NVIDIA Starts Teasing GeForce RTX 30 “Ampere” Series With Social media Viral

I missed the whole dedicated 3D accelerator phase entirely after getting soured on PC gaming because I got a 486 about a week before the first Pentium landed, after that it was Pentium this, Pentium that.

The first discrete GPU I bought was the FX 5200 (oh god). Things didn't start looking up until I got the 6600 GT, then I was hooked.

 

Do you think people will care if it has those shortcomings, as long as it is less money for similar performance? Again, in these current times many will look for the bargain.

 

well, Geforce256 had the revolutionary T&L (Transform and Lightning) engine, which could be compared to Ray-Tracing - but that was a new thing in Turing, and it's very unlikely that they can present anything groundbreaking this time, so I don't think it's a good comparison from their part

 

21 years and 21 days will be the length of time it will take to pay off your credit card bill for the 3080/90 TI BTW I'm going big Navi I just refuse to pay more than my computer build costs for a Nvidia flagship gpu!!!

 

Since when gpus are supposed to do
something about gameplay? You mean physix?

Where do you see 1080ti at 400$ ? are you talking about stuff used for mining?

 

yes... exactly how?

 

Great new high powered cards but where are the games ? PC gaming is dead on arrival with no new games

 

We are trying to gain UHD stable and Nvidia is at the forefront with this no 3rd party beats DLSS 3.0 currently with infereing support from the Tensor Cores.
This is Psychovisual Optimization and it's a very heavy Research Field compared to all the Simple Algorithms you put on the table here which Nvidia deprecated long time ago in favor of working on DLSS

It will bring 4K Gaming to the masses even on lower cost tier cards scaling up from 1080p/1440p without anyone noticing the difference if correctly Engine implemented and it will even improve the most simple reconstruction implementations in their visual IQ/Performance result.

it gives Nvidia currently a head on of ~3x Hardware Shrinking Generations in Silicon and fill that saved Render overhead with the RT cores.

It is the biggest Nvidia IO/Performance Cheat ever and it's being trained on their Mega Cluster, Hardware training itself.

We see the beginning of ML complementing Hardware inefficiencies or problems where Physically improving them as fast as we used to be because of the rising Quantum problem isn't as easy anymore as putting a small genius team together that train a massive amount of dataset to fix/improve a problem and where even personal costs now get relavated for the later cloud scaling and competitive efficiency.
And Nvidia also improved their internal compression around it additionally with ML.

And we have not yet any AMD/SONY answer to this AMDs approach is Async Hardware efficiency most efficient implementation visible in Idtech7 for now running on it.

This Battle will be very different this time DLSS 3.0 will be pushed on the Front Lines for Nvidia and they see it as a revolution being able to lower the Overhead so drastically to the goal reaching 8 million pixels in Realtime + Hybrid Raytraced for a decent Power output.

And this is what Nvidia Customers gonna pay for and how Nvidia is going to scale it in the Game Visual Experience Context in Price over the cards line of GPU scaling, keep a close eye on Cyberpunk 2077 behavior in the Reviews.

DLSS 3.0 is the first step to make you dependent on Nvidias Cloud structure the first step in their Transition.

AMD improved the GPU significantly the 5700XT captured the 1080 and 2070 Super at lower price point but DLSS 3.0 is another hurdle now, which is much harder to beat without extensive training on really heavy Clusters.

And you will see Nvidia Showcasing their Killer App implementations here Control and Cyberpunk 2077 running decently at 4K on as low as the upcoming 2060 replacement.

With the possibility to upgrade any TAA supporting Game going ahead after greenlighting the TAA 3rd party Engine implementation.

In the End this might become a Fight between Nvidias and AMDs Cluster and training/inferring efficiency Power.

When AMD should surprise us with their own Solution being ready for Launch.

 

NVIDIA Announces GeForce Special Event Featuring CEO Jensen Huang

https://nvidianews.nvidia.com/news/nvidia-announces-geforce-special-event-featuring-ceo-jensen-huang

 

That's on my birthday.

