Rivdoric
Member
Every few months: "NVIDIA doesn't care about gaming."
Every couple of years: NVIDIA releases a gaming GPU that blows everything else out of the water.
It now also blows itself & PSUs out of the water.
Every few months: "NVIDIA doesn't care about gaming."
Every couple of years: NVIDIA releases a gaming GPU that blows everything else out of the water.
I'll give up PC gaming before I even consider buying one of those.Chineses companies will supply gpus for us.
![]()
China Just Made a GPU That's Powerful Enough for Gaming, But There's a Catch
This 12GB VRAM GPU is about as powerful as an RTX 4060, but it's priced like a 5060 Ti.uk.pcmag.com
Yeah, I don't see Nvidia dropping gaming GPUs, but I could see them only having say x80 as the top model for a while.Percentage wise, but still a significant source of revenue.
Yeah, I don't see Nvidia dropping gaming GPUs, but I could see them only having say x80 as the top model for a while.
Which means gaming is completely irrelevant to them.
Data centers aren't going away; they'll only increase.Yeah, possible. Resource scarcity is really the problem overall and gaming isn't the cash cow data centers are. I'm sure Nvidia would love to be able to sell gaming GPUs as much as they have in the past, but that just may not be possible while this AI data center build up continues.
Data centers aren't going away; they'll only increase.
Yep.
Nvidia: Improving drivers, DLSS, frame gen, Path tracing, Switch 2, etc...
Some people: Nvidia doesn't care about gaming anymore...
hasnt intel given up already?They don't care about gaming for a while, CUDA is the proof of it only for bitcoin farmers
I seriously hope to AMD and Intel graphics crushing them because of this
"give up" maybe is a strong word. I can say that they acknowledge that they can't really compete with the others (at least for now)hasnt intel given up already?
Yes. And they are going to put them in orbit. Energy won't be a problem there. The main bottleneck will be chips.Data centers aren't going away; they'll only increase.
AI is already a global imperative; those who don't embrace it will be nobody.
DGX Spark is widely considered to be a worthless crap.
Purely from GPU / VRAM pespective, it's mostly these things that are widely accepted as go-to choices:
- RTX 3090 (with 2 of them, you can run some strong local models at 48GB VRAM)
- 5090, 4090, 4090 48GB mod (obviously expensive af to get more than 1 card)
- 7900 XTX (using llama.cpp Vulkan instead of CUDA)
- RTX Pro 6000 Blackwell (one 96GB card is enough, super expensive)
- stacking a few RTX 5060 Ti 16GB cards (slower than RTX 3090 by 1.5 to 2 times?)
While all the gamers whine about "AI enthusiasts" stealing their GPUs, those "AI enthusiasts" largely don't give a shit about anything other than a bunch of used RTX 3090 cards (at least when they aren't into image/video generation; for that stuff, 5090 would be quite significantly better).
I don't see a 6080 being almost 2x the current 5080. PS6 is rumored what between 5070Ti and 5080, closer to the Ti?In the end we all know once new nvidia gpu family drops in 2027 sometimes tons of ppl gonna upgrade, especially if we actually compare specs and that 1500$(streetprice, not fake msrp) 6080 is 2x stronger from upcoming ps6, i know i will![]()
3090's still too cheap for what it offers.DGX Spark is widely considered to be a worthless crap.
Purely from GPU / VRAM pespective, it's mostly these things that are widely accepted as go-to choices:
- RTX 3090 (with 2 of them, you can run some strong local models at 48GB VRAM)
- 5090, 4090, 4090 48GB mod (obviously expensive af to get more than 1 card)
- 7900 XTX (using llama.cpp Vulkan instead of CUDA)
- RTX Pro 6000 Blackwell (one 96GB card is enough, super expensive)
- stacking a few RTX 5060 Ti 16GB cards (slower than RTX 3090 by 1.5 to 2 times?)
While all the gamers whine about "AI enthusiasts" stealing their GPUs, those "AI enthusiasts" largely don't give a shit about anything other than a bunch of used RTX 3090 cards (at least when they aren't into image/video generation; for that stuff, 5090 would be quite significantly better).
I don't see how ps6 can match a 5070 Ti with just 160w. And that 160w has to be shared with a cpu. An anaemic cpu, sure, but it still takes power away from the gpuI don't see a 6080 being almost 2x the current 5080. PS6 is rumored what between 5070Ti and 5080, closer to the Ti?
Edit: if 6080 is 2x and has say 24GB+ VRAM, I am going to upgrade as well, no question.
8GB VRAM is complete shit.You will have to buy a Gabe Box and just be happy
I seriously hope to AMD and Intel graphics crushing them because of this
DGX Spark is widely considered to be a worthless crap.
