Tuesday, September 20th 2022
NVIDIA Project Beyond GTC Keynote Address: Expect the Expected (RTX 4090)
NVIDIA just kicked off the GTC Autumn 2022 Keynote address that culminates in Project Beyond, the company's launch vehicle for its next-generation GeForce RTX 40-series graphics cards based on the "Ada" architecture. These are expected to nearly double the performance over the present generation, ushering in a new era of photo-real graphics as we inch closer to the metaverse. NVIDIA CEO Jensen Huang is expected to take center-stage to launch these cards.
15:00 UTC: The show is on the road.
15:00 UTC: AI remains the center focus, including how it plays with gaming.
15:01 UTC: Racer X is a real-time interactive tech demo. Coming soon.
15:02 UTC: Future games will be simulations, not pre-baked- Jensen Huang
15:03 UTC: This is seriously good stuff (RacerX). It runs on a single GPU, in real-time, uses RTX Neural Rendering
15:05 UTC: Ada Lovelace is a huge GPU
15:06 UTC: 76 billion transistors, over 18,000 shaders. 76 billion transistors, Micron GDDR6X memory. Shader execution reordering is major innovation, as big as out-of-order execution for CPUs, gains up to 25% in-game performance. Ada built on TSMC 4 nm, using 4N, a custom process designed in together with NVIDIA.
There's a new streaming multiprocessor design, with a total of 90 TFLOPS. Power efficiency is doubled over Ampere.
Ray Tracing is on the third generation now, with 200 RT TFLOPS and twice the triangle intersection speed.
Deep Learning AI uses 4th gen Tensor Cores, 1400 TFLOPS, "Optical Flow Accelerator"
15:07 UTC: Shader Execution Reordering similar to the one we saw with Intel Xe-HPG
15:08 UTC: Several new hardware-accelerated ray tracing innovations with 3rd gen RTX.
15:09 UTC: DLSS 3 is announced. It brings with it several new innovations, including temporal components, and Reflex latency optimizations. Generates new frames without involving the graphics pipeline.
15:11 UTC: Cyberpunk 2077 to get DLSS 3 and SER. 16 times increase in effective performance using DLSS 3 vs. DLSS 1. MS Flight Simulator to get DLSS 3 support
15:13 UTC: Portal RTX, a remaster just like Quake II RTX, available from November, created with Omniverse RTX Remix.
15:14 UTC: Ada offers a giant leap in total performance. Everything has been increased 40 -> 90 TFLOPS shader, 78 -> 200 TFLOPS RTX, 126 -> 300 TFLOPS OFA, 320 -> 1400 TFLOPS Tensor.
15:17 UTC: Power efficiency is more than doubled, but power goes up to 450 W now.
15:18 UTC: GeForce RTX 4090 will be available on October 12, priced at $1600. It comes with 24 GB GDDR6X and is 2-4x faster than RTX 3090 Ti.
15:18 UTC: RTX 4080 is available in two versions, 16 GB and 12 GB. The 16 GB version starts at $1200, the 12 GB at $900. 2-4x faster than RTX 3080 Ti.
15:19 UTC: New pricing for RTX 30-series, "for mainstream gamers", RTX 40-series "for enthusiasts".
15:19 UTC: "Ada is a quantum leap for gamers"—improved ray tracing, shader execution reordering, DLSS 3.
15:20 UTC: Updates to Omniverse
15:26 UTC: Racer X demo was built by a few dozen artists in just 3 months.
15:31 UTC: Digital twins would play a vital sole in product development and lifecycle maintenence.
15:31 UTC: Over 150 connectors to Omniverse.
15:33 UTC: GDN (graphics delivery network) is the new CDN. Graphics rendering over the Internet will be as big in the future as streaming video is today.
15:37 UTC: Omniverse Cloud, a planetary-scale GDN
15:37 UTC: THOR SuperChip for automotive applications.
15:41 UTC: NVIDIA next-generation Drive
15:02 UTC: Future games will be simulations, not pre-baked- Jensen Huang
There's a new streaming multiprocessor design, with a total of 90 TFLOPS. Power efficiency is doubled over Ampere.
Ray Tracing is on the third generation now, with 200 RT TFLOPS and twice the triangle intersection speed.
Deep Learning AI uses 4th gen Tensor Cores, 1400 TFLOPS, "Optical Flow Accelerator"
15:26 UTC: Racer X demo was built by a few dozen artists in just 3 months.
333 Comments on NVIDIA Project Beyond GTC Keynote Address: Expect the Expected (RTX 4090)
7700XT is expected to perform similarly or better than 3090, which has ~69% better raster in relative performance. If RDNA3 architecture brings this kind of uplift, which it should, then AMD needs to convince people to buy 7700XT instead of used 3090 or new 4070.
