Introduction

Gainward GeForce RTX 4070 Ti SUPER Phoenix GS is a new performance segment graphics card based on NVIDIA’s latest high-end GPU. The new RTX 4070 Ti SUPER comes from a three-part mid-lifecycle refresh of the RTX 40-series Ada generation, focusing mainly on the higher end of the product stack. The RTX 4070 Ti SUPER is recommended by NVIDIA for 1440p high refresh-rate gaming with maxed out settings and ray tracing. The RTX 4070 series of graphics cards are very much capable of gameplay at 4K Ultra HD, if you’re a little careful with your game settings, or use features such as DLSS that come with pretty much every new AAA title these days; or even the newer DLSS 3 Frame Generation, which nearly doubles frame-rates by drawing alternate frames entirely using AI.

The new GeForce RTX 4070 Ti SUPER replaces the RTX 4070 Ti from the product stack, at its $800 price-point, which is now no longer in production, and will be phased out of the market at lower prices. The RTX 4070 Ti SUPER comes with significantly upgraded specifications, most notably, its increased 16 GB of memory across a wider 256-bit memory bus; as opposed to 12 GB over 192-bit for the original RTX 4070 Ti. This is a straight 33% increase in memory bandwidth and size. The extra 4 GB of memory is what makes us excited about this card’s prospects with 4K gaming. Besides the extra memory, the RTX 4070 Ti SUPER gets 10% more shaders, and 20% more ROPs. All this comes from NVIDIA’s switch to the larger AD103 silicon from the AD104 that had been maxed out to create the RTX 4070 Ti.

NVIDIA carved the RTX 4070 Ti SUPER out of the AD103 by enabling 66 out of 80 SM, worth 8,448 CUDA cores; 264 Tensor cores, 66 RT cores, and 264 TMUs. NVIDIA enabled 96 out of the 112 ROPs present on the silicon, which is still a big uplift from the 80 on the RTX 4070 Ti; although the on-die L2 cache size remains the same at 48 MB out of the 64 MB available on the silicon. The GPU runs at 2610 MHz boost, and the memory at 21 Gbps, which is lower than the 22.5 Gbps of the RTX 4080, but still results in a 33% higher memory bandwidth when compared to the RTX 4070 Ti, thanks to that 256-bit memory bus.

The SUPER moniker only represents performance uplifts at existing price-points, no new features are introduced. The RTX 4070 Ti SUPER is still based on the existing Ada Lovelace graphics architecture. Ada introduces a new generation CUDA core with increased IPC and support for new math formats, as well as shader execution reordering, a feature that speeds up ray tracing. The new 3rd generation RT core, besides increased ray intersection performance, debuts displaced micro-meshes, which enables greater complexity in ray traced objects. The new optical flow accelerator component enables the GPU to draw alternate frames entirely using AI, cutting out the main raster machinery, which is how DLSS 3 Frame Generation works.

Gainward’s GS line of graphics cards go all the way back to the GeForce 4 Ti days, where it represented “Golden Sample,” the company’s highest state of factory OC. On the RTX 4070 Ti SUPER, some 15 generations ahead, the Gainward GS brand means overclocked speeds of 2670 MHz, compared to 2610 MHz reference. The Gainward RTX 4070 Ti SUPER Phoenix features a simple triple-slot, cooling solution with a triple fan setup. Gainward hasn’t given out a price, but we are estimating a price of $870, which is what we are using in our cost-performance calculations.