The Ampere architecture has officially been unveiled, and Nvidia is making a strikingly bold claim about its performance capabilities. The corporation claims that their newly released $549 mid-range gaming GPU outpaces the Nvidia RTX 4090, but one wonders what’s behind this seemingly unexpected announcement.
Nvidia takes centre stage at CES, unveiling its comprehensive range of RTX 5000 gaming GPUs, including the show-stopping flagship RTX 5090, alongside a plethora of intriguing innovations and products. Notwithstanding this, the corporation’s bold statement about the RTX 5070 was perhaps the most stunning.
The NVIDIA RTX 5070 features 6,144 CUDA cores, a 192-bit memory interface, and is equipped with 12GB of dedicated video random access memory (VRAM). The new GPU comparison shows a striking similarity to its current-generation counterpart, the RTX 4070, boasting 5,888 CUDA cores, a 192-bit memory interface, and 12GB of dedicated video random access memory. In the meantime, the RTX 4090 boasts an impressive 16,384 CUDA cores, supported by a robust 384-bit memory interface and an ample 24GB of dedicated video random access memory (VRAM). Despite novel architectural refinements and the adoption of faster GDDR7 VRAM in the RTX 5070, there remains a significant gap to overcome.
Nvidia’s claim that the RTX 5070 achieves “4090 performance at $549” comes with a crucial caveat: it’s only possible thanks to synthetic intelligence. Specifically, the card’s access to 40TFLOPs of AI processing power and Nvidia’s new neural rendering and DLSS 4 technologies enable this feat.
NVIDIA’s introduction of DLSS 4, alongside its latest GPUs, appears to be the key catalyst driving this significant efficiency boost. Utilizing artificial intelligence, this technology generates novel pixels to augment upscaling and rendering decreased-resolution frames into higher-definition ones, subsequently enabling the creation of entirely new frames.
As DLSS progresses beyond its body-era counterpart, DLSS 3, it appears to potentially create up to three additional frames rather than just one. Without sacrificing quality, you can achieve fourfold the perceived value of your work without incurring the costs of a physical body, or double the return on investment compared to previous iterations of your art form.
Nvidia’s presentation lacks clarity on whether their claimed efficiency gains for the RTX 5070 compared to the RTX 4090 account for the latter’s potential use of a single-generation DLSS 3, or if it was only utilizing standard rendering without DLSS?
If that’s the case, the corporation’s declaration is glaringly dishonest. Notwithstanding the assumption that the previous iteration applies equally to the new GPU, the RTX 5070’s raw performance appears to offer a substantial uplift compared to its predecessor, the RTX 4070.
While it’s uncertain what the future holds, one thing seems clear: AMD is gearing up for a fierce battle in the GPU market with the launch of its flagship card, the AMD Radeon RX 9070 XT, which directly competes with NVIDIA’s RTX 5070. Once again, AMD introduces its latest iteration of upscaling and ray tracing technology, FSR 4, alongside its new graphics card, setting the stage for a thrilling competition to determine which GPU will dominate the market in this highly competitive price range?
