NVIDIA TU104 Graphics Processing Unit (GPU)

Overview
Manufacturer
NVIDIA
Series
GeForce 20
Release Date
July 23rd, 2019
Graphics Processing Unit
GPU Model
TU104-450 (TU104)
Architecture
Turing
Fabrication Process
12 nm FinFET
Die Size
545 mm2
Transistors Count
13.6B
Transistors Density
25M TRAN/mm2
CUDAs
3072
TMUs
192
ROPs
64
Clocks
Base Clock
1650 MHz
Boost Clock
1815 MHz
Memory Clock
1936 MHz
Effective Memory Clock
15488 Mbps
Memory Configuration
Memory Size
8192 MB
Memory Type
GDDR6
Memory Bus Width
256-bit
Memory Bandwidth
495.6 GB/s
Physical
Interface
PCI-Express 3.0 x16
Power Connectors
1 × 6-pin, 1 × 8-pin
Thermal Design Power
250 W
Recommended PSU
600 W
Maximum Temperature
89 °C
SLI Support
2-way
Display Outputs
DisplayPort
3 ×
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
105.6 GPixels/s
Texture Fillrate
316.8 GTexel/s
SPFP Performance
11.2 TFLOPS
Performance per W
44.6 GFLOPS/W
Performance per mm2
20.5 GFLOPS/mm2


 ModelGPUClocksMemory
NVIDIA GeForce RTX 2080 SUPER TU104-450 1650 / 1936 MHz 8 GB GDDR6
NVIDIA Quardro RTX 5000 TU104 1620 / 1750 MHz 16 GB GDDR6
NVIDIA GeForce RTX 2080 TU104-400/A 1515 / 1750 MHz 8 GB GDDR6
NVIDIA GeForce RTX 2080 (Mobile) N18E-G3 1380 / 1750 MHz 8 GB GDDR6
NVIDIA GeForce RTX 2080 Max-Q (Mobile) TU104 735 / 1750 MHz 8 GB GDDR6
NVIDIA GeForce RTX 2070 SUPER TU104-410 1605 / 1750 MHz 8 GB GDDR6
NVIDIA GeForce RTX 2060 SUPER TU104-410 1470 / 1750 MHz 8 GB GDDR6
NVIDIA GeForce GTX 2080 Founders Edition (V)
2944 Cores
TU104 1515 / 1750 MHz 8 GB GDDR6