Super or 2080S is the second-fastest desktop GPU in NVIDIA’s current Turing line-up after the GeForce RTX 2080 Ti (mid 2019). The RTX 2080 (Desktop) is a one-year-old update with significantly improved performance. On the TU104 chip, the GPU has 3,072 cores, up from 2,944. At 15.5 GHz, the VRAM remains 8 GB of GDDR6 (496 GB/s, instead of 14 GHz and 448 GB/s).
NVIDIA’s TU104 GPU has two new technologies: DLSS and RTRT, which together might provide more realistic lighting effects than previous GPUs based on the Pascal architecture (if the games supports it). Additionally, the RTX 2080 supports HDR, SMP, and H.265 video decoding and encoding on DisplayPort 1.4 and HDMI 2.0b (PlayReady 3.0).
In comparison to the Nvidia Titan XP and RTX 2080, the RTX 2080 Super is marginally better. To put it another way, the RTX2080S is best suited for 4k gaming with high to maximum detail settings (depending on the game). However, the playable resolution will be reduced if raytracing technologies are used.
However, maybe the most fascinating change with the RTX 2080 Super is not even the GPU itself, but rather what is being done on the memory front by NVIDIA to feed the beast. GDDR6-equipped video cards have delivered with 14Gbps RAM since their debut a year ago. At the time of its release, 14Gbps was the fastest speed grade available from the main memory providers, but it wasn’t on the roadmap. NVIDIA’s new card is the first to include 16Gbps of RAM, which has been in the works for some time.
NVIDIA has chosen Samsung’s latest 16Gbps, 8Gbit processors for the introduction of their new cards. Because of the card’s 256-bit memory bus, the RTX 2080 Super has the same 8GB of VRAM as the standard RTX 2080, but can take use of the increased bandwidth. It’s an uncommon move for NVIDIA, but they’re only operating the RTX 2080 Super at 15.5 Gbps instead of the 16 Gbps that the RAM is capable of in this case.
When it comes to the Super cards, NVIDIA says that this is because they wanted to make it as easy as possible for their board partners (including themselves) to utilise the same boards and coolers from their vanilla RTX 2080 cards to improve their production. This means that the RTX 2080 Super uses a PCB originally designed and certified for 14Gbps GDDR6. To get 16Gbps memory clocks, a new PCB seems to be required, which is not something board manufactures are likely to want to spend the money on for a 3% increase in memory clockspeed (and less on performance).
RTX 2080 Super tends to be a lot more GPU-bound than other NVIDIA cards because of this and the increased TDP. There isn’t much memory bandwidth left over for the fully equipped GPU, but the RTX 2080 Super makes up for it thanks to its faster memory rates. As a result, overclocking is likely to become increasingly one-dimensional, with GPU clockspeeds taking the lead by a large margin.
Recommended Readings (Visualr)