Purely from GPU / VRAM pespective, it's mostly these things that are widely accepted as go-to choices:
- RTX 3090 (with 2 of them, you can run some strong local models at 48GB VRAM)
- 5090, 4090, 4090 48GB mod (obviously expensive af to get more than 1 card)
- 7900 XTX (using llama.cpp Vulkan instead of CUDA)
- RTX Pro 6000 Blackwell (one 96GB card is enough, super expensive)
- stacking a few RTX 5060 Ti 16GB cards (slower than RTX 3090 by 1.5 to 2 times?)
While all the gamers whine about "AI enthusiasts" stealing their GPUs, those "AI enthusiasts" largely don't give a shit about anything other than a bunch of used RTX 3090 cards (at least when they aren't into image/video generation; for that stuff, 5090 would be quite significantly better).
Yeah, the whole point is that 128gb is enough to fit most models, you can run CUDA and the overall inference speed doesn't really matter, it's not meant for that.Spark was designed for developers. It's essentially a replica of what you find in a data center rack, making application development much easier.
You can draw an analogy with the devkits for game consoles.
Like other non profit companies, Nvidia cares about money.
Right now, PC gaming isn't as profitable as AI.
Which means gaming is completely irrelevant to them.
PS6 APU 280mm2? on N3P, that's for the CPU and GPU + SOC with 120-130W power budget for the GPU? While RTX 6080 may use 350-400mm2 die on N3P, 1152GB/s memory bandwidth and 300W TDP.I don't see a 6080 being almost 2x the current 5080. PS6 is rumored what between 5070Ti and 5080, closer to the Ti?
Edit: if 6080 is 2x and has say 24GB+ VRAM, I am going to upgrade as well, no question.
No idea how big of a gap we gonna see, 5080 is only around 15% faster from 5070ti so indeed from 15% to 100% is super unlikely(around 50% would be probable), but u never know- we getting 3nm process node, so could be some srs performance jump(ofc price jump will be massive nonethelessI don't see a 6080 being almost 2x the current 5080. PS6 is rumored what between 5070Ti and 5080, closer to the Ti?
Edit: if 6080 is 2x and has say 24GB+ VRAM, I am going to upgrade as well, no question.
PS6 at best 5070. New Xbox around 5080.I don't see a 6080 being almost 2x the current 5080. PS6 is rumored what between 5070Ti and 5080, closer to the Ti?
Edit: if 6080 is 2x and has say 24GB+ VRAM, I am going to upgrade as well, no question.
9070XT is actually ahead now days of 5070Ti in Raster. The latter is better at RT and path tracing. And even with RT some games (Sony ports and Capcom especially) run better on 9070XT.PS6 at best 5070. New Xbox around 5080.
People said PS6 apu doesn't have same Power as 9070 XT so 5070 is a good given
Physics is gone, there's nothing special left about nvidia that AMD can't do.
you can't have two cards and 1 memory pool for ai, that's not how it works.DGX Spark is widely considered to be a worthless crap.
Purely from GPU / VRAM pespective, it's mostly these things that are widely accepted as go-to choices:
- RTX 3090 (with 2 of them, you can run some strong local models at 48GB VRAM)
- 5090, 4090, 4090 48GB mod (obviously expensive af to get more than 1 card)
- 7900 XTX (using llama.cpp Vulkan instead of CUDA)
- RTX Pro 6000 Blackwell (one 96GB card is enough, super expensive)
- stacking a few RTX 5060 Ti 16GB cards (slower than RTX 3090 by 1.5 to 2 times?)
While all the gamers whine about "AI enthusiasts" stealing their GPUs, those "AI enthusiasts" largely don't give a shit about anything other than a bunch of used RTX 3090 cards (at least when they aren't into image/video generation; for that stuff, 5090 would be quite significantly better).
I think 5070 is a safe bet. Maybe better in Ray tracing and even better with pssa 3.09070XT is actually ahead now days of 5070Ti in Raster. The latter is better at RT and path tracing. And even with RT some games (Sony ports and Capcom especially) run better on 9070XT.
HUB just did a video on this very topic. So if PS6 is going to be close to 9070XT, it's going to be very near 5070Ti.
It's because people see it only makes up a small part of revenue that they seem to ignore the fact that the small part is still tens of billions of dollars yearly revenue.
Last year was 18 billion on gaming GPUs alone. Still a fraction of datacenter sales, but still a staggering amount of money.
Buy 2 and tape them together, 16gb steam machine pro8GB VRAM is complete shit.
9070XT is actually ahead now days of 5070Ti in Raster. The latter is better at RT and path tracing. And even with RT some games (Sony ports and Capcom especially) run better on 9070XT.
HUB just did a video on this very topic. So if PS6 is going to be close to 9070XT, it's going to be very near 5070Ti.
They are also talking about arming the data centers for defense or putting them underground, as fear of drones from "domestic terrorists", ie a "peasants revolt". This was noted by the head of black rock Larry Fink, aka own nothing and be happy fame.Bro, in the States, they are starting to use imminent domain to seize properties destined for data center use.
It's not a wide spread thing atm but it's a growing concern of mine.