Tesla K80

Kepler 2.0 Tesla K Series Data Center Server GDDR5 2014 NVIDIA

Key Metrics

Launch Date: 2014-11-17
CUDA Cores: 2,496
Memory: 12 GB GDDR5
Memory Bandwidth: 240.6 GB/s
TDP/TBP: 300 W
Interface: PCIe 3.0 x16

Save this GPU, then open another GPU page to compare them side by side.

GPU Specs

Identity
Entry Type	Data Center GPU
Official Name	NVIDIA Tesla K80
GPU Codename	GK210
Architecture	Kepler 2.0
Launch Date	2014-11-17
Market Segment Notes	Data Center in Tesla K Series
Core Specs
GPU Die	GK210
Chip Variant / Package	BGA-2152
Process Node	TSMC 28 nm
Transistor Count	7.1 billion
Die Size	561 mm²
CUDA Cores	2496
SM Count	0
TMUs	208
ROPs	48
Base Clock	562 MHz
Boost Clock	824 MHz
Pixel Fillrate	42.85 GP/s
Texture Fillrate	171.4 GT/s
FP32 Throughput	4.11 TFLOPS
FP64 Throughput	1.37 TFLOPS
Memory
Memory Size	12 GB
Memory Type	GDDR5
Memory Bus Width	384-bit
Memory Clock	1,253 MHz
Memory Bandwidth	240.6 GB/s
L2 Cache	2 MB
L1 Cache	16 KB
Board and Display
TDP / TBP	300 W
Recommended PSU	700 W
Power Connectors	8-pin EPS
Interface	PCIe 3.0 x16
Slot Width	Dual-slot, 267 mm length
Card Length	267 mm
Board Partner Model	Reference or OEM specification
Reference Design	NVIDIA reference design
Software
CUDA Compute Capability	3.7
Shader Model	5.1
DirectX Support	DirectX 11.1
Vulkan Support	Vulkan 1.2
OpenGL Support	OpenGL 4.6
OpenCL Support	OpenCL 3.0

Performance Context

Performance context should focus on CUDA/HPC workloads and should not use gaming tests as representative unless clearly labeled as unsupported hobbyist use.

Architecture Context

Architecture notes should explain the dual-GPU GK210 design, 24 GB total memory as 12 GB per GPU, passive cooling, PCIe Gen3, and 300 W board power from NVIDIA board specs.

Used-Market Caution

Used-market notes should warn about passive data-center cooling, no normal display outputs, platform support, old CUDA/toolchain support, dual-GPU memory split, and power/case requirements.

Sources

Primary / manufacturer sources

NVIDIA Newsroom, NVIDIA Unveils World's Fastest Accelerator for Data Analytics and Scientific ComputingPrimary launch source for Tesla K80 positioning.Open source
NVIDIA, Tesla K80 GPU Accelerator Board Specification PDFPrimary board-spec source for dual-GPU layout, memory split, cooling, PCIe, and power.Open source
NVIDIA, Tesla K80 GPU Accelerators datasheet PDFPrimary datasheet source for accelerator positioning and feature summary.Open source
NVIDIA, Tesla K80 product pageOfficial product page for data-center accelerator context.Open source
NVIDIA Unveils World's Fastest Accelerator For Data Analytics And Scientific ComputingOfficial November 17, 2014 unveiling and shipping announcement for Tesla K80Open source
NVIDIA Tesla K80 GPU Accelerator board specificationOfficial dual-GPU memory, memory-interface, 300 W power-limit, and CPU 8-pin connector specificationsOpen source

Independent reference sources

TechPowerUp GPU SpecsModel-level specification reference for this GPU entryOpen source

Reference databases

RightNow GPU DatabaseOpen GPU database used to cross-check specifications A reference database: good for cross-checking specifications, but not proof of the original release details on its own.Open source

Performance Context

Architecture Context

Used-Market Caution

Tesla K10

Tesla K20c

Tesla K20m

Tesla K20s

Primary / manufacturer sources

Independent reference sources

Reference databases