GPU

NVIDIA Tesla P40

Edit@2 months ago

Intergrated Memory(VRAM)
Capacity

24 GB

(GDDR5 384-bit)

Bandwidth

347 GB/s

49 Token/s

Vector Compute
FP64
367.40 G
FP32
11.76 T
FP16
X
BF16
X
INT32
NI
INT8
X

NVIDIA Tesla P40 General-Purpose Float-Point performance (Vector Performance / Scalar Performance)

FP64: 367.40 GFLOPS

FP32: 11.76 TFLOPS

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

NVIDIA Tesla P40 AI performance (Tensor Performance / Matrix Performance)

Hardware Specs
NVIDIA Tesla P40 is a 16nm chip, has 11800 million transistors, launched by NVIDIA at 2016. It has 24 GB built-in(On-Board/On-Chip) memory with bandwidth up to 347 GB/s. It has 3840 general-purpose ALUs(CUDA cores/Shader cores).
Process Node
16 nm
Launch Year
2016
Vector(CUDA) Cores
3840
Matrix(Tensor) Cores
Core Frequency
1303 ~ 1531 MHz
Cache
3MB