 

I don't see a real bargain to be honest. The 2070S is a better card than the 5700XT, and Nvidia always prices matches in similar performance levels. If there is a driver/feature disparity, the bargain is not there.
Bargain is getting a card like the 2060 with DLSS 2.0, running games with raytracing OK.

Turing was probably the most revolutionary thing out of Nvidia after the original GeForce and whichever one it was that had Pixel Shaders (I believe it was the GeForce 3, not certain, it's been too long).

 

Turing though could be seen as both a failure to their GPGPU Promises or a learning process that more fixed function in tandem will be needed without redesigning everything from scratch in a not doable timeframe

Turings Async disadvantage was the Power Distribution it was a pure Beta Design playing on Cost efficiency, this is fixed now and the final results of this Hardware side fix when everys core work clashes in 1 Frame together will be interesting to see.

This thing brought us DLSS 3.0

https://blogs.nvidia.com/blog/2020/07/29/mlperf-training-benchmark-records/

And it wont surprise me it the DLSS 3.0 Team gets a lot of Presentation time for their Achievements on Sep 1

 

Right now, you may not see a bargain but i meant for what may happen depending on the performance of RDNA2.

Seeing as RDNA made a lot of efficiency improvement over Vega, i'd say it would be good to hold off and see what AMD counters with unless there are features you must have now that wouldn't be available with AMD.

 

No. This hints that the name of the cards will be 2180 and not 3080

 

I don't know what that ML record has to do with Nvidia managing to inject DLSS2.0 in a generic manner. I really hope they manage it, but one thing is not like the other.

RDNA seemed like a pipe cleaner and a way to recuperate some design costs while the "real" architecture (ie RDNA 2.0) was still on the way. It literally feels like a draft someone decided to publish because the book wasn't ready on time.

 

Looking forward to getting a 3080ti so I can utilize the HDMI 2.1 port on my LG 65" C9 OLED "monitor".

120Hz @ 4K, heck yeah!

Going to build a new wood/aluminium PC case for it, that will look very cool

 

Ha. I had a pair of Diamond Monster Voodoo 2's in SLI before going to nvidia as well. and my one deviation has been the 9700 pro also. volt modded the crap out of that thing.

 

Considering there was 3080 on the real leaked cards, this is a nope.

 

This is the thing this war will be for AMD about how fast will they be able to counter DLSS 3.0 inferring it efficiently from RDNA 2 maybe with the Help of Sony R&D they both work much closer together since the whole Playstation 4 Design and share R&D.
Many things we see in RDNA 2 is a corporational effort between AMD/SONY R&D especially on the Hardware side and we gonna see more.
And you wonder what Sony is up to in General here after Cell.
One thing for sure it makes jensen shaking in his shoes ehh leather jacket, and that Samsung "forced" him after the lost lawsuit to buy Ampere production resources he doesn't like much as well

But anyway all this is super beneficial for us consumers we gonna see a nice boost in overall 4K and VR Performance even at the more balanced sweetspot Energy Efficiency/Performance/IQ and more then the General 10 FPS gain + possible additional 10 FPS from crazy overclocking which becomes though less and less as bruteforce efficiency is dieng slowly at those resolutions and more and more compute workloads we see in engines

 

Also i want to add something

We have never been in a better situation of code quality parity for AMD/Nvidia as well since AMD moved to a Scalar architecture from VLIW this is also why DLSS 3.0 is so important for Nvidia as since Nvidia lost the the Consoles completely the code quality/performance for AMD was forced to improve significantly and wee see these results daily improving on the Consoles and being brought over to the PC and Nvidia predicted this long time ago happening
And they upping their exclusivity R&D on it but in the End how their 500 Engineers gonna survive vs million of DEVs can ML be really the answer ?
ML is also nothing exclusive at all to Nvidia so their Dominance is swindling faster now.
They are at the absolute highest point currently but we all know what follows after the PEAK

The catchup to DLSS 3.0 will happen just a matter of time training time to be exact.