RTX 4000 mobile series officially released.

[1610ftw]

22 minutes ago, VEGGIM said:

 

"With the RTX 4070 laptop GPU, this is 2,400 MHz, which are in 11 of the 14 games with an average consumption of well below 115 watts. With the RTX 4060 laptop GPU it is 2.610/2.625 MHz – in 10 of 14 games this clock is already reached at an average of less than 115 watts.

This behavior is fundamentally different from that of the two larger GPUs, which operated at over 150 watts in each game in the test, because they were able to convert the additional electrical budget available via Dynamic Boost into even higher clock rates. The maximum possible clock (2.355 MHz for the 4090 and 2.520 MHz for the 4080) was reached only in Far Cry 6. So the clock was almost never the limit, but with the smaller variants this is the rule."

Nvidia GeForce RTX 4070, 4060 & 4050 Laptop GPU Review - ComputerBase

 

What a blunder - I guess we will see if Nvidia makes a few changes with all the bad reviews pouring in.

 

[1610ftw]

35 minutes ago, Papusan said:

 

 

Nice start with 4070 Mobile. Now you'll may see what to expect from 4060 laptops. What a failure. And what a money grab. 

 

 

Gigabyte Aorus 15X Review: Brawny Raptor Lake, Laggard RTX 4070

Graphics performance fails to impress.

Conclution... It seems like for every step forward; the laptop also took a step backward.

 

What did the review editor expect? He failed to see that nvidia don't want avesome for laptops. No profits in that. They want that you upgrade every year. Small increments in performance for each new gen is to be expected. And the laptop manufacturers hail that. 

 

I think Nvidia made a boo-boo there. They were probably shooting for a mildly disappointing 10 to 15% increase and instead completely missed the boat. At least 3070Ti laptops will fly off the shelves now...

[Shark00n]

This generation of mobile GPUs would be a much easier pill to swallow if the 4090 was a 4080, 4080 a 4070 and 4070 a 4060.

 

Pricing would also need obvious adjusting.

 

They'd never do it though.

 

 

 

I went from moderately excited for this gen and really considering an upgrade (from my 11th gen + 3080 as it can't really keep up with 4K gaming) to a definite skip. 

[1610ftw]

1 hour ago, Shark00n said:

This generation of mobile GPUs would be a much easier pill to swallow if the 4090 was a 4080, 4080 a 4070 and 4070 a 4060.

 

Pricing would also need obvious adjusting.

 

They'd never do it though.

 

They could have just kept everything as it is now but with names adjusted to one tier lower.

Then introduce the 4090 chip in super high end laptops that can go up to at least 250W and preferably more if the laptop manufacturer is crazy enough.

 

That would have been fun. Now there is only some fun in the Intel and soon AMD CPU's but the GPUs are at best meh.

 

 

 

1 hour ago, Shark00n said:

I went from moderately excited for this gen and really considering an upgrade (from my 11th gen + 3080 as it can't really keep up with 4K gaming) to a definite skip. 

 

Before I would have thought that the 4070 would be reasonably cheap and a bit better than last generations 3080 Ti but now it really looks like it is the least interesting card for this generation. As I said somewhere else if one isn't adverse to (for me) hideous gaming looks the Asus Strix G18 looks interesting with 2.5K with the 4080 and I also have to mention the Tongfang/Uniwill water cooling solutions with 16 and 17" screens that seem to be priced quite well.

 

Other than that lots of high priced stuff that does not really add a lot with regard to performance.

 

 

[Papusan]

2 hours ago, Shark00n said:

Pricing would also need obvious adjusting.

 

They'd never do it though.

For desktop cards we will see prices going down. 

 

NEWS

AMD Radeon RX 7900 and NVIDIA GeForce 40 GPUs are slowly getting cheaper in Europe

1 hour ago, 1610ftw said:

They could have just kept everything as it is now but with names adjusted to one tier lower.

Then introduce the 4090 chip in super high end laptops that can go up to at least 250W and preferably more if the laptop manufacturer is crazy enough.

They could just do that with the 4080 Mobile. No need to introduce new SKU. But add in 4090 into the GPU stack was meant to increase ASP. And higher SKU numbers will boost more sales... So

[VEGGIM]

1 hour ago, 1610ftw said:

 

They could have just kept everything as it is now but with names adjusted to one tier lower.

