GPU

NVIDIA Tesla V100 SXM2 16 GB

Edit@true years ago

Intergrated Memory(VRAM)
Capacity

16 GB

(HBM2 4096-bit)

Bandwidth

1132 GB/s

161 Token/s

Vector Compute
FP64
8.18 T
FP32
16.35 T
FP16
32.71 T
BF16
X
INT32
16.35 T
INT8
X

NVIDIA Tesla V100 SXM2 16 GB General-Purpose Float-Point performance (Vector Performance / Scalar Performance)

FP64: 8.18 TFLOPS

FP32: 16.35 TFLOPS

FP16: 32.71 TFLOPS

INT32: 16.35 TOPS

Matirx Compute
FP64
X
FP32
X
FP16
130.83 T
FP8
X
TF32
X
BF16
X
INT16
X
INT8
X
INT4
X

NVIDIA Tesla V100 SXM2 16 GB AI performance (Tensor Performance / Matrix Performance)

FP16: 130.83 TFLOPS

Hardware Specs
NVIDIA Tesla V100 SXM2 16 GB is a 12nm chip, has 21100 million transistors, launched by NVIDIA at 2019. It has 16 GB built-in(On-Board/On-Chip) memory with bandwidth up to 1132 GB/s. It has 5120 general-purpose ALUs(CUDA cores/Shader cores) and 640 matrix cores(Tensor cores) .
Process Node
12 nm
Launch Year
2019
Vector(CUDA) Cores
5120
Matrix(Tensor) Cores
640
Core Frequency
1245 ~ 1597 MHz
Cache
6MB