Following NVIDIA’s announcement of its latest GeForce RTX 20-series GPUs on Monday, its board partners also introduced their custom-built offerings based on the new processors. As is usually the case, the first wave of products do not feature extreme frequencies or highly-custom designs – and indeed many of them look like they’ll be using NVIDIA’s reference PCB – but they still offer various customizations and cooling changes relative to NVIDA’s own “reference” Founders Ediiton card. Today we are taking a look on these cards and briefly compare their characteristics.
NVIDIA’s GeForce RTX 20-series lineup will initially consist of three models: the RTX 2080 Ti based on the TU102 GPU as well as the RTX 2070 and the RTX 2080 powered by the TU104 processor. The flagship GeForce RTX 2080 Ti GPU features 4352 CUDA cores, a 352-bit memory bus, and carries 11 GB of GDDR6 memory. Meanwhile, the GeForce RTX 2070 and 2080 boards have 2304 and 2944 cores (respectively), a 256-bit memory bus, and carry 8 GB of GDDR6.
Makers of graphics cards cannot activate any additional CUDA cores, but what they can do is increase default and boost frequencies of GPUs, or install high-performance cooling systems, thus altering and improving performance of their products compared to reference video cards designed by NVIDIA. Without further ado, let’s jump right to custom-designed graphics cards based on the TU102 and TU104 GPUs.
Custom GeForce RTX 2080 Ti Graphics Cards
NVIDIA’s reference PCB for the GeForce RTX 2080 Ti, which is being used in their own factory overclocked Founders Edition card, feature a 13-phase VRM and a double-wide dual-fan cooler. The chip developer’s Founder’s Edition will ship with the GPU clocked at 1350/1635 MHz (base/boost), but for its partners NVIDIA recommends maximum burst frequency of up to 1545 MHz. In a bid to differentiate themselves from NVIDIA’s own design (while ideally fetching a similar price), they’ll be implementing their own factory overclocks along with different (and usually larger) coolers.