience.
So the 4090 is 420 watts on a 384-bit bus. What is the 5090 going to be with 48 more SMs on a 512-bit bus? I know one thing. That cooler is going to be MASSIVE.
But what about the leaks that the 5090 is going to be a slim 2 slotter? Nvidia has some magic going on.
More SMs, but on N3 instead of N4. I think it will have / need the same cooler
Dude the XTX literally pulls more wattage than a RTX 4090 when gaming. The 4090 is usually in the 350 watt range nice try.
Now seems 5090 will be 448bit bus
So essentially Qualcomm wants to completely DESTROY the PC Desktop industry. Keep this shite for laptops.........
Maybe and most probably they are gonna do this.
GB202: RTX 5090/Ti,5080/Ti.
GB203: RTX 5070Ti and might be RTX 5070.
GB205: RTX 5070, RTX 5060 Ti.
GB206: RTX 5060Ti,RTX 5060.
GB207: RTX 5050.
NVidia should just get out of consumer GPU's. And NOBODY wants the stupid AI/NPU shit in their computers. I'm looking at YOU, Microsoft.
By looking at the rumored specs, it feels like the 5080 will be around 1200usd and the 5090 will be around 2500usd, that leaves room for 5080Ti and 5080 Super.. my thoughts.
It would be funny, if AMD decided to release the 8000 series they canceled after nvidia confirms if those are the real specs.
nvidia anouncing next gen already reminds me of intel when they hit their peak and began giving us only small 10% upgrades. It will get worse i promise you they already had their 6000 series rdy before they launched 5000.
"The END is HERE!" Let's see how well this ages. 😂
I’m still on a 3060Ti and probably won’t be able to upgrade for at least 6 months.
Maybe just like with the 9000 series CPUs they’re working on cutting down power consumption while increasing performance a little and maybe redesign architecture a bit to use AI processing more and optimize performance in other ways besides just adding cores (like faster RAM and larger cache or faster bus speeds ect)
We’ll just have to wait and see
are we talking about NUC sized systems? If so, then heat would be a consideration. Whatever card you put in there, best have a way to exhaust its own hot air.
Why are you mentioning the Auto overclocking feature like it's a new addition. It LITERALLY says right in front of you while you speak about it that it is a returning feature and it's something that the Geforce Experience has been capable of for more than 2 years. I believe the auto overclocking features were added specifically for the advent of RTX GPU in particular.
I get you're in a rush to get this content out, but at least try and read the articles you're copying for content and reiterate the content with your own perspective before uploading a nothingburger statement.
ARM instruction set is so big, it should be considered a monopoly and turned into an open standard.
Very disappointing. I was hoping manufacturers were going to start pushing PCIe 5.0 and taking advantage of it.
On the nVidia part of talking about the next datacenter chip. Well you have to order your cards years in advance. With order forms for a chip that isn't out yet in some cases being common. Mostly for the cloud sector not the local to the data center compute. Yes there is a difference but in this it mostly is on how many cards are bought and used 24/7/365 at or close to 100% in custom designed for the programs it is to run computers.
So most likely they already sold every blackwell chip for the data center they are planning on making in the first few years.
Qualcomm CPU has no where near the power required for serious desktop or general pc use. Notebooks and low power units they would be ok.
The limited instruction set requires more instructions to get things done so they BS to people a bit with their instuctions per sec etc as it is irrrellevent. Also windows basically runsin emulation mode on these CPUs so will never be as fast (lol) as X86 . THey would need to rewrite vast chunks of the OS to truly take advantage.
The ARM Snapdragon means nothing to a PC desktop user until or unless it can do all the PC apps and games. I am hoping that will be the case in the future. It would be really cool with a new form of CPU as in ARM for PC gaming/desktop. I mean those suckers if they actually make use of being able to push more power for the performance. It could be beastly in performance.
It's not like they need to sit with some power budget of a laptop. Of course that doesn't mean we want some heat and power demon.
Snapdragon should just stay on laptop for now
I've seen this whole song and dance with X86 is dead! ARM and RISC-V are gunning for X86, but all X86 needs to do to compete is drop a ton of legacy ISA. I think Intel is doing something like X86S and I wouldn't be surprised if AMD isn't doing something similar.
