Tesla K40 GPU is equipped with 12 GB of memory and ideal for the most demanding HPC and big data problem sets. Tesla K40 has Nvidia GPU boost feature which increases the clock speed of all CUDA cores, providing a 20% to 30% performance boost for many common applications

Features and Benefits:

• NVIDIA GPU Boost: On-demand performance boost to attain up to 25% additional application speedup
• Streaming Multiprocessor (SMX): Perform 3X the workload with the same power budget.
• NVIDIA Kepler Architecture: Accelerate all your applications with the world’s fastest, most efficient HPC architecture.
• ECC Memory: Address a critical requirement for computing accuracy and reliability in supercomputing and data centers.
• System Monitoring Features: Manage GPU processors in computing systems using widely used cluster/grid-management solutions.

Specification	Description
GPU Chips	GK110B
Cuda Cores	2880
Single Precision Performance	4.2 TFLPOS
Double Precision Performance	1.439 TFLPOS
Memory Size	12GB GDDR5
Memory Bandwidth(ECC off)	288 Gbytes/sec
Memory I/O	384-bit
Memory Clock	3004 MHz
SMX Units	15
GPU Boost Feature	Yes
Architecture features	SMX, Dynamic Parallelism, Hyper-Q
System Interface	PCI Express 3.0 x16
Energy Star Enabling	Yes
Compute Capabilty	3.5
Semiconductor Manufacturing Size	28nm
Wattage (TDP)	235W
GPU Applications	> Reservoir simulation  > CAE (structural analysis)  > Molecular dynamics, Numerical analytics  > Computational visualization (ray tracing)
Thermal Solution	Ultra-quiet Active Fansink
Form Factor	110 mm (H) × 265 mm (L) - Dual Slot, Full-Height