Then introduce the 4090 chip in super high end laptops that can go up to at least 250W and preferably more if the laptop manufacturer is crazy enough.

 

Unfortunately it's a pipe dream due to die size. Its double of ad 103's. Mobo's would have to ditch ton of things to fit that along with the cpu. And MXM gpus rn are small. And horizontal space is a premium. 99% 4090 would only go into workstations. And even so, the quadro A series is for that. 

[1610ftw]

2 hours ago, VEGGIM said:

Unfortunately it's a pipe dream due to die size. Its double of ad 103's. Mobo's would have to ditch ton of things to fit that along with the cpu. And MXM gpus rn are small. And horizontal space is a premium. 99% 4090 would only go into workstations. And even so, the quadro A series is for that. 

 

This is the wrong thinking for an ultra high end card - of course it will only be in a few select laptops. But these days we just pretend that a card is ultra high end in order to then sell many more then if there was a truly high end offer.

Motherboards would not have to ditch a ton of things if they finally went back to stacking things in laptops - HP does this in an extremely compact 16" chassis with 4 memory sticks and 4 full size NVME slots and a vapor chamber so an 18" version of that chassis surely could handle the 4090 chip as most of its additional real estate could go to the GPU and that would be a lot of real estate going from 16 to 18".

 

But yes this will probably stay a pipe dream. As brother @Papusan points out the numbers make a lot more sense with what we have now - pretending that something is special that can be built and accommodated much more easily is a sure fire way to make a lot more money. It helps that these pretender chips can be put into almost every half decent laptop chassis while a truly high end chip would be relegated to only a few units.

[KING19]

7 hours ago, saturnotaku said:

3070 Ti vs 4070 - the only real improvement the 40-series realizes is from the use DLSS frame generation. Otherwise, the two GPUs are pretty much equal, at least when comparing models with maximum power limits (140W for the 4070 and 150W for the 3070 Ti).

 

 

 

I am a Wizard!.

 

Talk about castration at the highest level... Also worse performance on QHD+ than the RTX 3070ti because of it being only on a 128bit bus..

 

Like i said before they're making DLSS 3 and frame generation as the main selling points for the mid range gaming laptops and sadly people are going to fall for it.

[VEGGIM]

1 hour ago, 1610ftw said:

 

This is the wrong thinking for an ultra high end card - of course it will only be in a few select laptops. But these days we just pretend that a card is ultra high end in order to then sell many more then if there was a truly high end offer.

Motherboards would not have to ditch a ton of things if they finally went back to stacking things in laptops - HP does this in an extremely compact 16" chassis with 4 memory sticks and 4 full size NVME slots and a vapor chamber so an 18" version of that chassis surely could handle the 4090 chip as most of its additional real estate could go to the GPU and that would be a lot of real estate going from 16 to 18".

 

But yes this will probably stay a pipe dream. As brother @Papusan points out the numbers make a lot more sense with what we have now - pretending that something is special that can be built and accommodated much more easily is a sure fire way to make a lot more money. It helps that these pretender chips can be put into almost every half decent laptop chassis while a truly high end chip would be relegated to only a few units.

I don't remember a 16 inch laptop that fits a 608mm+ die and 384 bit bus. There was only 1 laptop that did and it had to make so much sacrifices. That was the studiobook one. Even so, look at this thing. 

[1610ftw]

2 hours ago, VEGGIM said:

I don't remember a 16 inch laptop that fits a 608mm+ die and 384 bit bus. There was only 1 laptop that did and it had to make so much sacrifices. That was the studiobook one. Even so, look at this thing. 

 

Yes I know the studiobook. I was talking about the Zbook 16 G9. It does not hold that kind of die but it still has an MXM style GPU and all that memory and storage (4+4). If you would scale that chassis to 18" the HP could use almost the entire added real estate on the motherboard for a bigger die and even free some additional space by going with a BGA instead of MXM solution that would be better suited to handle the 4090 chip and its bandwidth requirements.

 

Again I do not think this will happen but it would be possible to do that.

 

[Papusan]

nvidia joking with their customers. New has to be better, LOOL @Mr. Fox😀 

 

 

 

I expect people accept being scammed... 14:27 The 4070 is a 3070 Ti?

 

Even the blind man Dave is able to se there is something wrong with the new and shiny...

