• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

NVIDIA 4000 Series

banbom44 said:
Any of you 4090 owners heard of a bloke on Youtube called Bang4buck pc gamer? whenever i get a new graphics card i watch his videos but i can never get as many FPS as him sometime he gets miles more than me i wonder if any of you lads can beat him he must have some special sauce his cards always run really cool and boost really well as well.
Click to expand...


I think he needs to take a look at his channel name, a 4090 is bang for buck now?
 
Purgatory said:
Don't worry about it Kaap, I'm still waiting for his so called reviews he's seen that a 4090 beats dual 3090s in certain games that use the 3090s in SLI. So far from my testing and comparing I'm not seeing it even on a 5950x at stock and 3090s at stock. Also have first hand testing now that 2 x 3090s are still faster for my work than a single 4090 and still faster in some games and benchmarks that can use both the 3090s.

He throws out stuff and never shows proof or backs it up in a logical way.
Click to expand...

Exactly. And I am still waiting for him to smash his 3090 with a hammer :cry:

Would not worry about what Grim says @Kaapstad

I mean he genuinely thinks the reason the 4090 is so fast is because it has 24GB of VRAM. No jokes, here is his post I quoted. Maybe after he smashes his 3090 with a hammer, might help if he whack’s his head with it gently :p:cry::D
 
arknor said:
Bang for buck doesn't sound very family friendly :p

I wonder where the term originated
Click to expand...
it comes from military spending on firepower, it literally comes from the idea that if you have nukes you don't need as big of a standing army as a deterrent, bang as in explosives, buck as in dollars
 
Last edited:
What are peoples expectations for the 4080? I have heard some people say it will be almost as powerful, but to me it looks shockingly underpowered compared to the 4090. Then again, I have no idea what I am talking about. Lol.
 
ljt said:
In short - Overpriced and certainly nowhere near 4090 performance
Click to expand...
Oh, so I am not so clueless as I thought I was, lol.

It does make me wonder what NVIDIA's policy will be with all of this. The 4000 series are grossly over-priced. Will they greatly reduce prices later on? I can't see it. Yet nor can I see them selling a lot at those prices.
 
Purgatory said:
Don't worry about it Kaap, I'm still waiting for his so called reviews he's seen that a 4090 beats dual 3090s in certain games that use the 3090s in SLI. So far from my testing and comparing I'm not seeing it even on a 5950x at stock and 3090s at stock. Also have first hand testing now that 2 x 3090s are still faster for my work than a single 4090 and still faster in some games and benchmarks that can use both the 3090s.

He throws out stuff and never shows proof or backs it up in a logical way.
Click to expand...
But there are only 20 odd games that support linked mda.. it's impractical from a gaming perspective. There still aots that does unlinked mda which should work with dual 4090s, which could make sense for someone who is active in the community. The upgrade path from dual 3090s looks pretty straightforward
 
As it was handed over to the developers in game to take advantage of multi GPU 'support' is now can games be bothered to accomodate, not the hardware being the problem IMO. Its also when the card makers limit you from being able to SLI that makes you factor in are they locking people out so you buy other products. As he has shared last gen you could use SLI at the top tier, this gen they didnt. He pointed out its likely because they want you to use thier A6000 etc.
 
I am only pointing this out from a gaming perspective. There's no point hanging on to a 3090 sli setup

Regarding multigpu being handed over to developers, that's completely true and no one's blocking them from using it. It's just that the dx12 development environment is a lot more complex than dx11 while the dev community hasn't matured enough to harness the greater functionality that dx12 offers. We should be satisfied with the fact that games are being released on dx 12 platform and constrain our expectations on fancy stuff like multi GPU support
 
Last edited:
brostradamus said:
But there are only 20 odd games that support linked mda.. it's impractical from a gaming perspective. There still aots that does unlinked mda which should work with dual 4090s, which could make sense for someone who is active in the community. The upgrade path from dual 3090s looks pretty straightforward
Click to expand...

I'm not using them for gaming, it's a workstation for my work and mess about on it with some games, normal use is for work and then MSFS only really. NVLINK on 3090s is pointless for the few games that support it unless one of the games is something you really like a lot.


^... people are trying already to see what they can get out of two 4090s as you see but problem maybe the no NVLINK support for some commercial APPS or in house apps/tools.

We will be trying dual 4090s at work soon too and see if we can get our devs to use the full VRAM 48GB in parallel without the NVLINK as they don't have it like 3090s/ti's and 3090s/ti's can pool memory. So far they see 2 x 24GB blocks and basically limiting some workloads to 24GB which is the issue as we need 48GB and soon more for some projects, so looking like back to Quadros (A series) for some projects and means we need to grab Ampere A series not ADA as again they have no NVLINK too. So looking like we need to soon grab some RTX A6000's ..

Ampere 48GB with NVLINK so 96GB available VRAM (RTX A6000)

www.nvidia.com

NVIDIA RTX A6000 Powered by Ampere Architecture | NVIDIA

Unlock The Next Generation Of Revolutionary Designs And Immersive Entertainment Experiences
www.nvidia.com

Ada 48GB no NVLINK so max 48GB available VRAM (RTX 6000)

www.nvidia.com

NVIDIA RTX 6000 Ada Generation Graphics Card

Powered by the NVIDIA Ada Lovelace Architecture.
www.nvidia.com


Basically they even neutered their A series (quadro class) cards this time with max 48GB available and no NVLINK.

They were even making a big deal about the Ampere A6000 NVLINK recently and selling the 96GB feature and clearly trying to distract from the latest version the ADA RTX 6000 is basically half the VRAM available and no NVLINK a downgrade for high VRAM users ..
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom