NVIDIA made its third and final entry to the GeForce RTX 20-series Super family of graphics cards with the GeForce RTX 2080 Super. The company debuted the series with the new RTX 2060 Super and RTX 2070 Super launched earlier this month to preempt AMD's Radeon RX 5700-series "Navi," which beat the original RTX 2060 and RTX 2070 at their price points. NVIDIA also augmented the series with the GeForce RTX 2080 Super despite no apparent competition from AMD. The Radeon VII is a limited-edition product that's reportedly being retired with the advent of Navi. So why did NVIDIA feel the need to launch the RTX 2080 Super?
The answer to the question lies in the inadvertent mess NVIDIA created in its product stack with the RTX 2070 Super. The company had maxed out the "TU106" to build the original RTX 2070, and had to tap into the larger "TU104" silicon to carve out the RTX 2070 Super by disabling one of its six GPCs (graphics processing clusters) and end up with 2,560 out of 3,072 CUDA cores physically present. Our reviews of the RTX 2070 Super show it to have a lot of performance over the RTX 2070 non-Super, ending up just 6-8 percent behind the original RTX 2080 in performance. This presents a problem for NVIDIA as it can no longer sell the RTX 2080 at $700 when a $500 card is that close and can be overclocked to close the gap. It now had to find a way to continue selling a $700 graphics card without having to tap into the much larger "TU102" silicon. We hence have the RTX 2080 Super, which continues to be based on the "TU104".
NVIDIA created the original RTX 2080 out of the "TU104" by enabling just 2,944 out of 3,072 CUDA cores. The new RTX 2080 Super is configured with all 3,072 CUDA cores and proportionate increases in other components, such as 192 TMUs, 384 Tensor cores, and 48 RT cores. The second enhancement NVIDIA gave the RTX 2080 Super is higher GPU clock speeds. The GPU Boost frequency of the card is set at 1815 MHz, a 6 percent gain from the 1710 MHz of the original RTX 2080. Lastly, NVIDIA made the memory faster by running it at 15.5 Gbps and using 16 Gbps-rated GDDR6 memory chips compared to the 14 Gbps memory clock of the original RTX 2080. This yields a memory bandwidth of 496 GB/s, which is roughly 11 percent gained. The card still draws power from a combination of 8-pin and 6-pin PCIe power connectors, although the TDP is increased to 250 W from the 225 W of the original.
With these changes, NVIDIA is looking to increase the performance gap within its product stack, between the RTX 2070 Super and RTX 2080 Super, back to levels that exist between the original RTX 2070 and RTX 2080, just so there isn't a vast gorge between the $500 RTX 2070 Super and the $1000 RTX 2080 Ti.
In this review, we have with us the GeForce RTX 2080 Super Founders Edition, the de facto reference-design card based on this GPU. It features NVIDIA's lavish dual-fan cooling solution with an added chrome embellishment and clock speeds of 1815 MHz GPU Boost with 15.5 Gbps memory. NVIDIA is pricing this card at US$700 and enabled its add-in card partners to price their cards around this mark, too.
|RX Vega 64||$400||4096||64||1247 MHz||1546 MHz||953 MHz||Vega 10||12500M||8 GB, HBM2, 2048-bit|
|GTX 1080 Ti||$700||3584||88||1481 MHz||1582 MHz||1376 MHz||GP102||12000M||11 GB, GDDR5X, 352-bit|
|RX 5700 XT||$400||2560||64||1605 MHz||1755 MHz||1750 MHz||Navi 10||10300M||8 GB, GDDR6, 256-bit|
|RTX 2070||$440||2304||64||1410 MHz||1620 MHz||1750 MHz||TU106||10800M||8 GB, GDDR6, 256-bit|
|RTX 2070 Super||$500||2560||64||1605 MHz||1770 MHz||1750 MHz||TU104||13600M||8 GB, GDDR6, 256-bit|
|Radeon VII||$680||3840||64||1802 MHz||N/A||1000 MHz||Vega 20||13230M||16 GB, HBM2, 4096-bit|
|RTX 2080||$630||2944||64||1515 MHz||1710 MHz||1750 MHz||TU104||13600M||8 GB, GDDR6, 256-bit|
|RTX 2080 Super||$700||3072||64||1650 MHz||1815 MHz||1940 MHz||TU104||13600M||8 GB, GDDR6, 256-bit|
|RTX 2080 Ti||$1100||4352||64||1350 MHz||1545 MHz||1750 MHz||TU102||18600M||11 GB, GDDR6, 352-bit|