If you can buy 3090 for less than $700 on ebay, AMD will need to propose convincing price for 7700XT to attract buyers. It cannot cost more than used 3090, so I expect $600-$650 max. But this is far away, in H1 2023. Things can change a bit.
I will hold off until a game with some insane RT visuals comes out (that I actually want to play). My 3080 is more than enough for rasterization. Maybe they will drop the prices next year. No way they can keep this up after the 30 series is gone.
And hey, what is a "4080 - 12" with a 192-bit bus? What's next? A 4060 with 64-bit bus? lol
If I were AMD, I would launch an RX 7600 XT with "4080-12" performance for $649 and call it a day.
At the most, AMD has a chance and nibbling away at NVIDIA's market share.
Let's not forget that the NVIDIA GPU chip has excellent adoption as a compute engine (AI, etc.) and that NVIDIA's Gaming business is now smaller than their Data Center business.
As for that last point, it has nothing to do with Geforce sales nor desktop GPU marketshare.
Have to wait for reliable tests, checking the technology, but it shows that currently the duopoly, previously cryptocurrencies turned out to be deadly for players and gave huge, disproportionate profits for Nvidia, AMD. Playing on PC becomes an abstractio
If it were to be confirmed, I dont fully understand the producers and publishers of games. They limit the potential number of customers with questionable technology. Currently, RT looks like an on / off in terms of lighting, shadows etc. Yet many titles without RT looked great. Then such a technological leap would make sense - but it looks like a lack of support. It's like you buy a new iPhone and after a year your iPhone 12, 13 is no longer usable...for basic functions (performance drop by several dozen percent). With an update, you lose the usefulness of your phone even in a year or two. You have to switch to new products after the premiere of the new series because the life cycle is extremely short.
Someone will say - great, we have a huge performance increase. And I will say - the RTX 2000 series gave a very small increase in performance in games without RT. There were a few issues with RT games and the performance was mediocre. At the same price over the previous series, barely 7-10%, when the GTX 1070-1080Ti cost more, but the performance gains were much greater.
For example
www.techpowerup.com/review/nvidia-geforce-gtx-1070/24.html what we get for 379$ ;)
Even the next RTX series as the most efficient Nvidia RTX 3090Ti card costing $ 2,000 in 4K with RT in some games we only had 25-50 frames. For $ 2000! If now the leap in performance was so large only with DLSS 3.0 it would potentially mean that Nvidia gets the profits from cryptocurrencies, didnt have to worry about what players thinkin . One market is over* cryptocurrencies - now in if you want to play, everyone "has" to buy a new series.
The better reviewers typically compared a number of games just on pure 3D rasterization. They then add some comparisons with RT on but no enhancements like DLSS or FSR. Finally there are some comparisons between DLSS and FSR, particularly with the handful of titles that support both. With each passing week, there are more gaming titles that support these two technologies.
And we will soon see the additional of XeSS to the mix.
It's not like hardware reviewers are being blindsided by DLSS 3.0.
For the most part, today's graphics cards have enough 3D rasterization performance on a single PCB which is why NVLink wasn't even included on the Ada cards. Super sampling technologies like DLSS, FSR, and XeSS are most beneficial when ray tracing is enabled.
So, stock power limits match the 30 series, though there are strong indication that AIB partner cards will dramatically exceed this. FE coolers are huge chunguses, except for the 12GB which ... doesn't have one? Uh, okay? 4080 12GB should be very, very noticeably slower than the 16GB.
With more gaming titles adopting RT and DLSS/FSR (and now XeSS), the performance comparison now takes multiple charts, tables, and graphs. The old paradigm of "run this 3D benchmark utility to compare scores" from five years ago is increasingly obsolete.
And cherry picking one or two games (MS Flight Simulator, Cyberpunk 2077, Red Dead Redemption, whatever) isn't particularly useful. The better analyses use a battery of games to average out architectural advantages and disadvantages of each manufacturer, whether it be DX11 vs. DX12 or other newer features like super sampling.
I don't know about anyone else here but I happen to play some older titles on newer hardware, including games that aren't always part of a graphics card review.
More interesting will be upper mid rage and mid range. Do not forget that AMD will also offer DP 2.0 ports, PCIe 5.0 interface x16 and x8, as well as more VRAM, for more future-proof products.It seems to me that it's 7800 that will try to be competitive with 4080 12GB, 7800XT with 4080 16 GB, and 7700XT with 4070.
If, when the sub-$500 cards are launched, they can be run on a reasonable PSU that someone with a last-gen PC might have (so probably a 650-850W PSU), then we can tell for sure if Nvidia have lost the plot or not.
Reference 3090 with 350W TBP needed 2 PCIe 8pins and 4080 with 320W needs 3?
Ouch and us Americans think we got it bad....
Also maybe this was mentioned before but the 12GB is a partner only model.
I still hold my position - not impressed.