home > GRID K100

 
GRID K100

Specifications of the NVIDIA GRID K100 GPU


NVIDIA GRID K100 Graphics Processing Unit for server market segment utilizes Kepler architecture, and was released in June 2013. The GPU is manufactured on 28 nm technological process. The card has graphics clocked at 850 MHz. It also boasts 192 CUDA cores, 16 texture units, together with 16 ROPs. The GRID K100 incorporates 256 MB of DDR3 memory. Because the memory runs at 891 MHz, and uses 128 bit bus, the effective memory bandwidth is 28.51 GB/s. The GPU supports PCI Express 3.0 interface. The maximum power consumption of this model is 130 Watt.

Below you will find detailed specs of the NVIDIA GRID K100 graphics card.

Name / Brand / Architecture

Manufacturer:NVIDIA
Model:GRID K100
Reference card?:Yes
Target market segment:Virtual Desktop Server
Die name:GK107
Architecture:Kepler
Fabrication process:28 nm
Transistors:1.27 billion
Bus interface:PCI-E 3.0 x 16
Launch date:June 2013
Price at launch:$63

Frequency

Graphics clock:850 MHz

Memory specifications

Memory size:** MB
Memory type:DDR3
Memory clock:891 MHz
Memory clock (effective):1782 MHz
Memory interface width:128-bit
Memory bandwidth:28.51 GB/s

Cores / Texture

CUDA:3.0
CUDA cores:192
ROPs:16
Texture units:16
RAMDACs:400 MHz

Electric characteristics

Maximum GPU temp:96 C
Maximum power draw:130 W

Video features

Maximum digital resolution:1920 x 1200
Anti-Aliasing technologies:FSAA

Performance

Pixel fill rate:13.6 Gigapixels/s
Texture fill rate:13.6 Gigatexels/s
Single presision compute power:326.4 GFLOPS

External connectors

Power connectors:1 x 6-pin

Other features / Support

Other features:Direct Compute 5.0
Open CL support:1.1
OpenGL support:4.4
DirectX support:11.0
Shader model:5

Notes


Identical vGPUs share the GPU resources, and have their own dedicated VRAM.
The GRID K1 can service 8 GRID K100 virtual GPUs per physical GPU, or 32 per card.
The GRID K100 is a virtual GPU served by the GRID K1 graphics card.
Email:info@Mobilegpu.com  All rights reserved.