 

[VEGGIM]

This looks interesting.
https://twitter.com/JarrodsTech/status/1628922420898992129?s=20

[SapphiraTriX298]

Yep definitely want to see more data.

[Etern4l]

On 2/24/2023 at 6:05 AM, VEGGIM said:

This looks interesting.
https://twitter.com/JarrodsTech/status/1628922420898992129?s=20

 

BTW the top TS Graphics score for the 3090 Ti on 3DMark is ...  

 

 

Spoiler

25,400.

 

[ssj92]

7 hours ago, Etern4l said:

 

BTW the top TS Graphics score for the 3090 Ti on 3DMark is ...  

 

 

  Hide contents

25,400.

 

considering this is a kingpin card this is about as high as it gets. 

 

will be fun to see where 4090 lands in the right hands 😉

 

I've seen over 20k on 4080 so far which is a good start. 

[Etern4l]

2 hours ago, ssj92 said:

considering this is a kingpin card this is about as high as it gets. 

 

will be fun to see where 4090 lands in the right hands 😉

 

I've seen over 20k on 4080 so far which is a good start. 

 

Not sure. People have scored 25k with a 3090 alone. The OC doesn't look too extreme, and - if the description is factual - no chilled WC or LN2 was used.

 

Anyway, this is mostly relevant for gamers. The 3090 Ti beats the 4090 mobile soundly on specs, and actually in many games and gaming benchmarks at 4K.

 

https://www.notebookcheck.net/NVIDIA-GeForce-RTX-4090-Laptop-GPU-Benchmarks-and-Specs.675091.0.html

 

There is no comparison between the two devices really, apart from energy efficiency: 8GB more VRAM, +50% memory bandwidth, some 20% more theoretical FP performance.

 

https://www.techpowerup.com/gpu-specs/geforce-rtx-4090-mobile.c3949

https://www.techpowerup.com/gpu-specs/geforce-rtx-3090-ti.c3829

 

Nvidia should have allowed at least 20% more power for high-end 4090 laptops. 

[ssj92]

24 minutes ago, Etern4l said:

 

Not sure. People have scored 25k with a 3090 alone. The OC doesn't look too extreme, and - if the description is factual - no chilled WC or LN2 was used.

 

Anyway, this is mostly relevant for gamers. The 3090 Ti beats the 4090 mobile soundly on specs, and actually in many games and gaming benchmarks at 4K.

 

https://www.notebookcheck.net/NVIDIA-GeForce-RTX-4090-Laptop-GPU-Benchmarks-and-Specs.675091.0.html

 

There is no comparison between the two devices really, apart from energy efficiency: 8GB more VRAM, +50% memory bandwidth, some 20% more theoretical FP performance.

 

https://www.techpowerup.com/gpu-specs/geforce-rtx-4090-mobile.c3949

https://www.techpowerup.com/gpu-specs/geforce-rtx-3090-ti.c3829

 

Nvidia should have allowed at least 20% more power for high-end 4090 laptops. 

4080 is good with 175w, it performs well. 

 

4090 needs 225w at least so show it's strength. nVidia would be smart to push a vBIOS update to allow at LEAST 200w to widen the gap between 4080 and 4090 mobile. 

[Etern4l]

1 hour ago, ssj92 said:

4080 is good with 175w, it performs well. 

 

4090 needs 225w at least so show it's strength. nVidia would be smart to push a vBIOS update to allow at LEAST 200w to widen the gap between 4080 and 4090 mobile. 

 

Actually, who knows if that extra power would help, given the severe memory bandwidth limitation. I mean it's a 50% cut vs 3090Ti, so even a bit more vs 4090 desktop. 

 

Electronics Mag 17 scored 21.5K in "beast mode"... 

 

 

[ratchetnclank]

The performance of the 4090 is good for a mobile part but the price is just beyond stupid. It's not worth it any way. Even if I had money to burn, the performance delta between the desktop part in the laptop is massive yet the mobile part is even more expensive somehow?

 

Nvidia been smoking that crack pipe too long. I really hope AMD comes and kicks their arse with the next generation on price and performance because Nvidia have been on top for too long and consumers are getting a fucking awful deal because of it.

[SapphiraTriX298]

I don’t think that Nvidia is the only one to blame.

 

A fully kitted and powered 4090 mobile can still be had at ~$3000 as previously pointed out in this thread. That is notably less than top laptops of the previous 2 years, without taking into account inflation.

