NVIDIA TU106 Graphics Processing Unit (GPU)

Overview
Manufacturer
NVIDIA
Series
GeForce 20
Release Date
August 20th, 2018
Board Model
NVIDIA PG160
Graphics Processing Unit
GPU Model
TU106-400/A (TU106)
Architecture
Turing
Fabrication Process
12 nm FinFET
Die Size
445 mm2
Transistors Count
10.8B
Transistors Density
24.3M TRAN/mm2
CUDAs
2304
TMUs
144
ROPs
64
Clocks
Base Clock
1410 MHz
Boost Clock
1620 MHz
Memory Clock
1750 MHz
Effective Memory Clock
14000 Mbps
Memory Configuration
Memory Size
8192 MB
Memory Type
GDDR6
Memory Bus Width
256-bit
Memory Bandwidth
448.0 GB/s
Physical
Interface
PCI-Express 3.0 x16
Power Connectors
1 × 8-pin
Thermal Design Power
175 W
Recommended PSU
500 W
Maximum Temperature
89 °C
SLI Support
Unsupported
Display Outputs
DVI-I DualLink
1 ×
DisplayPort
2 ×
HDMI 2.0
1 ×
API Support
DirectX
12.0
Mantle
-
Vulkan
1.0
OpenGL
4.5
OpenCL
2.1
Shader Model
5.0

Performance
Pixel Fillrate
90.2 GPixels/s
Texture Fillrate
203 GTexel/s
SPFP Performance
6.5 TFLOPS
Performance per W
37.1 GFLOPS/W
Performance per mm2
14.6 GFLOPS/mm2


 ModelGPUClocksMemory
NVIDIA GeForce RTX 2070 TU106-400/A 1410 / 1750 MHz 8 GB GDDR6
NVIDIA Quardro RTX 4000 TU106 1215 / 1625 MHz 8 GB GDDR6
NVIDIA GeForce RTX 2060 TU106-300 1365 / 1750 MHz 6 GB GDDR6
NVIDIA GeForce GTX 2070 Founders Edition (V)
2304 Cores
TU106 1410 / 1750 MHz 8 GB GDDR6