GPU

NVIDIA GeForce RTX 3050

Edit@2 months ago

Intergrated Memory(VRAM)
Capacity

8 GB

(GDDR6 128-bit)

Bandwidth

224 GB/s

32 Token/s

Vector Compute
FP64
142.20 G
FP32
9.10 T
FP16
9.10 T
BF16
9.10 T
INT32
4.55 T
INT8
X

NVIDIA GeForce RTX 3050 General-Purpose Float-Point performance (Vector Performance / Scalar Performance)

FP64: 142.20 GFLOPS

FP32: 9.10 TFLOPS

FP16: 9.10 TFLOPS

BF16: 9.10 TFLOPS

INT32: 4.55 TOPS

Matirx Compute
FP64
X
FP32
X
FP16
18.20 T
36.39 T
FP8
X
TF32
9.10 T
18.20 T
BF16
18.20 T
36.39 T
INT16
X
INT8
72.79 T
145.57 T
INT4
145.57 T
291.14 T

NVIDIA GeForce RTX 3050 AI performance (Tensor Performance / Matrix Performance)

FP16: 18.20 TFLOPS, with sparsity: 36.39 TFLOPS

TF32: 9.10 TFLOPS, with sparsity: 18.20 TFLOPS

BF16: 18.20 TFLOPS, with sparsity: 36.39 TFLOPS

INT8: 72.79 TOPS, with sparsity: 145.57 TOPS

INT4: 145.57 TOPS, with sparsity: 291.14 TOPS

Hardware Specs
NVIDIA GeForce RTX 3050 is a 8nm chip, has 12000 million transistors, launched by NVIDIA at 2022. It has 8 GB built-in(On-Board/On-Chip) memory with bandwidth up to 224 GB/s. It has 2560 general-purpose ALUs(CUDA cores/Shader cores) and 80 matrix cores(Tensor cores) .
Process Node
8 nm
Launch Year
2022
Vector(CUDA) Cores
2560
Matrix(Tensor) Cores
80
Core Frequency
1552 ~ 1777 MHz
Cache
2MB