Introduction

Gainward GeForce RTX 4070 SUPER Ghost is the company’s most affordable custom-design RTX 4070 SUPER graphics card making its debut today. It’s priced at the NVIDIA baseline MSRP, and is targeted at those who choose graphics cards based on GPUs, want them at MSRP, and who don’t particularly need factory OC or elaborate cooling. The Gainward Ghost is designed to maximize case compatibility, as it is strictly two slots thick, and about 27 cm long, which should fit most mid-tower cases, even with a front-facing radiator in place. The new GeForce RTX 4070 SUPER is part of a mid lifecycle refresh of the RTX 40-series Ada, the SUPER brand extension denotes increased performance at given price points, using existing silicon, and so there are no new features as such.

The new GeForce RTX 4070 SUPER, much like the rest of the RTX 4070 family, is being recommended by NVIDIA for maxed out gaming at 1440p, including with ray tracing, although our testing shows that even the original RTX 4070 can tackle today’s games at 4K Ultra HD at reasonable frame rates, which you can only improve by dialing down the eye-candy a bit, or engaging DLSS. On supported games, you also get DLSS 3 Frame Generation, which nearly doubles frame-rates, by drawing entire alternate frames entirely using AI, and without involving the raster pipeline.

The new GeForce RTX 4070 SUPER displaces the RTX 4070 from its $599 price point, which now drops to $549. There exists a rather big performance gap between the original RTX 4070 and the $800 RTX 4070 Ti, which rival AMD capitalized on with the Radeon RX 7800 XT, which could be one of the reasons NVIDIA configured the RTX 4070 SUPER with almost as many shaders as the RTX 4070 Ti. The RTX 4070 SUPER is based on the same AD104 silicon as the two, enabling 56 out of 60 streaming multiprocessors (SM) present on the silicon. This results in 7,168 CUDA cores, 224 Tensor cores, and 56 RT cores, which is a 21% increase over those of the original RTX 4070—not bad considering it’s coming at the same $599 MSRP. Besides these, NVIDIA has increased the ROP count from 64 on the RTX 4070, to the full 80 available on the silicon. The on-die cache size has been increased to 48 MB as on the RTX 4070 Ti. The memory sub-system is carried over, at 12 GB of 21 Gbps GDDR6X across a 192-bit memory bus.

The new Ada Lovelace graphics architecture debuts generational gains in performance and energy efficiency, thanks to the new 5 nm EUV foundry node. The new generation CUDA core, besides the usual IPC gains, now supports shader execution reordering, which benefits ray tracing workloads. The new 3rd generation RT core improves ray intersection performance, as well as introduces support for displaced micro-meshes, a feature that should increase the geometric complexity of ray traced objects. The new optical flow accelerator is a required hardware resource for DLSS 3 Frame Generation to work. NVIDIA also re-architected the memory sub-system to place larger on-die caches, which allows NVIDIA to narrow the memory bus width, while tapping into higher density GDDR6X memory chips, to increase memory size.

The Gainward RTX 4070 SUPER Ghost features a simple install-and-forget design at the $600 baseline price, with a large aluminium fin-stack heatsink that’s ventilated by a pair of fans; a 2-slot thick cooler, NVIDIA reference clock speeds of 1980 MHz GPU base frequency, and 2475 MHz boost; and an untouched 21 Gbps GDDR6X memory speed.