RTX 50 Mobile Lineup Leaked, Nvidia Antitrust Probe, Ryzen 9000 and X3D, and so much more!!!!!!!!!!1

If AMD would just fix their VR performance in their drivers and I would've taken a 6800 XT instead of the 4070 I ended up taking for 50€ more.

Maybe someone should call these vendors that we want better CPUs and GPUs instead of NPUs.

This is not the solution for Recall. Hacker can access to someone's computer and activate Recall. Then return after some time in the future and get access to our data. Recall should not be implemented in Windows at all. It can be an option to download for those who wants this feature on their computers

WHAT THE FUCK THEY STILL WON'T INCREASE THE VRAM FK THAT

Yeah I think gamers need to realize we have come to a fork in the road.

The hardware you need to game, a GPU, is mostly needed for AI processing now. It makes these same companies more money. They have limited fabrication sites. Limited output. They literally lose money right now catering to you.

AMD and Intel are both looking for ways to bridge that problem without cutting into their profit.
Nvidia has publicly stated they are flirting with the idea of completely moving to AI production.
That 17% market share in the GPU space could more than double their profits if they shift it to AI.

Yes, they make money on gamers.
Yes, they make more than TEN TIMES the amount by selling those same chips to AI big data sites. That is not an exaggeration.

Is AI a bubble? No. Sorry. As a programmer of of now 29 years, we have merely hit the very beginning of the exponential AI boom.
The end of that boom is either the end of humanity, or the demarcation point where humanity has been enhanced by AI in ways that we cannot likely even fathom at this point.

Likely merging at some point. We are at the fourth turn now which is why everything is going crazy in politics, because

The goal is the second option.

When I hear that the NPU code takes up massive die-space... I hear that there is a massive amount of compute in their NPU code. Because when you are on a 3nm process, that is what massive die-space is.

So, much as every other microsoft call for unification, this will start slow and once programmers (AI, to be honest. Within a year most programming will be done by AI) realize the potential applications, you will see a massive adoption of the technology.

DirectX came out in 1995, with windows 95. It wasn't really adopted by most game programmers until the late 90's, early 2000s. Even then, there was a massive push towards OpenGL because VoodooFX had better drivers in OpenGL/GL-IDE.

ID Tech 3 was still OpenGL in 1999, and didn't use directx until 8.1b. (After June, 2002) If i remember correctly.
Unreal Engine 3, the other major engine finally officially adopted DirectX in 2004.

So it took nearly a decade for Microsoft to take partial share in the industry, with massive hardware support. (And if anyone remembers,
This may annoy Lisa Su because "Where is my results to showcase now Microsoft?"
But in time this will pay off massively.

Thank you for the news as always ❤

Xbox Handheld with the Gamepass and the 360 library would be insane.

its gonna be more than 15 percent hit trust me

Thanks for streaming all this interesting news together. Complete with solid insights!

Why would Nvidia give you more memory on the low end when you keep buying them? Memory is dirt cheap but die space to address it isn't. (laughably narrow buses) Nothing is going to change until you stop buying them. If you don't want AMD I get it, I've had more than a few bad experiences with them, but buy your GeForce used or something.

How do Microsoft force the "recall feature" on us, is it forced in windows 11 update or something?

Nvidia is torturing their customer with vram deficiency

GDDR 7 vs GDDR 6

5080 16gb vram
5080ti 24gb vram
5090 24gb vram
5090 ti super 32gb vram

Disappointed in the laptop vram if you want 16 or even 12gb those pcs are so expensive. I think low end should be 8 with the choice of 12, mid range 12 with the choice of a 16 variant and high end 16 all the way.

funny i thought amd and microsoft had super tight connections 😎

Glad Micro$py is walking back Recall. Thanks for vloggers like you for spreading the word about this thing far and wide!

