Blackwell gpus

[Tinek]

A GPU is a pretty good asset I've found. I sold my 4090 for £1699, early November, while I speculated it was still worth a lot. So this next upgrade isn't going to cost me much at all.

Even if it's X2? hardly not much when Nvidia is a monopoly and can do anything that want to that price.

 

[Hostile_18]

Even if it's X2? hardly not much when Nvidia is a monopoly and can do anything that want to that price.

If its more than £2000 they probably will sell out day 1, but long term it wouldn't be viable unless they are giving up on the game market entirely IMO.

For me personally its strictly for gaming only. Having more frames, that you might not even notice, isn't worth it for that kind of premium. I think most gamers would be happy with any top end GPU TBH.

 

[mrk]

If the 5080 is half the price of a 5090 for above 4090 performance

I suspect this will not be the case given it has lower specs than the 4090, the only way it can be better than a 4090 is if it uses this "advanced DLSS" to AI it's way over a 4090, that is only assuming "Advanced DLSS" is exclusive to 50 series. And then we will have a new group of consumers who cry "FAKE POWER!!!!!!!!!!!" who suddenly shake hands with the "FAKE FRAMES!!!!!!!!!" brigade and the tweed wearing "FAKE RESOLUTION!!!!!!!!!!!" croissant bashers.

Or something like that.

 

[Hostile_18]

I suspect this will not be the case given it has lower specs than the 4090, the only way it can be better than a 4090 is if it uses this "advanced DLSS" to AI it's way over a 4090, that is only assuming "Advanced DLSS" is exclusive to 50 series. And then we will have a new group of consumers who cry "FAKE POWER!!!!!!!!!!!" who suddenly shake hands with the "FAKE FRAMES!!!!!!!!!" brigade and the tweed wearing "FAKE RESOLUTION!!!!!!!!!!!" croissant bashers.

Or something like that.

I'm a little out of my depth speaking here, but is generational uplift not a thing, so less can essentially do more?

Spec don't always tell the full story, like a 4090 having +75% cuda cores, but that only translating to +30% performance etc.

 

[brostradamus]

idk if trump is going to reverse/revise the china chip ban, but right now the assumption is that the 5080 should be within those restrictions

 

[mrk]

I'm a little out of my depth speaking here, but is generational uplift not a thing, so less can essentially do more?

Spec don't always tell the full story, like a 4090 having +75% cuda cores, but that only translating to +30% performance etc.

It was, but hasn't been since the move from 30 to 40 series and there is no sign that this will change going forwards.

 

[mrk]

Remember, the more you buy, the more you save

 

[Tinek]

If its more than £2000 they probably will sell out day 1, but long term it wouldn't be viable unless they are giving up on the game market entirely IMO.

As I mentioned in earlier post, neither 4090 nor 5090 nor Titan cards in the past were actually primarily gaming cards. There's a lot of semi-pro and even pro use for these, which is why they have so much vRAM which isn't useful for games at all yet. People like that will be happy to pay even 3k+ because it pays for itself back rather quickly, considering cards aimed purely at pro use are much more expensive as is anyway. Just like scalped 4090 sold, 5090 will sell as long as it's cheaper than pro card and still good enough for (semi)pro work. Gamers will be a minority of buyers and it's never been aimed at them specifically, hence price won't be adjusted to average gamer either. Which is why my bet is on price increase comparing to 4090.

For me personally its strictly for gaming only. Having more frames, that you might not even notice, isn't worth it for that kind of premium. I think most gamers would be happy with any top end GPU TBH.

Agreed. But again, gamers do not seem to be main target here by Nvidia. Never have been.

 

[Tinek]

I'm a little out of my depth speaking here, but is generational uplift not a thing, so less can essentially do more?

Not anymore, no. Physics said enough is enough, now we get very incremental improvements, smaller each generation. It's why Nvidia pushes so hard towards AI solutions now, as there's simply no way for them to push RT/PT to good place with performance brute force anymore.