Ah yes how is Arm supposed to beat x86? Unless x elite beats x3d gaming at i3 prices, or i9 productivity at i3 efficiency barely anyone is going to buying it if they value their money.
Should I upgrade from 2080ti to 4090? I’m not poor. Just need advice.
It depends on what you use it for. If you want to max out all settings in all games and play on a 1440p+ panel, then yes, the 4090 is a good option and future-proof for a minimum of 3-4 years. It's also good for video editing and rendering. Compared to the 2080 Ti, it's a considerable jump. Previously, I had a 2070 and upgraded to the 4090. I feel the difference massively.
@@battleprincess2701 4& editing and 4k gaming. I wa t to do both.
Last architectures were EPIC from Intel and before that Alpha 64s from DEC.
my guess is that theyre saving it for the super and Ti varients.
When there is no competition the end result is current nvidia
Only company capable of ditching everything and starting fresh, is Apple. And unlike other companies that only claim they can too, they’ve actually proven to be able to do it.
I'm enthusiastic but for me it means quality. I expect that graphic card gives good performance below 200W.
Price is not a problem. Problem is heat, noise level, unstability or short lived components.
SD elite wont run ALOT of extensions because ARM is RISC
Nvidia wants to keep shipping to China, they will at least keep xx80 cards neutered for the foreseeable future.
Can't wait to get a 5080 or 5090. My 1080's been great but time to upgrade
I am indeed kind of interested of having a mini-PC with a Snapdragon Elite CPU in it. I won't be running Windows on it though but Linux!
Previous day: RYZEN 9000 IS HUGE! KILLER CPUS!
Literally the next day: X86 ON DESKTOP IS DOOMED!!!
They can't keep on upping core count, since it affects power draw, cooling requirements and physical size. There will be an upper practical limit, and after that improvements will come in other ways, such as optimized logic and such things. The 4000 series was already hitting that practical limit. You can't just keep on making GPUs bigger and bigger, cases won't take them.
Having the same number of cores does NOT mean performance will be the same.
The new cards are going to smaller, meaning they are going to be more efficient and more practical to use than the last gen, which had a lot of issues relating to power draw and size.
Node size decreases so chip size decreases even with more cores. Cooling requirements vary. Maybe one GPU in the 4000 series was pushing those limits.
RDNA 4 and Battlemage should launch later this year and are said to be competitive in the midrange segment. I don't think drivers would be nearly as big of a problem for Battlemage as they were for Alchemist. After RDNA 4 and Battlemage, RDNA 5 and Celestial should be expected to launch after that, hopefully not too far after. AMD is expected to return to the high end and enthusiast class segment with RDNA 5, and Celestial shouldn't be too far away from launching then as well. They reckon Intel will compete in the high end and enthusiast class with Celestial, which should bring the heat to Nvidia and AMD, so hopefully there will be strong competition in the gaming GPU market within a few years. There is still hope.
First off I have been following you for a really long time and have wanted to say I am ok with what ever form of video you put out even if it is a talking head. You always do a great job putting together the info you are sharing with us.
As long as x86 is the best CPU arch to play games on, ARM arch can wait.
Since the GB202 is such a massive difference to the GB203 I have a feeling the 5080 will share that with the 5090. It wouldn't make sense otherwise. They would want and need something for yield. All the failed full or almost full 5090 could be made into 5080s. The only thing speaking differently about it might be the 512-bit bus It would depend on memory configurations.
What would it be? 16GB/20GB? for 5080. Does that even work?
Think about it there would be a bigger gap between 5090 and 5080 by a LARGE margin more than 4090 vs 4080 that was already close to 30%. I mean it would be 50%.
The 4090 would still be faster than a 5080 or could be considering still 84SM vs 144SM.
In one way I love that hardware slows down. It means that devs will have to rely less in future hardware to run their games well and start giving a damn about performance
AMD:
sells better products for the same price
NVidia:
sells the same products at a "better" price
RTX 5000 series rumor news and you showed fake website and Nvidia not mentioned RTX 5000 series.
Thanks!