 

4080 laptops are coming in around $2500. That $500 delta between the two top graphics cards is probably close to the true difference in cost between the 2 chips.

 

The rest is not Nvidia, it is the Notebook manufacturers. New chassis designs to allow for the Mini LED and other high end screen options, (which at this point require different hardware in the chin of the laptop, not compatible with prior designs), exotic cooling options like vapor chambers, etc (and the desire to make as much profit on them as possible).

 

TongFang, managed to take two models from last year without needing extravagant redesigns of the prior-mentioned manufacturers and thus are able to focus on volume; a single design with limited hardware variation outside of the GPU which cut down added cost, enabling them to undercut the competition in price. They also did not take the premium cost for the 13980hx, choosing instead the more economical 13900hx. All of these features add up.

 

So with these other manufacturers you are paying for the hype, name, higher-end features, and redesign costs, which will likely carry over into the next generation as well. This all based on the idea that people remain willing to spend higher and higher amounts from previous years.

 

Therefore, I do agree that the current laptops in the $4k and up range are overpriced for what they are, and that they will probably see decreasing prices as we move into the summer to parallel what we are seeing in the desktop world. This will be the result of overestimation of the buying power of the general public and increased emphasis on value per performance in the same.

 

I am thinking that the first round of these very expensive laptops will sell out fairly quickly, then the next will sit notably longer, too long, as the number of people willing to fork out that kind of money will drop off.

[1610ftw]

Nvifdia would not be to blame for 5000$ + prices but they would be to blame if they offer their "4090" mobile chip at 4090 desktop prices.

 

They are also to blame for not allowing a higher TGP in laptops - let customers and manufacturers decide.

 

And what is it with crippling of the 4050, 4060 and 4070 - talk about deceiving the customer and manufacturers:

 

 

https://www.notebookcheck.net/Nvidia-s-TDP-values-for-the-RTX-4050-4060-and-4070-Laptop-do-not-matter-for-gaming.698161.0.html

 

I like how they limited power for both Time Spy and gaming so that people get a pretty good idea about the gaming performance while running Time Spy - You think you have a 140W GPU laptop but the max that your GPU is really using is 100 to 105W - can't make that up! 

 

 

[VEGGIM]

1 hour ago, 1610ftw said:

Nvifdia would not be to blame for 5000$ + prices but they would be to blame if they offer their "4090" mobile chip at 4090 desktop prices.

 

They are also to blame for not allowing a higher TGP in laptops - let customers and manufacturers decide.

 

And what is it with crippling of the 4050, 4060 and 4070 - talk about deceiving the customer and manufacturers:

 

 

https://www.notebookcheck.net/Nvidia-s-TDP-values-for-the-RTX-4050-4060-and-4070-Laptop-do-not-matter-for-gaming.698161.0.html

 

I like how they limited power for both Time Spy and gaming so that people get a pretty good idea about the gaming performance while running Time Spy - You think you have a 140W GPU laptop but the max that your GPU is really using is 100 to 105W - can't make that up! 

 

 

There was actually a reason why for that. It reaches its max clocks faster than it can use the wattage.  Remember that computerbase.de image I shared?

[1610ftw]

19 minutes ago, VEGGIM said:

There was actually a reason why for that. It reaches its max clocks faster than it can use the wattage.  Remember that computerbase.de image I shared?

 

Must have missed that indeed.

 

I wonder how they manage to use maximum power in Furmark then and to what effect - heating up the room in winter?

[saturnotaku]

My enthusiasm for RTX 4000 is an inverse bell curve. On the low end, I'm excited to see what the 4050 can do in small/thin-and-light laptops that aren't necessarily designed for gaming. I'm hoping it will appear in models like the Lenovo Slim 7/7i Pro X. The 4060 and 4070 aren't doing anything for me especially since 3070 Ti models will perform the same or better in games without DLSS frame generation while being significantly cheaper right now. The 4080 and 4090 are pretty solid and will be even more so once the discounts start rolling in.

[Papusan]

Moore's law is Dead rambling about Nvidia's newest laptop scam... 56:51 RTX 4070 / 4060 / 4050 Laptop Reviews

 

What a Joke. And the 4080/4090 series laptops ain't less of a Joke. Overpriced scam.