Published on May 23, 2023
Spider-Man invites itself in our tests of RTX 4060 Ti, thanks to ZOTAC, who sent us the graphics card RTX 4060 Ti 8GB Twin Edge OC Spider-Man: Across The Spider-Verse, you will understand, this is a custom version Twin Edge, with a specific bundle, inspired by the upcoming animated feature film by Sony.
The ZOTAC card is based on an AD106 GPU, with 4332 Cuda Cores and 8 GB of GDDR6 memory (18 Gbps). The GPU runs at 2310 Mhz with a slightly overclocked Boost to 2550 Mhz, compared to 2535 Mhz for the Founders Edition, while the memory is at 9000 Mhz. The RTX 4060 Ti 8 GB Twin Edge OC Spider-Man: Across The Spider-Verse is offered at the MSRP price, i.e. 439 euros in France. Will ZOTAC's map capture us in its web?
Two years have passed since the Ampere architecture of NVIDIA, a little more even if we were teasing. The advent of a new architecture was therefore expected, here is Ada Lovelace, the third architecture natively managing Ray Tracing technology. Chronologically, let's remember that in September 2018, NVIDIA inaugurated its new RTX range, with its Turing architecture, followed by the Ampere architecture, two years later. The RTX 2000 used an engraving fineness of 12 nm FFN, for the RTX 3000, NVIDIA had called on Samsung and its engraving fineness of 8 nm. For its new cards, the greens are returning to TSMC, in order to take advantage of an engraving fineness of 4 nm, this new fineness makes it possible to reach a record number of transistors, Turing proposed a maximum of 18.6 billion, Amp� re 28 billion, Ada Lovelace announces 76.3 billion transistors at most.
NVIDIA has already declined Ada Lovelace via 4 models, namely the RTX 4090, RTX 4080, RTX 4070 Ti and RTX 4070. Cards that offer a big evolution in terms of Cuda Cores with 18432, 9728, 7680 units and 5888 Cuda Cores. The new RTX 4060 Ti comes in rough ranks with 4332 Cuda Cores on the counter.
The first four references benefited from GDDR6X memory, a standard common to the RTX 3070 Ti, 3080, 3080 Ti, RTX 3090, RTX 3090 Ti but also to the RTX 4070, RTX 4070 Ti, RTX 4080 and RTX 4090. Remember that the GDDR6X standard allows the processing of more signals at the same time, 4 against 2 for the GDDR6 standard. The RTX 4060 Ti uses GDDR6 memory
The Ada Lovelace architecture is a continuation of the Ampere and Turing architectures, with large GPC blocks. Each GPC block contains six TPC blocks. Each TPC block has two SMs, each SM offers 128 Cuda Cores, one third-generation RT Core and four fourth-generation Tensor Cores. The GPC blocks are distributed around the L2 cache, the quantity of which has greatly increased with this new architecture, an RTX 3090 Ti had 6144 KB of L2 cache, the RTX 4090 has 73728 KB, the RTX 4080 has 65536 KB, the RTX 4070 Ti has 49152 KB and the RTX 4070 36864 KB, which explains, at least in part, why the memory buses, less impressive than on some previous generations, do not penalize the new RTX 4000s in any way. The new RTX 4060 Ti is equipped with 32768 KB of L2 memory!
NVIDIA also announces:
3rd Generation Ray Tracing Cores
With the invention of ray tracing (RT) cores, NVIDIA has made real-time ray tracing a reality in video games. These dedicated processing cores have been specifically designed to handle the most intensive ray tracing workloads. Ada's 3rd generation ray tracing cores have twice the yield for intersections with triangles, increasing MAXIMUM RT-TFlops performance up to 2,8x higher are.
Ada's 3rd Gen RT Cores also implement Opacity Micro-Map and Micro-Mesh Engines, two new hardware units for game acceleration. demanding ray tracing-based workloads used in concert with the new software techniques detailed below.
4th Generation Tensor Cores
NVIDIA Tensor Cores enable and accelerate transformative AI technologies, such as NVIDIA DLSS (which is available in 216 released games and apps), and the new NVIDIA DLSS 3 frame rate multiplier Millions of GeForce RTX and NVIDIA RTX users also leverage Tensor Cores to enhance their broadcasts, as well as their video and voice calls in the free NVIDIA Broadcast app.
New optical flow accelerator
NVIDIA DLSS 3 image generation technology is based on Ada's all-new Optical Flow Accelerator, which feeds pixel motion data from frames data to the DLSS neural network, generating new images on the GPU, helping to speed up performance even in constrained scenarios by the CPU.
Two AV1 encoders
GeForce RTX 4000 graphics cards incorporate two of our new 8th generation NVIDIA encoders, now compatible with AV1 encoding, offering new possibilities for livestreamers, editors and call enthusiasts video.
For livestreamers, the AV1 system improves encoding efficiency by 40%. your livestreams will appear to use 40% more bitrate. OBS Studio, the leading streaming application on the market, is working with NVIDIA to enable AV1 encoding in its next software release, scheduled for October. OBS also optimizes encoding pipelines to reduce the overall load of all NVIDIA GPUs by 35%. Video calling also gets the benefits of high-quality live streaming, with Discord adding AV1 support later this year.
For video editors, both encoders are up to twice as fast, saving creators hours and hours each week. We collaborated with DaVinci Resolve, Voukoder - a popular plug-in for Adobe Premiere Pro - and Jianying - the most popular video editing application in China - to enable this feature in major video editing applications y. Updates will be available in October. And if you want to capture high-resolution content for your videos, GeForce Experience users with GeForce RTX 40-series graphics cards can now use NVIDIA ShadowPlay to capture gameplay in 8K and � 60 frames per second in HDR.