GeekRadar / The comparison desk

COMPARE.

GPU

NVIDIA

Tesla K20X

Graphics processor

Video memory: 6 GB^*
Memory bus: 384 bit^*
Rated memory bandwidth: 249.60 GB/s^*

No matched benchmark results are available for this entry yet.

Processing & platform

Architecture: Reported value 1: Kepler^*
Reported value 2: Kepler^*
Reported value 3: Kepler^*
Reported value 4: Kepler GK110^*
Reported value 5: Kepler^*
Process node: 28 nm^*
GPU cores: 2,688^*
GPU base clock: 732 MHz^*
Reported GPU frequency: 732 MHz^*
Reported graphics memory frequency: 1,300 MHz^*
Graphics interface: PCIe 3.0 x16^*
Video memory: 6,144 MB^*
Reported chip die area: 561 mm²^*
Chip/package TDP: 235 W^*
Transistors: 7,080 million^*
Graphics processor: GK110^*
Texture mapping units: 224^*
Raster output units: 48^*
Chip foundry: TSMC^*
Card or module format: Dual-slot^*
Declared GPU TDP: 235 W^*
Suggested system PSU: 550 W^*
External power connectors: 1x 6-pin + 1x 8-pin^*
DirectX / feature level: 12 (11_0)^*
OpenGL: 4.6^*
OpenCL: 3.0^*
Vulkan: 1.2.175^*
CUDA compute capability: 3.5^*
Shader model: 6.5 (5.1)^*
GPU L2 cache: 1,536 KB^*

Memory & storage

Memory options: 6^*
Video memory: 6 GB^*
Video memory type: Reported value 1: GDDR5^*
Reported value 2: GDDR5^*
Reported value 3: GDDR5^*
Reported value 4: GDDR5^*
Reported value 5: GDDR6^*
Reported value 6: GDDR5^*
Reported value 7: GDDR5^*
Memory bus: 384 bit^*
Rated memory bandwidth: 249.60 GB/s^*

Power specifications

Thermal design power: 235 W^*

System configuration

Detailed specifications

Specifications 1 · 11 fields

Manufacturer: Nvidia^*
GFLOPS: 3,935^* GFLOPS
TDP: 235^*
Class: Datacenter^*
Name: Tesla K20X^*
Year: 2012^*
Fab: 28^*
Transistors (mln): 7,080^*
Die size: 561^*
Memory size: 6,144^*
Memory speed: 249.6^*

Specifications 2 · 24 fields

Architecture: Kepler GK110^*
Integrated: No^*
L2 Cache: 1536KB^*
Manufacturer: Nvidia^*
Max Power: 235 Watts^*
Memory: 6,144^* MB
Memory Bandwidth: 249.6GB/sec^*
Memory Bus: 384 Bit^*
Memory Speed: 1,300^* MHz
Memory Type: GDDR5^*
Name: Tesla K20X^*
Notebook GPU: No^*
Open GL: 4.5^*
Process: 28nm^*
Release Date: 12-Nov-2012^*
Resolution WxH: 4096x2160^*
SLI Crossfire: Yes^*
Shader: 5^*
VGA Connection: 0^*
Core Speed: 732^* MHz
DVI Connection: 0^*
Dedicated: Yes^*
Direct X: DX 12.0^*
HDMI Connection: 0^*

Specifications 3 · 8 fields

Product Name: Tesla K20X^*
GPU Chip: GK110^*
Released: Nov 12th, 2012^*
Bus: PCIe 3.0 x16^*
Memory: 6 GB, GDDR5, 384 bit^*
GPU clock: 732^* MHz
Memory clock: 1,300^* MHz
Shaders TMUs ROPs: 2688 / 224 / 48^*

Specifications 4 · 12 fields

Foundry: TSMC^*
Vendor: NVIDIA^*
FP32 GFLOPS: 3,935^* GFLOPS
FP64 GFLOPS: 1,312^* GFLOPS
Product: NVIDIA Tesla K20X^*
Type: GPU^*
Release Date: 2012-11-12^*
Process Size (nm): 28^* nm
TDP (W): 235^* W
Die Size (mm^2): 561^* mm²
Transistors (million): 7,080^*
Freq (MHz): 732^* MHz

