The GPU Specifications
Turing TU116 GPU Specifications
The GPU TU116 for the GeForce GTX 1660 (Ti) contains 6.6 billion transistors localized on the die. In comparison, Pascal had close to 12 billion transistors on a die size of 471mm2. Gamers will immediately look at the shader processors; the Quadro RTX 8000 has 4608 of them enabled and since everything with bits is in multitudes of eight and while looking at the GPU die photos; it has 24 SMs (streaming multiprocessors) each holding 64 cores = 1536 Shader processors. This GPU is fabbed on an optimized 12nm TSMC FinFET+ node and the full GPU is unlocked.
GeForce | RTX 2080 Ti FE | RTX 2080 Ti | RTX 2080 FE | RTX 2080 | RTX 2070 FE | RTX 2060 FE | GTX 1660 Ti |
---|---|---|---|---|---|---|---|
GPU | TU102 | TU102 | TU104 | TU104 | TU106 | TU106 | TU116 |
Node | TSMC 12 nm FFN | ||||||
Die Size mm² | 754 | 545 | 445 | ||||
Shader cores | 4352 | 4352 | 2944 | 2944 | 2304 | 1920 | 1536 |
Transistor count | 18.6 Billion | 18.6 Billion | 13.6 Billion | 13.6 Billion | 10.8 Billion | 10.8 Billion | 6.6 Billion |
Base frequency | 1350 MHz | 1350 MHz | 1515 MHz | 1515 MHz | 1410 MHz | 1365 MHz | 1500 MHz |
Boost frequency | 1635 MHz | 1545 MHz | 1800 MHz | 1710 MHz | 1710 MHz | 1680 MHz | 1770 MHz |
Memory | 11GB GDDR6 | 11GB GDDR6 | 8GB GDDR6 | 8GB GDDR6 | 8GB GDDR6 | 6GB GDDR6 | 6GB GDDR6 |
Memory frequency | 14 Gbps | 14 Gbps | 14 Gbps | 14 Gbps | 14 Gbps | 14 Gbps | 12 Gbps |
Memory bus | 352-bit | 352-bit | 256-bit | 256-bit | 256-bit | 192-bit | 192-bit |
Memory bandwidth | 616 GB/s | 616 GB/s | 448 GB/s | 448 GB/s | 448 GB/s | 336 GB/s | 288 GB/s |
L2 Cache | 5632 KB | 5632 KB | 4096 KB | 4096 KB | 4096 KB | 4096 KB | 4096 KB |
RT cores (x64) | 68 | 68 | 46 | 46 | 36 | 30 | - |
Tensor cores | 544 | 544 | 368 | 368 | 288 | 240 | - |
Texture units | 272 | 272 | 184 | 184 | 144 | 120 | 96 |
ROPs | 96 | 96 | 64 | 64 | 64 | 48 | 48 |
TDP | 260W | 250W | 225W | 215W | 185W | 160W | 120W |
Power connector | 2x 8-pin | 2x 8-pin | 8+6-pin | 8+6-pin | 8-pin | 8-pin | 8-pin |
NVLink | Yes | Yes | Yes | Yes | - | - | - |
Max Therm degree C | 89 | 89 | 89 | ||||
price | $ 1199 | $ 999 | $ 799 | $ 699 | $ 599 | $ 349 | $ 279 |
Graphics memory - GDDR6
Allow me to quickly inject a paragraph here. Another difference in-between Volta and Turing is graphics memory. HBM2 is a bust for consumer products, at least it seems and feels that way. The graphics industry at this time is clearly favoring the new GDDR6. It’s easier and cheaper to fab and add, and at this time can even exceed HBM2 in performance. The previous GeForce GTX 1080 with the latest GDDR5X memory could run 11 Gbps, often tweakable towards the 12 Gbps range. GDDR6 graphics memory will be faster and more energy efficient. The memory is advancing on GDDR5X (Graphics Double Data Rate (DDR)) but with a memory bandwidth of 14 Gbit/s it offers almost twice as much as GDDR5 (not GDDR5X) offers. In the near future, GDDR6 could transfer data at 16Gbps (bits per second), which is twice as fast as regular GDDR5. The GeForce GTX 1660 Ti (6GB 192-bit) will be paired with 12 Gbps GDDR6.