Nvidia CEO Jensen Huang on Tuesday morning announced the RTX 3080 as Nvidia’s “new flagship” GPU, confirming leaked rumors and a month of card design. This was compounded by a glut of other hardware, app and software partnership announcements, but arguably the largest number in today’s event came in terms of performance and price.
Huang alleged that the RTX 3080 will “double” the performance of the RTX 2080 GPU, starting at $ 699, with the sale going on September 17. Subsequently, Huang announced the RTX 3070, whose power would reportedly exceed the $ 1,200 + RTX 2080 Ti starting at just $ 499.
Following these, the three-slot RTX 3090 was announced without any price aspirations, but debuted at a price of $ 1499 with a launch on 24 September. This was followed by a sequel reel of fans playing RTX-enabled 3D games recently at 8K resolution and “60 frames per second.”
“Pascal’s friend: updating is safe”
The announcements came with a flurry of Nvidia-handled specs, especially tricked into Nvidia’s performance estimates for each card, so you’ll see in the gallery above how Huang’s performance estimates may vary, which is in question. Varies may vary depending on the software. But in at least one case, it investigates the claim. Although Huang claims the RTX 3070 will surpass the 2080 Ti, one of the charts replaces it as a “1440p, 60fps” card, enabling the card top-bottom with RTX effects – but it is quite honest From is the performance you can expect. Current RTX 2080 TI. Its 4K / 60fps threshold is easy to reach with either RTX effects or a DLSS turn up approach.
In a 45-minute presentation, Huang repeatedly referenced Nvidia’s two-year promotional campaign about RVX technologies and described how Huang called the “final limit to rasterization.” The jump, from Turing to the umpire process, combined with a leap in GPU architecture led to the pitch, with Huang claiming the following adventure: “For all my Pascal gaming friends: It’s safe to update now.”
The 19.8 billion to 28 billion transistor of Temping initiates the ampere process, and also supports what Huang calls “RTX3080 (10GB) and” as a four-level pulse amplitude modulation G6X memory Describes. RTX 3090 (Multiple 24GB). The RTX 3070, meanwhile, will sport 8GB of GDDR6 memory, with no “X” there. The CUDA cores for each are calculated as follows: 10,496 on RTX 3090; 8,704 on RTX 3080; And 5,888 on RT88 3070. All of them are a significant jump from the 4,352 CUDA cores on the RTX 2080 Ti from the end of 2018.
We have to wait longer for a whitepaper to break the RTX 3080 and other technical aspects of the 3090 before the September launch. While we’d love to point out the leaked specs posted on Monday from GPU manufacturer Gainward, these are CUDA core counts, so we’re catching on to those numbers moment by moment. We are also curious to see how the PCI-e 4.0 support of the RTX 3000 line will be a performance factor, and whether keeping anyone on a high-end Intel CPU would harm the previous generation of motherboards that PCI-e Maxes out at 3.0.
The rest of the event information can be found in another gallery, and it revolves around software that will take advantage of RTX-specific technologies. The most interesting of these is Nvidia RTX IO, which promises to streamline the distribution of all types of computing assets directly to the CPU that might otherwise be routed to the CPU. (Nvidia would love to do this for you, as its Tensor cores ensure which type of computing is to be handled in favor of the GPU.
Obviously, we want to know more – and to go hands-on with the results. (Or, in the case of VR on high-resolution headsets like the Valve Index and HP Revere G2, with the results to go eye to eye.) Hopefully, we have good ones to share with ARS Technica readers on that front There will be news. End of september.
Image listed by nvidia