Nvidia made a splash at CES by introducing its revamped lineup of RTX 40 series GPUs. The new “Super” cards offer significant improvements in terms of performance and memory upgrades compared to their predecessors.
One of the standout features of the new lineup is the RTX 4070 Super and RTX 4070 TI Super, which boast impressive improvements. The 4070 Ti Super, in particular, features a higher base clock and 768 additional CUDA cores compared to the standard 4070 Ti. However, the most remarkable enhancement lies in the memory department. The 4070 Ti Super now packs 16 GB of GDDR6x memory with a wider 256-bit bus. The increased memory capacity is particularly essential for running larger or higher precision models, especially in on-device generative AI applications.
Moreover, the 12 GB RTX 4070 Super, while being the least expensive option among the new cards, offers a 20% increase in CUDA cores compared to the original 4070. Despite not receiving a boost in VRAM or a wider memory bus, it comes close to the performance of the RTX 4070 Ti.
Nvidia has refrained from offering direct performance comparisons with its non-super counterparts, instead highlighting the improvements over its older hardware. According to Nvidia’s claims, the RTX 4070 Ti Super delivers approximately 60% more speed than the 3070 Ti without relying on AI upscaling or frame generation tricks. Additionally, the less powerful RTX 4070 Super is said to have similar performance to the RTX 3090.
Both the RTX 4070 Super and RTX 4070 Ti Super are set to hit shelves this month, with the former being available on January 17 and the latter on January 24.
At the top of the lineup is the RTX 4080 Super, priced at $999. It offers modest improvements, such as an additional five percent of CUDA cores and a slightly higher boost clock, compared to the original 4080. Nvidia claims this card can deliver about 40% higher performance than the older 3080 Ti without the use of deep learning super sampling upscaling. When enabled, the new card can supposedly handle twice as many frames as its predecessor.
The RTX 4080 Super, although more powerful, still falls short of Nvidia’s flagship 4090. Nvidia enthusiasts will have to wait for the launch of the Blackwell architecture later this year to discover more details about the successor to the 4090.
The source of the article is from the blog macnifico.pt