GPU

NVIDIA GeForce RTX 3080

Edit@6 days ago

Intergrated Memory(VRAM)
Capacity

10 GB

(GDDR6X 320-bit)

Bandwidth

760 GB/s

108 Token/s

Vector Compute
FP64
465.10 G
FP32
29.77 T
FP16
29.77 T
BF16
29.77 T
INT32
14.88 T
INT8
X

NVIDIA GeForce RTX 3080 General-Purpose Float-Point performance (Vector Performance / Scalar Performance)

FP64: 465.10 GFLOPS

FP32: 29.77 TFLOPS

FP16: 29.77 TFLOPS

BF16: 29.77 TFLOPS

INT32: 14.88 TOPS

Matirx Compute
FP64
X
FP32
X
FP16
59.54 T
119.07 T
FP8
X
TF32
29.77 T
59.54 T
BF16
59.54 T
119.07 T
INT16
X
INT8
238.14 T
476.28 T
INT4
476.28 T
952.57 T

NVIDIA GeForce RTX 3080 AI performance (Tensor Performance / Matrix Performance)

FP16: 59.54 TFLOPS, with sparsity: 119.07 TFLOPS

TF32: 29.77 TFLOPS, with sparsity: 59.54 TFLOPS

BF16: 59.54 TFLOPS, with sparsity: 119.07 TFLOPS

INT8: 238.14 TOPS, with sparsity: 476.28 TOPS

INT4: 476.28 TOPS, with sparsity: 952.57 TOPS

Hardware Specs
NVIDIA GeForce RTX 3080 is a 8nm chip, has 28300 million transistors, launched by NVIDIA at 2020. It has 10 GB built-in(On-Board/On-Chip) memory with bandwidth up to 760 GB/s. It has 8704 general-purpose ALUs(CUDA cores/Shader cores) and 272 matrix cores(Tensor cores) .
Process Node
8 nm
Launch Year
2020
Vector(CUDA) Cores
8704
Matrix(Tensor) Cores
272
Core Frequency
1440 ~ 1710 MHz
Cache
5MB

Comment without registration

Share your experience with NVIDIA GeForce RTX 3080 / Found an Error? Help Us Improve!