The topic title says it all. Back in June when we talked to NVIDIA about it, we asked and they said they had no plans for a SUPER model GeForce RTX 2080 Ti. Then again, if it is possible, you know they will right? So something interesting has happened. A new entry in the changelog of AIDA64 beta is creating a bit of chatter.
Recent release notes from AIDA64 in beta version 6.00.5151 indicated this Nvidia Geforce RTX new graphics card based on TU102. The GPU is obviously is used in the upper high-end segment of cards is used in the RTX 2080 Ti, Titan RTX, Quadro RTX 8000 and Quadro RTX 6000. Check that cryptic listing entry, interesting right.
Update: November 20th 2019. Again new chatter is hitting the web, a tweet from info leaker kopite7kimi is hinting towards an RTX 2080 Ti SUPER. While it is the most unsubstantiated rumor this far, it indicates a full 4,608 CUDA cores, 576 tensor cores, and 72 RT cores, and thus 288 TMUs and 96 ROPs. The Super would get fast 16 Gbps GDDR6 memory.
Take it with a grain of salt, or two.
On the other side, Radeon Big Navi could be presented at CES 2020, a Geforce RTX 2080 Ti Super would be the perfect answer from Nvidia. Below a current product overview versus Super products in the RTX series, let's add a 'Titan RTX'.
model | Titan RTX | Geforce RTX 2080 Ti FE | Geforce RTX 2080 Super | Geforce RTX 2080 FE | Geforce RTX 2070 Super | Geforce RTX 2070 FE | Geforce RTX 2060 Super | Geforce RTX 2060 FE |
---|---|---|---|---|---|---|---|---|
Code Name | TU102-400 | TU102-300A-Kx | TU104-450 | TU104-400A | TU104-410 | TU106-400A | TU106-410 | TU106-400 (A) |
Chip size | 754 mm² | 754 mm² | 545 mm² | 545 mm² | 545 mm² | 445 mm² | 445 mm² | 445 mm² |
Transistors Graphics chip (million) | 18,600 | 18,600 | 13,600 | 13,600 | 13,600 | 10,800 | 10,800 | 10,800 |
manufacturing processes | 12 nm FFN | 12 nm FFN | 12 nm FFN | 12 nm FFN | 12 nm FFN | 12 nm FFN | 12 nm FFN | 12 nm FFN |
Graphics Processing Clusters (GPCs) | 6 | 6 | 6 | 6 | 6 or 5 | 3 | 3 | 3 |
ALUs / SM / TMU / ROPs | 4,608 / 72/288/96 | 4,352 / 68/272/88 | 3,072 / 48/192/64 | 2,944 / 46/184/64 | 2,560 / 40/160/64 | 2,304 / 36/144/64 | 2,176 / 34/136/64 | 1,920 / 30/120/48 |
Dedicated INT32 ALUs | 4608 | 4352 | 3072 | 2944 | 2560 | 2304 | 2176 | 1920 |
Raytracing cores | 72 | 68 | 48 | 46 | 40 | 36 | 34 | 30 |
Tensor cores | 576 | 544 | 384 | 368 | 320 | 288 | 272 | 240 |
GPU base clock (MHz) | 1350 | 1350 | 1650 | 1515 | 1605 | 1410 | 1470 | 1365 |
GPU boost clock in games (MHz) | 1770 | 1635 | 1815 | 1800 | 1770 | 1710 | 1650 | 1680 |
FP32 / FP64 performance (TFLOPS) | 16.31 / 0.51 | 14.23 / 0.45 | 11.15 / 0.35 | 10.60 / 0.33 | 9.06 / 0.28 | 7.88 / 0.25 | 7.18 / 0.22 | 6.45 / 0.20 |
Level 2 cache (KiB) | 6144 | 5632 | 4096 | 4096 | 4096 | 4096 | 4096 | 3072 |
Memory connection (bit) | 384 | 352 | 256 | 256 | 256 | 256 | 256 | 192 |
Memory data-rate | 14.0 | 14.0 | 15.5 | 14.0 | 14.0 | 14.0 | 14.0 | 14.0 |
Memory type | GDDR6 | GDDR6 | GDDR6 | GDDR6 | GDDR6 | GDDR6 | GDDR6 | GDDR6 |
Memory transfer (GB / s) | 672.0 | 616.0 | 496.1 | 448.0 | 448.0 | 448.0 | 448.0 | 336.0 |
Storage capacity (MiB) | 24576 | 11264 | 8192 | 8192 | 8192 | 8192 | 8192 | 6144 |
PCI Express power connectors | 2 × 8-pin | 2 × 8-pin | Each 1 × 8- / 6-pin | Each 1 × 8- / 6-pin | Each 1 × 8- / 6-pin | 1 × 8-pin | 1 × 8-pin | 1 × 8-pin |
Typical power consumption | 280 watts | 260 watts | 250 watts | 225 watts | 215 watts | 185 watts | 175 watts | 160 watts |
Retail price | 2,699 eur | 1,259 eur | 739 eur | 849 eur | 529 eur | 629 eur | 419 eur | 369 eur |
Updated: Geforce RTX 2080 Ti Super Rumors Surface again