Report an Error

NVIDIA Tesla K80

Graphics Processor
GK210 x2
Cores
2496 x2
TMUs
208 x2
ROPs
48 x2
Memory Size
12 GB x2
Memory Type
GDDR5
Bus Width
384 bit x2
GPU Chip
GPU
Top
Top
The Tesla K80 was a professional graphics card by NVIDIA, launched in November 2014. Built on the 28 nm process, and based on the GK210 graphics processor, in its GK210-885-A1 variant, the card supports DirectX 12.0. The GK210 graphics processor is a large chip with a die area of 561 mm² and 7,100 million transistors. Tesla K80 combines two graphics processors to increase performance. It features 2496 shading units, 208 texture mapping units and 48 ROPs per GPU. NVIDIA has placed 24,576 MB GDDR5 memory on the card, which are connected using a 384-bit memory interface per GPU (each GPU manages 12,288 MB). The GPU is operating at a frequency of 562 MHz, which can be boosted up to 824 MHz, memory is running at 1253 MHz.
Being a dual-slot card, the NVIDIA Tesla K80 draws power from 1x 8-pin power connectors, with power draw rated at 300 W maximum. This device has no display connectivity, as it is not designed to have monitors connected to it. Tesla K80 is connected to the rest of the system using a PCI-Express 3.0 x16 interface. The card measures 267 mm in length, and features a dual-slot cooling solution.

Graphics Processor

GPU Name
GK210
GPU Variant
GK210-885-A1
Architecture
Kepler 2.0
Foundry
TSMC
Process Size
28 nm
Transistors
7,100 million
Die Size
561 mm²

Graphics Card

Release Date
Nov 17th, 2014
Generation
Tesla
(Kxx)
Production
End-of-life
Bus Interface
PCIe 3.0 x16

Clock Speeds

Base Clock
562 MHz
Boost Clock
824 MHz
Memory Clock
1253 MHz
5012 MHz effective

Memory

Memory Size
12 GB
Memory Type
GDDR5
Memory Bus
384 bit
Bandwidth
240.6 GB/s

Render Config

Shading Units
2496
TMUs
208
ROPs
48
SMX Count
13
L1 Cache
16 KB (per SMX)
L2 Cache
1536 KB

Theoretical Performance

Pixel Rate
42.85 GPixel/s
Texture Rate
171.4 GTexel/s
FP32 (float) performance
4.113 TFLOPS
FP64 (double) performance
1,371 GFLOPS (1:3)

Board Design

Slot Width
Dual-slot
Length
10.5 inches
267 mm
TDP
300 W
Suggested PSU
700 W
Outputs
No outputs
Power Connectors
1x 8-pin
Board Number
P2080 SKU 200

Graphics Features

DirectX
12.0 (11_1)
OpenGL
4.6
OpenCL
1.2
Vulkan
1.1.103
CUDA
3.7
Shader Model
5.1

GK210 GPU Notes

L1 Cache is configurable from 16 KB up to 48 KB per SMX

GK210 chips have 5 Rasterizers each capable of 8 pixels per clock and are limited to 40 pixels per clock and because of that it can't feed all 48 ROPs when they all require data at the same time.
Luckily, additional ROPs can be used when for example MSAA is enabled (because it doesn't require additional data from rasterizers, while giving more work to ROPs).