Discussion in 'Frontpage news' started by Hilbert Hagedoorn, Sep 1, 2020.
Waw, you do realise there's folks here that know what they're talking about?
5% waw ..
I'd wait for official reviews when it comes to performance.
Where are you even pulling 35% from?
Its using like 50% more power than 2080.
Did you check the link I posted a few posts above ?
The mighty 3080 is barely 35%-40% faster (on average) than a 2080 S, a bit more in 4K, a bit less in 1080p
Let's add the 5% to the original 2080 (non Super), this still makes it -barely- 50% faster than 2080, certainly not 2x (twice) as Jensen advertised !
I mean we already know it's twice as fast in Doom and the presentation slide clearly says up to twice as fast.
That's not to mention that this card is probably bottlenecked in tons of situations. In the other thread people were saying it's only 25% faster based on the entry level timespy benchmark from 2010 lol
The focus was on 4k I believe in JHH presentation.
I want to believe.
Why would they compare the cards bottlenecked? Makes no sense.
Absolutely on point, we're clearly on the same boat here.
True, you dont want to compare gpus while they are bottlenecked by anything - issue is that testing the 2080 vs the 3080 at 4k rather than lower resolutions paints the 3080 in a more favourable light, as it has a wider bus width, which does matter for 4k and higher, while less so at 1440p and lower.
Imo an ideal situation would be testing it in games that aren't cpu bottlenecked at all (doom actually shouldnt be cpu bottlenecked at all), while also pairing it with the fastest possible cpu, and testing it at 1080p and 1440p aswell, which are the resolutions the vast majority of 3080 buyers play at... would give a more accurate reflection of what performance you can expect from it.
Yep Im getting a 3090 strix, and just hoping that the dx12 cpu optimization in cyberpunk will be excellent, so my old 4930k can still do 60 fps In all honesty though, if it is properly multithreaded, it should be np at all - it still pulls well above 60 fps in all other proper dx12 / vulkan titles =D
Some people buy fast GPUs for high-refresh rate, you know ?
If a new generation is not providing a significant benefit to that, why bother upgrading ?
I'm not saying that these new GPUs aren't fast, they are very fast, but they are not AS fast as advertised.
Advertisement = Cherry picked best case scenario.
Reality = Somewhere in the middle.
Nah I disagree, it's like saying the new Bugatti Veryon is nearly twice as fast as the last one, but I'm not bringing it on any fast roads, so it's not, and Bugatti is false advertising it.
Some games simply don't scale in performance. Look at New Dawn - at 1080p the 2060s is almost as fast a 2080Ti. No one should use that as metric for comparison. The game clearly just doesn't scale for high FPS. People buying fast GPUs for high refresh rate aren't looking at Farcry: New Dawn as the goto experience - they are playing games like CSGO, Valorant, COD, whatever the latest is that are designed to achieve high FPS rates.
In benchmarks designed to scale or have the CPU bottlenecks removed, it does scale. 3Dmark Timespy at 4K, for example - it's 1.7x faster than a 2080s. 2080s at 4K .05x faster than a 2080. So overall it's about 1.75x faster than a 2080 - not exactly 2x but pretty close, closer than the 1.3-1.4x you stated.
Digital Foundry video shows Doom doing over 2x, Quake 2 RTX going over 2x, Boarderlands 3, 1.8x, SOTR is ~1.7x and Control 1.7x - 1.8x.
So we have two clear cut examples of it doubling a 2080 and multiple others where it punches within 20% of doubling it. The Nvidia slide clearly states up to 2x performance:
Maybe they are cherry picking the best examples but you're cherry picking the worst examples. I wouldn't necessarily say the card is "twice as fast as a 2080" but if someone said "it's up to twice as fast as a 2080" I'd agree with it. It's definitely within that range.
That´s how you sell a product, you show at his best, every company does the same. That´s why we should wait for reviews before pulling the trigger.
... you mean that "advertorial" ?
Of course, but fanboys will be fanboys. They don't listen to reason.
DF will loose all credibility from me if those claims do not end up being true. Their sponsored videos lately are really cringy. How can you trust someone that got a bag of money to say something.
Well, that´s their problem...
For how much 3090 gonna be better than 3080? From 5 to 10% at best?
It's not just their problem.
You see, with so many fanboys shelling out big $$$ for ultra-overpriced cards like 2080 Ti or 3090, the entire stack gets marked up !
It wasn't so long ago when an x60 card from Nvidia was just $199. Example: GeForce GTX 560 May 17, 2011 $199
Look at them now! $349 ? $399 ? Next year 4060 will cost $499 ?
So many are praising Ampere for being cost effective, but it is really not ! It's the same inflated price as the overpriced Turing !
Pascal was inflated as well, btw ! So was Maxwell... this has been going on for a decade, just up, never down.
Soon GPUs will be more expensive than entire cars if this keeps going much longer.
Of course, they are doing what every business should be doing, milk their customers. That's why the business exists, to make money !
The problem is that people have stopped "voting" with their wallets and just let themselves be milked...
I for one will not not participate in this milking. Will stick with my existing GPU until it dies of old age or the tech has moved so much that I literally can't start a new game with it, on any setting (like it happened with move from DX9 to DX11)