GPU

NVIDIA GeForce RTX 3080 Ti

Edit@4 months ago

Intergrated Memory(VRAM)
Capacity

12 GB

(GDDR6X 384-bit)

Bandwidth

912 GB/s

130 Token/s

Vector Compute
FP64
532.80 G
FP32
34.10 T
FP16
34.10 T
BF16
34.10 T
INT32
17.05 T
INT8
X

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

FP64: 532.80 GFLOPS

FP32: 34.10 TFLOPS

FP16: 34.10 TFLOPS

BF16: 34.10 TFLOPS

INT32: 17.05 TOPS

Matirx Compute
FP64
X
FP32
X
FP16
68.20 T
136.40 T
FP8
X
TF32
34.10 T
68.20 T
BF16
68.20 T
136.40 T
INT16
X
INT8
272.79 T
545.59 T
INT4
545.59 T
1091.17 T

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

FP16: 68.20 TFLOPS, with sparsity: 136.40 TFLOPS

TF32: 34.10 TFLOPS, with sparsity: 68.20 TFLOPS

BF16: 68.20 TFLOPS, with sparsity: 136.40 TFLOPS

INT8: 272.79 TOPS, with sparsity: 545.59 TOPS

INT4: 545.59 TOPS, with sparsity: 1091.17 TOPS

Hardware Specs
NVIDIA GeForce RTX 3080 Ti is a 8nm chip, has 28300 million transistors, launched by NVIDIA at 2021. It has 12 GB built-in(On-Board/On-Chip) memory with bandwidth up to 912 GB/s. It has 10240 general-purpose ALUs(CUDA cores/Shader cores) and 320 matrix cores(Tensor cores) .
Process Node
8 nm
Launch Year
2021
Vector(CUDA) Cores
10240
Matrix(Tensor) Cores
320
Core Frequency
1365 ~ 1665 MHz
Cache
6MB