Happy Summer Break! I hear your concern about the 9950X3D being more gaming-powerful than the 9800X3D (for example). To me, it just makes more sense that you pay more and you get more performance. I'm not rich, btw. I like low price/high performance, too. But I also want a Porsche 911 Turbo convertible that I may never have ;-) That all said, I'm hoping the reason the higher numbered 3D parts will perform better in games isn't just due to more cache being thrown at them. Sure, I'll take that as a factor, just preferring they have ironed out any scheduling issues.

Nvidia angers me

if microsoft makes something opt-in, they will surely remind of that option whenever it's possible and these reminders will not be easy to deactivate, but that's still better, I guess.

Why would I even consider upgrading if they lock me with low V-Ram and bad software....

Fuck you Nvidia

It looks like people are simply ignorant to me. Why would you keep buying a nvidia GPU with very low V-ram when you can buy a very nice AMD GPU with a great amount of V-ram, it simply doesn't make any sense.

Dear Daniel, please never stop to be how you review tech topics of the pc. You represent at the same time a good friend with good advices and point of view with knowledge. Thanks for your effort and information. Keep on it! Greetings from Austria 🇦🇹

All these AI news, I just have to jump in with some perspective as a former dev. But this will help anyone who works with a lot of data in their day-to-day.

Snapdragon has had NPU for years and years so their developer stack is way more mature than Intel or AMD. On the PC side, only NVidia has more solid dev tools for AI but they don't make x86 CPUs. So Microsoft has to go with Qualcomm to get Copilot+ up and running before anyone else.

Just having the hardware isn't enough. The drivers and software stack needs to be robust enough to be actually useful. AMD don't even have AI enhanced FSR even though Radeon 7000 has the hardware.

What is AI good for? A LOT of stuff on your mobile phones are AI enhanced especially those with Snapdragon SoC. Voice recognition, live camera enhancement (face focus, beautification, white balance, noise reduction, colour grading), OCR of text and QR code, face unlock, video upscaling, battery monitoring & charging.

On the PC, many of these features will become available + potentially indexing of file and file content. I think Recall is great! Can't wait to use it to improve my productivity since I work with a lot of data. OCR happening in the background for my files and content generation will reduce time to search. I may want to have a third monitor dedicated to Recall.

Auto SR is another great feature. CPUs are generally way cheaper than GPU these days. Upscaling of streaming video is another great feature. I use the Video enhance feature in Edge browser to upscale & improve SDR content on my HDR display.

Remember we are still months away from Intel & AMD CPUs with 40+ TOPS to be generally available. That's why the AM5 X870 were pushed back to rush out the CPUs first.

I'm a bit jealous of my wife's 5800X3D coz I'm using the 5900X and it's not maximising my 7900XTX performance. Upgrading to a 9900X3D or 9950X3D will be great for my gaming & productivity so I'd probably wait till early next year once X870 mobos launch to upgrade those components.

Why upgrade my CPU? I expect Copilot+ to come to x86 in less than a year and my 5900X can barely handle Star Citizen too.

And I'm worried about marketing speak too especially now there are Zen 5c parts thrown into the mix. So far the 9000 series is exactly as we expected. If AMD is able to solve the heat problem by moving the 3D Vcache onto the substrate rather than on top of a CCD, we can get high L3 cache + high clocks. That is wishful thinking of course. Without a new Socket, I doubt AMD can move the dies around too much due to die to socket trace length limitations.

It may be possible if the Infinity fabric is upgraded to the latest version of TSMC COWOS but the cost will be crazy. It also seems like TSMC has maxed out their COWOS capacity for 2024. Just my speculations.

As bad as this trend is, the sad reality is that the geforce gpus are still more reliable (aside from the cable failure on 4090s), "trusted" and energy efficient than radeon gpus are, regardless of vram capacity and price. most people dont care about it as they still run games on 1080p medium (or thereabout) and hence buy the "budget" cards, as long as they just "work". At this point, the only ones that can take down nvidia is nvidia themselves.

