Introduction

PNY GeForce RTX 4070 leads the company's custom-design RTX 4070 lineup from the front. As a company, PNY has come a long way from a memory products vendor for photographers, to an OEM focusing on professional-visualization graphics cards. The company has seriously stepped up its game in the consumer graphics segment with the RTX 40-series Ada generation. The card in this review is called simply the PNY GeForce RTX 4070, there's no brand extension as such. It is intended to be sold at the NVIDIA MSRP of $600, alongside every other RTX 4070 card we are reviewing for you today. This is because NVIDIA wants to encourage its board partners to have decent custom-design cards at its baseline price, and has given the MSRP cards an exclusive review publication date. Reviews of premium factory-overclocked cards priced above the MSRP go live tomorrow (April 13), which is when all RTX 4070 cards go on sale.



The GeForce RTX 4070 Ada is designed for maxed out gameplay across all genres, at resolutions of up to 1440p with native resolution; or high refresh-rate competitive e-sports gameplay at 1080p or 1440p. 4K Ultra HD gameplay is doable if you can find the right settings (or use GeForce Experience to do so); but it isn't the recommended resolution for this card. There is plenty of inventory of previous-generation graphics cards around this price, but none of them come with new features such as Optical Flow Accelerator, dual video-acceleration, or DLSS 3—a technology that generates entire alternate frames using AI, without involving the graphics rendering machinery, to nearly double frame-rates, which should have a transformative impact on games that are optimized for the feature.

The GeForce RTX 4070 is based on the same AD104 silicon as its "Ti" compatriot that launched earlier this year. While the RTX 4070 Ti maxes out the silicon, enabling all 60 streaming multiprocessors, the RTX 4070 only gets 46, which works out to 5,888 CUDA cores, 184 Tensor cores, 46 RT cores, and 184 TMUs. The ROP count is reduced from 80 to 64, too. The memory is unchanged, though—you get 12 GB of 21 Gbps GDDR6X memory across a 192-bit memory interface, with 504 GB/s of memory bandwidth on tap, that's greater than the 448 GB/s the previous-generation RTX 3070 has. The memory bus widths may appear narrower across the RTX 40-series, spare the top RTX 4090, but this is because NVIDIA has redesigned the memory sub-system at an architectural level, with much larger on-die caches on the GPU. There's still a 50% increase in memory size over the RTX 3070.

The PNY RTX 4070, as we mentioned, features a simple, minimalist board design for those who just want an RTX 4070 at MSRP to install and begin gaming, and don't care all that much about premium features such as overclocking headroom or factory-overclocked speeds. Its compact board design, with a dual-slot cooler, should be particularly appealing to those building compact or SFF gaming PCs. You also get the convenience of a legacy 8-pin PCIe power connector—that's right, NVIDIA allowed board partners to use a single 8-pin PCIe power connector design instead of the newer 16-pin 12VHPWR. The typical graphics power (TGP) being under 200 W makes this possible, as the power configuration is rated for 225 W.