Introduction

Zotac GeForce RTX 4070 SUPER Trinity Black is a mid-range custom design for the GeForce RTX 4070 Super. NVIDIA today gave its RTX 40-series a mid lifecycle refresh with the new RTX 40 SUPER series. This brings you more performance, and hence more value, at existing price points. The new RTX 4070 SUPER we’re reviewing today displaces the RTX 4070 from its official RTX 4070 MSRP, which now starts at $550. Zotac is going beyond what the MSRP cards offer, with their Trinity Black triple-fan cooling solution. The GPU itself is not overclocked, which should prove beneficial for temperatures and fan noise. NVIDIA is recommending the new RTX 4070 SUPER for the same gaming scenario as the other RTX 4070 series cards—maxed out AAA gaming at 1440p with ray tracing enabled; although our testing over the months have consistently shown even the older RTX 4070 to be capable of 4K Ultra HD gameplay. You could turn the eye-candy just a touch down, or take advantage of the DLSS feature, including the newer DLSS 3 Frame Generation, which is now supported across an impressive number of AAA and e-sports titles.

NVIDIA designed the RTX 4070 SUPER to be closer in performance to the RTX 4070 Ti than it is further from the original RTX 4070. This is done by increasing the shader counts by a neat 21%, enabling the full 48 MB L3 cache, as well as all of the 80 ROPs on the 5 nm AD104 silicon that the RTX 4070 series was based on; which means the RTX 4070 SUPER enables 93% of the available shaders, in contrast to the RTX 4070, which only has 76% of them. The RTX 4070 Ti maxes this chip out, but is being retired by NVIDIA, and replaced by the new RTX 4070 Ti SUPER, which will come out next week.

The Ada Lovelace graphics architecture takes advantage of the 5 nm EUV foundry node, and new on-silicon technologies, to generationally increase performance over the RTX 30-series Ampere. The new Ada CUDA core, besides IPC uplifts, supports the new shader execution reordering feature, which should benefit ray tracing workloads. The new 3rd generation RT core, besides increased ray intersection performance, supports displaced micro-meshes, which should increase the complexity of ray traced objects. The new optical flow accelerator component is what allows the GPU to draw alternate frames entirely using AI, which is what powers DLSS 3 Frame Generation.

NVIDIA carved the RTX 4070 SUPER out of the AD104 by enabling 56 out of 60 SM available on the silicon, working out to 7,168 CUDA cores, 224 Tensor cores, 56 RT cores, and 224 TMUs. The card gets the full 48 MB of L2 cache available on the silicon, as well as all 80 ROPs, which should significantly improve the memory sub-system, as well as raster performance. The memory itself, however, is unchanged from the RTX 4070 Ti and RTX 4070—you get 12 GB of 21 Gbps GDDR6X memory across a 192-bit memory bus, yielding 504 GB/s of bandwidth. This may seem low, but this is where the large on-die L2 cache steps in, to compensate for the bandwidth shortfall. 12 GB is still a generational gain in memory size over even the previous generation RTX 3080.

Zotac RTX 4070 SUPER Trinity Black comes with the new IceStorm 2.0 cooling solution, which is a large triple-slot cooler that uses a heavy dual fin-stack heatsink that’s ventilated by three fans. The GPU, however, sticks to NVIDIA reference clock speeds of 2475 MHz boost, and 21 Gbps GDDR6X-effective memory, the company has priced the card at $650. Compared to the RTX 4070, NVIDIA has increased the TGP power limit by 10% to drive all those extra shaders, which means that the single 8-pin PCIe you found on some of the MSRP RTX 4070 cards now makes way for a full 12VHPWR connector. An adapter cable is included.