Report an Error

NVIDIA GB202

NVIDIA's GB202 GPU uses the Blackwell 2.0 architecture at TSMC. GB202 supports DirectX 12 Ultimate (Feature Level 12_2). For GPU compute applications, OpenCL version 3.0 and CUDA 9.1 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 24576 shading units, 768 texture mapping units and 192 ROPs. Also included are 768 tensor cores which help improve the speed of machine learning applications. The GPU also contains 192 raytracing acceleration cores.
Further reading: Blackwell Architecture Whitepaper

Graphics Processor

Released
2025
GPU Name
GB202
Architecture
Blackwell 2.0
Foundry
TSMC
Process Type
4N FinFET
Process Size
0 nm
Transistors
unknown
Die Size
unknown

Graphics Features

DirectX
12 Ultimate (12_2)
OpenGL
4.6
OpenCL
3.0
Vulkan
1.3
CUDA
9.1
Shader Model
6.7
WDDM
3.1
Tensor Cores
5th Gen
RT Cores
4th Gen
NVENC
9th Gen
NVDEC
6th Gen
PureVideo HD
VP13
VDPAU
Feature Set M

Render Config

Shading Units
24576
TMUs
768
ROPs
192
SM Count
192
FP16 Units
24576
FP64 Units
384
INT32 Units
12288
Tensor Cores
768
RT Cores
192
SFUs
3072
TPCs
96
GPCs
12
Tex L1 Cache
64 KB per SM
L1 Cache
128 KB per SM
L2 Cache
90112 KB
Max. TDP
500 W

All Blackwell 2.0 GPUs

NVIDIA GPU Architecture History

Graphics cards using the NVIDIA GB202 GPU

Name Chip Memory Shaders TMUs ROPs Base Clock Boost Clock Memory Clock
NVIDIA GeForce RTX 5090
GB202 28 GB 20480 640 192 2235 MHz 2520 MHz 2500 MHz

GB202 GPU Notes

Ray Tracing Cores: 4th Gen
Tensor Cores: 5th Gen
NVENC: 9th Gen
NVDEC: 6th Gen
PureVideo HD: VP13
VDPAU: Feature Set M

Latest Drivers:
Windows 10 / 11:
GeForce Release: Latest
Quadro Release: Latest
Data Center Release: Latest
Jun 29th, 2024 15:30 EDT change timezone

Latest GPU Drivers

New Forum Posts

Popular Reviews

Controversial News Posts