Report an Error

NVIDIA TU104B

Default
TU104B
Block Diagram
Block Diagram
SM Diagram
SM Diagram
NVIDIA's TU104B GPU uses the Turing architecture and is made using a 12 nm production process at TSMC. With a die size of 545 mm² and a transistor count of 13,600 million it is a very big chip. TU104B supports DirectX 12 Ultimate (Feature Level 12_2). For GPU compute applications, OpenCL version 3.0 and CUDA 7.5 can be used. Additionally, the DirectX 12 Ultimate capability guarantees support for hardware-raytracing, variable-rate shading and more, in upcoming video games. It features 3072 shading units, 192 texture mapping units and 64 ROPs. Also included are 384 tensor cores which help improve the speed of machine learning applications. The GPU also contains 48 raytracing acceleration cores.
Further reading: Turing Architecture Whitepaper

Graphics Processor

GPU Name
TU104B
Codename
NV164
Architecture
Turing
Foundry
TSMC
Process Size
12 nm
Transistors
13,600 million
Density
25.0M / mm²
Die Size
545 mm²
Released
Jun 8th, 2020

Graphics Features

DirectX
12 Ultimate (12_2)
OpenGL
4.6
OpenCL
3.0
Vulkan
1.2
CUDA
7.5
Shader Model
6.6
PureVideo HD
VP10
VDPAU
Feature Set J

Render Config

Shading Units
3072
TMUs
192
ROPs
64
SM Count
48
FP16 Units
6144
FP64 Units
96
INT32 Units
3072
Tensor Cores
384
RT Cores
48
SFUs
768
TPCs
24
GPCs
6
Tex L1 Cache
32 KB per SM
L1 Cache
64 KB per SM
L2 Cache
4096 KB
Max. TDP
110 W

All Turing GPUs

NVIDIA GPU Architecture History

Graphics cards using the NVIDIA TU104B GPU

Name Chip Memory Shaders TMUs ROPs Base Clock Boost Clock Memory Clock
NVIDIA Quadro RTX 5000 Mobile Refresh 16 GB 3072 192 64 1035 MHz 1545 MHz 1750 MHz