- cross-posted to:
- technology@lemmy.world
- cross-posted to:
- technology@lemmy.world
You must log in or # to comment.
We don’t need faster GPUs as much as we need more VRAM. Double the memory instead of leaving it stagnant again.
It’s not just the lack of VRAM, it’s also Nvidia and their stupidity lowering the bit bus for lower tier cards compared to the last gen counterparts.
I dont understand the VRAM cuts… The RAM fabs have been cutting production because of low prices
I would love more VRAM so that i can have a GPU that can do a bit of gaming and dabble in some AI stuff. 100% agree i’d pay for more VRAM instead of horsepower
More memory means you can do real work with it, and enterprise AI training is a money printer that they’d be scavenging the shit out of with cards that are closer substitutes.
Honestly, the gap between the server parallel compute cards and the home video cards isn’t that large. 24GB on video cards, 80GB for a compute card.
That’s not even two binary orders of magnitude. That’s a narrow window to try to make their money from. Plus, some tasks can be subdivided and run on multiple GPUs, and they can’t segment up the market for those.
Like, in general, my bet is that when for most things that fit the above requirements of fitting in that window and having a task that can’t be subdivided, there’s probably enough room for algorithmic improvements to get two binary orders of magnitude of reduction in memory requirements.
But then you can do work with it, and that’s where the real money is at.
They should all be shipping with 32GB now… AMD is at least seeing the light by releasing some 24gb cards under $1k
Really hope Intel’s next generation of GPU silicon makes it a more realistic substitute - that would actually spice things up a lot, you basically won’t see real competition again until nVidia’s AI training dominance is in someone’s crosshairs
The 5800 will have a 128bit bussy and 6GB gram. The Ti version will only have it clocked higher and be able to actually address all of the ram.
128bit bussy? Say less
64bit?
will we be able to afford it?
Lol no. For AI, research and studio 3d rendering only.
*except Arnold. I don’t need render farm neckbeards downvoting me
deleted by creator
What sucks is the 4090 is an amazing GPU. It’s priced for how it performs (top dog bar none). It’s the lower end cards that are the problem. When a lower end card is a worse value per dollar than the flagship then something is horrendously wrong.
This. The 4090 absolutely crushes this generation of cards, hands down. And while irs expensive, it also feels like a “pay more get more” type deal. Even just dropping down to the 4080, it starts to raise questions about the price. It’s barely a step over the 3090 and 3080, but gets priced as if it’s hitting alongside the 4090
Bro I’m still using my 2060 Super
And I’m still here on my 1060 6GB rocking 40 fps on medium in Jedi: Survivor.
2060 user here as well. Could use a bit of an upgrade, but not much.
I’ve yet to encounter a game my 2080s couldn’t run at 144hz max graphics… Why are we still pumping out $800-1200 gpus when there’s nothing requiring that amount of power?
Must be nice hitting that 2010s backlog. I still got all many unplayed steam games. And don’t even get started with all the free epic games.
nothing requiring that amount of power
My simulators, Pimax, and three 4k 144hz monitors beg to differ.
Congratulations on not playing anything intensive, I guess?
It’s 144hz, not fps. For all we know he could be at like 30fps and it still counts.
Yeah bit you’ve never played Alan Wake 2 with turbo mode ray tracing in 28000p /s
(proud owner of a 2070s still chugging along here)
Recently upgraded both CPU and GPU for better VR performance. Before this my 3070 was struggling a bit with cyberpunk on the 4k TV. There’s also productivity and AI stuff people deal with these days.
There’s plenty of needs that require more and more power.
They still haven’t dumped all those chips they claimed on their last three earnings calls. Expect the 5000 series to have a familiar flavor.
I don’t get this argument at all. Unless you’re buying gpus based on their name (xx80, xx70 etc), 40 series have been consistently cheaper than 30 series at any performance level. Nobody’s going to think “$850 for a 4070ti is too much” and buy a 3090ti for $1500 instead.
$850 for a 4070ti IS too much.
its ridiculous for a -70 series.
And yet it’s still a better deal than the 30 series equivalent
New GPUs are supposed to perform better at the same price point as their last ones. If they didn’t then we’d be moving backwards.
Also the 4060’s MSRP is $30 less than the 3060, and yet it performs worse than the 3060 at times. So honestly we have been moving backwards.
New GPUs are supposed to perform better at the same price point as their last ones.
And they do, so?
What I think people are pissed at is that the prices of some cards have increased to the prices of higher segment cards from the previous generations. And the small generational gains on the middle/low end prices.
If you’re upgrading from a 30 series card 40 series might be a huge improvement, but if you aren’t you’ll still choose the slightly better option. Noone would buy a worse value card just because the newer card isn’t better value enough. There no logic behind the claim that Nvidia is trying to clear 30 series inventory by making 40 series worse. And the idea that they’d try to do that with the 50 series as well is absurd. This feels like people trying to hold on to hope that gpu prices will come down once some factor goes away.
Also here’s a comparison of last gen vs current gen for the interested:
-
4090 - 60% faster than 3090/ti at the same price point
-
4080 - 40% faster than 3080ti at the same price
-
4070ti - 15% faster than 3080ti at $400 less or 25% faster than 3080 at $100 more
-
4070 - equivalent to 3080 at $100 less or 20% faster than 3070ti at the same price
-
4060ti 8gb - a bit worse than 3070 at $100 dollars less or a bit faster than 3060ti at the same price
-
4060 - generally 10% faster than 3060 at the same price. Exceptions aren’t the rule
-
