Introduction

NVIDIA today launched the new GeForce RTX 4070 SUPER, and allowed us to share with you how cards priced at its MSRP perform, custom OC cards can be reviewed tomorrow. For this review, we have with us the PNY RTX 4070 SUPER Verto. If you’re in the market for a particular GPU and don’t care about the 1-3% performance difference that the more expensive overclocked cards command, you’d want to look for cards priced at the baseline. They are designed to meet the performance outlook of the GPU at a price NVIDIA is quoting, and they’re meant to be installed and forgotten about. PNY is a specialist with professional NVIDIA graphics cards, and with the RTX 40-series, the company has taken greater control over the design and manufacturing of its custom-design GeForce RTX graphics cards. This is reflected in a vast improvement in the quality, performance and acoustic tuning for this generation.

The NVIDIA GeForce RTX 4070 SUPER represents a mid lifecycle refresh for the RTX 40-series Ada generation, with a focus on the high end market segment. This sees the RTX 4070 SUPER which we’re reviewing today; the RTX 4070 Ti SUPER next week; and the RTX 4080 SUPER toward the end of the month. The now 5-SKU strong RTX 4070 series graphics cards are recommended by NVIDIA for maxed out gameplay at 1440p, including with ray tracing; although we’ve consistently found these to be capable of 4K Ultra HD gameplay, including with ray tracing, with a minimal amount of tweaking with the game settings; or using enhancements such as DLSS and the new DLSS 3 Frame Generation, both of which are growing in their lists of supported game titles.

The GeForce RTX 4070 SUPER is a significant specs upgrade over the RTX 4070. While it’s based on the same AD104 silicon, NVIDIA has unlocked 21% more shaders, and 25% more ROPs. The new RTX 4070 SUPER comes with 7,168 CUDA cores across 56 streaming multiprocessors, nearly maxing out the AD104. It also features 224 Tensor cores, 56 RT cores, and 224 TMUs. Interestingly, NVIDIA has also increased the L2 cache size to 48 MB, same as the RTX 4070 Ti. The GPU base frequency is set to 1980 MHz, with 2475 MHz boost, which this PNY Verto card doesn’t change. The memory sub-system is carried over from the RTX 4070 and the RTX 4070 Ti—12 GB of 21 Gbps GDDR6X across a 192-bit memory bus.

The SUPER brand extension only represents performance improvements at a given price-point, there are no new features to be had. This is still the cutting edge Ada Lovelace graphics architecture; which takes advantage of the 5 nm foundry node and on-silicon improvements for generational performance uplifts. Ada introduces a new CUDA core with support for newer math formats and shader execution reordering, which positively impacts ray tracing; a new generation RT core, with support for displaced micro meshes, increasingly the complexity of ray traced objects, and the new optical flow accelerator, which is a hardware requirement for DLSS 3 Frame Generation, which nearly doubles frame-rates by generating entire alternate frames entirely using AI, and without involving the raster hardware. The memory size is generationally increased, but at a narrower memory bus, compensated to an extent by faster memory. This is because NVIDIA has redesigned the memory sub-system by using larger on-die caches to cushion memory transfers, on the GPU.

The PNY RTX 4070 SUPER Verto features a compact, lightweight, and straightforward design meant to maximize case compatibility, with some effort made to keep the noise levels down. It features a strictly two-slot cooling solution, and a card length of 25 cm. NVIDIA has increased the total graphics power (TGP) of the RTX 4070 SUPER over the older RTX 4070 by 10%, to support the extra shaders; and so this card features a 12VHPWR power connector. An NVIDIA-designed adapter cable is included. PNY is pricing the card at the $600 baseline for this GPU.