I like the mouse pointer format

Yeah so sounds like unless you spend a bunch of money gaming laptops are gonna suck

Damn! 25% up from 7940HS!!! Luna Lake had better have a good skymont cores. Only reason to buy Intel is battery life.

If the 70-class has a 192-bit bus AGAIN, Nvidia can get bent. It's gonna be another skip generation. I can't wait to skip it! 👌

AMD would just outright win the battle if they just give tons of vram to the consumers, because Nvidia refuses to

Love watching the videos where real life is actually going on in the background ( kids being loud, like they tend to do lol) Keep it up man!

There's no reworking Recall. The data doesn't need to be encrypted, it needs to not be collected by an A.I. Why give people data to break into when you can just not give them data at all?

2017 was an amazing time to be a gamer - GTX1080Ti came out and AMD came with the banger RX580. Crazy how fast time flies.

It's fine guys, DLSS 4 is going to include VRAM generation and will be exclusive to RTX5k series GPU's.

The 40 series laptop was soo diappointing. 4070 laptop was 2.4% faster than a 3070TI laptop and had a 8GB Vram on a 128Bit bus downgraded from a 256Bit bus. You had to pay around $1000 more to get the better and more powerful 4080 laptop

This reminds me of a reddit comment i came across where some snob was trying to shame people for purchasing the 4080 and 4090 saying that the 50 series is gonna have more VRAM.....my my...how the turn tables have turned.

I believe I read somewhere that Microsoft has a 2024 exclusive deal with Qualcomm. I cannot remember the source but ya.

nvidia doing anti trust stuff? i just find out that YES they are an anti trust company. and i want to see how much wattages their new GPU line up feeds. will there be more things melting this time besides power sockets? nvidia confident this time is merely their AI "bubble" market. notice i said "bubble".
and how hard is it to make a good AI chip? nvidia advantage with their AI chip is their COMPANION programming API like CUDA which is tailored to make use the best of their AI chip's architecture. and ANYBODY could do the same. if they design their AI chip architecture efficiency matches with the special programming AI that will use it.
there are many developing AI chip company right now in china mainland. one of these days they will get what it takes(like the steps i mentioned), and they will make nvidia AI chip a complete overprice.
the same thing AMD or intel need to do when they are developing their AI chips. proprietary API first, and then design the AI chip according to those.
like any bubble, either could burst or deflate.

Imagine if AMD priced more aggressively. Not feasible I know but a world where the 7900XT was $600 could have been great

Phil can pretend all he wants. Microsoft has made it very clear that they're very invested in pushing gaming further into cloud. All of their purchases, even their low prices on the game pass are preparation for it. I was surprised at their aggressive move with the texture streaming in their first cod release. All working towards normalizing their subscription based nirvana. Even if the first handheld is local, it will only be out of necessity. I won't invest in a platform when I know the 2nd or 3rd version is going to the cloud.

Typically a 5090M is around the specs of a 5080 card.
You can see this 4090M is a 16 gig product, much more similar to the 4080 than the desktop 4090.
Works in the lower stack too the 4070M being comparable to the 4060(TI?) instead of the 4070 desktop with it's 12gigs.
So... 16-16-12 that's probably 5080, 5070 TI and 5070 in desktop space.

aghh 8gb is brutal :(

Can someone make mouse cursor out of Daniel Owen? Static daniel pointing up as the regular mouse. Him yapping when something loads and so on and on. Would be nice.

Look, they need to at least make sure the VRAM is enough and is on par with consoles at least. Maybe 16 GB is too much but 10-12 GB should be the bare minimum even for lower-mid range GPUs. We need to be able to match console levels of textures.

damn nvidia. 🤣
at least give 4070 mobile 10gb vram.🤣

I'm thinking of getting a rtx 4070 powered hp omen 16 this month, so should I get it or wait for the 50 series to hit the markets, I can do only one, meaning I cannot buy it now and exchange it for a new one since the exchange value of products in my country are shit