Intel never released anything significant in the 3-8000s because there wasn’t any competition
That Nvidia 50 series news is great! When 50 series releases, 40 series will go down in price. And since the hardware is essentially the same, and the only difference will probably be software-locked AI nonsense like DLSS4. You will essentially be able to buy a 50 series card for half the price.
If the 50 series flops and everyone tries to buy a 40 series, then the prices of the 40 series will actually go up.
Well the new architecture and faster ram with hopefully lower power/lower cost might be good? Sadly it will probably be the same ram on all cards other than the 5090.
The RTX 5060 will have 12gbs of vram.
@@ZackSNetwork Not according the leaks, it will be using the same memory sizes.
This might be a reason why im switching from Nvidia to Amd Nvidia overprices their products so much and spits customers in the face
4:38 "overclocking for lazy people" i got it, next
NVIDA 5000 series, +10% performance +50% price markup ! 😂👍
I'm jist gonna stay with my 3080 ti for years. Maybe I'll upgrade if 5090 is not too absurdly priced
I have a 4090 and I do not like what Nvidia is doing, if rumors are true. Them releasing the 5080 to drive sales for the 4090 is something I wish they wouldnt do but it seems to be the case. They better not release the 5080 first. I would be so pissed.
We can clearly see just how much Nvidia has been holding back from "gamers" with their industry focused AI accelerators. Why don't we move to the SXM socket type? SXM5 can deliver upto 700w, has more than 10x the bandwidth, and it has no need for power cables from the PSU. It can deliver its full 700w right through the board. This makes for much more efficient, and compact cooling solutions. At scale, the cost difference would be negligible.
Odds are they aren't going to nerf the prices to match
Sounds like Nvidia scrambling because of MI300 and MI350. They may have to trash a generation because AMDs chiplet design is leaps more powerful and interchangable then their dedicated motherboard GPU design.
I get amazed at some of these videos out there. It’s all speculation and rumours people and all about CZcamsrs getting views.
The only new feature I want from nvidia is for my 3090 to wake up after my pc goes to sleep. The rest of my pc wakes up, but no display.
Its not my monitor cuz my laptop soesnt have this problem when its hooked up to it.
& yes, Ive tried updating the driver... several times.
Not good for X86 should have their warning when M2 came out. Seriously, do both think they can rest on their laurels or they are the only competitors to the other? Apple gave the roadmap to Microosft.
ARM86 CPUs are pretty weak at iGPU performances tho
I often say that the choice of an operating system is the applications that it runs. I saw the snapdragon presentation. It looks like if you only run Microsoft office apps, snapdragon is for you. Will other developers release applications, maybe, but this is not the first time Microsoft has tried to replace the x86.
Honestly, people shit on Nvidia, and for good reason... monopolies are annoying, but there's a clear reason they have a stranglehold on the market. If you have the money for it, it's great
Really Nvidia, you stole EVGAS program. Also the only new thing on Nvidias video cards is GDDR7(not even x), how is this going to be better? I really hope AMD is pulling a fast one on all of and mostly Ngreedia
Ngreedia... i like it
Mmmm, I'm not surprised Nvidia hasn't kept up with the rumoured specs.
They have no need to, particularly in the lower end.
However, the fact that GB202 doesn't seem to have been nerfed also seems to be the same play they did for 40 series, make the flagship sooooo much further ahead in performance that it makes it better value than the next SKU down (5090 -> 5080).
Nvidia said it was Rubin 6 months ago by the way
The point of question of qualcom x CPU’s is for the customers if it’s possible to change the CPU in future or the CPU is fixed on the motherboard.
It's bga. Fixed to the board mini pcs.
@@tstager1978 then it will be a lot of electrical trahs in future, what is not acceptable and it is not a system which I would like to have. But thanks for your reply!
@@pju28 I agree totally. I've only seen 1 socketed arm chip and it was Chinese.
I thought lower power was the big deal about the rtx5000s
If your new to the current state of the computing market, then you would be surprised at Nvidias tactics. In a nutshell the tactic is to introduce superficial upgrades while maintaining or increasing the price on said hardware.. IE. 5090, 5090 Super, 5090 TI.. basically a rebadge and resell.
I'm thinking that AMD is going to release a sleaper and intentionally led Nvida down a tick-tock path.