Specifications 5 · 25 fields

proc transistors: 7,080 million^*
proc die size: 561^* mm²
cores: 2,688^*
memory clock: 1,300^*
ram: 6^*
name: NVIDIA Tesla K20X^*
ram type: GDDR5^*
bus width: 384^*
bus interface: PCIe 3.0 x16^*
ram bandwidth: 249.6^* GB/s
fp32 performance: 3.935^*
fp64 performance: 1.3117^*
tdp: 235^* W
suggested psu: 550^* W
slot width: Dual-slot^*
power connectors: 1x 6-pin + 1x 8-pin^*
api cuda: 3.5^*
is retail board: No^*
gpi value: 3.69^*
vendor: nvd^*
manufacturer: NVIDIA^*
arch name: Kepler^*
card release date: 2012-11-12^*
proc foundry: TSMC^*
proc process size: 28^* nm

Specifications 6 · 36 fields

CUDA compute capability

3.5^*

DirectX / feature level

12 (11_0)^*

OpenCL

3.0^*

OpenGL

4.6^*

Shader model

6.5 (5.1)^*

Vulkan

1.2.175^*

Architecture

Kepler^*

Graphics interface

PCIe 3.0 x16^*

Memory bus

384^* bit

Reported shader processors

2,688^*

Reported theoretical FP32 peak

3.935^* TFLOPS

Scope and conditions

Arithmetic density: Not stated^*

Reported theoretical FP64 peak

1.3117^* TFLOPS

Scope and conditions

Arithmetic density: Not stated^*

Declared L1 cache configuration

16 KB (per SMX)^*

Declared GPU L2 cache

1536 KB^*

Reported display outputs

No outputs^*

Theoretical pixel fill rate

40.99 GPixel/s^*

External power connectors

1x 6-pin + 1x 8-pin^*

Reported chip die area

561 mm²^*

Chip foundry

TSMC^*

Process node declaration

28 nm^*

Transistors

7,080 million^*

Graphics processor

GK110^*

Reported video memory

6^* GB

Rated memory bandwidth

249.6 GB/s^*

Video memory type

GDDR5^*

Raster output units

48^*

Reported shading units

2,688^*

Card or module format

Dual-slot^*

Suggested system PSU

550 W^*

2:4 tensor sparsity

No^*

Declared GPU TDP

235 W^*

Theoretical texture fill rate

164.0 GTexel/s^*

Texture mapping units

224^*

GPU cores

2,688^*

Video memory

6^* GB

Video memory type

GDDR5^*

Specifications 7 · 8 fields

Product Name: Tesla K20X^*
GPU Chip: GK110^*
Released: Nov 12th, 2012^*
Bus: PCIe 3.0 x16^*
Memory: 6 GB, GDDR5, 384 bit^*
GPU clock: 732^* MHz
Memory clock: 1,300^* MHz
Shaders / TMUs / ROPs: 2688 / 224 / 48^*

Specifications 8 · 11 fields

Manufacturer: Nvidia^*
GFLOPS: 3,935^* GFLOPS
TDP: 235^*
Class: Datacenter^*
Name: Tesla K20X^*
Year: 2012^*
Fab: 28^*
Transistors (mln): 7,080^*
Die size: 561^*
Memory size: 6,144^*
Memory speed: 249.6^*

Specifications 9 · 54 fields

Brand: NVIDIA^*
Top BUS WIDTH: 384^* bit
Graphics Processor GPU Name: GK110^*
Graphics Processor Architecture: Kepler^*
Graphics Features CUDA: 3.5^*
Render Config SMX Count: 14^*
Graphics Processor Process Size: 28^* nm
Graphics Processor Transistors: 7,080 million^*
Graphics Processor Die Size: 561^* mm²
Graphics Card Release Date: Nov 12th, 2012^*
Graphics Card Generation: Tesla Kepler (Kxx)^*
Graphics Card Successor: Tesla Maxwell^*
Graphics Card Production: End-of-life^*
Name: Tesla K20X^*
Graphics Card Bus Interface: PCIe 3.0 x16^*
Clock Speeds GPU Clock: 732^* MHz
Clock Speeds Memory Clock: 1300 MHz 5.2 Gbps effective^*
Memory Memory Size: 6^* GB
Memory Memory Type: GDDR5^*
Memory Memory Bus: 384^* bit
Memory Bandwidth: 249.6^* GB/s
Render Config TMUs: 224^*
Top GRAPHICS PROCESSOR: GK110^*
Render Config ROPs: 48^*
Theoretical Performance Pixel Rate: 40.99 GPixel/s
Theoretical Performance Texture Rate: 164.0 GTexel/s
Board Design Slot Width: Dual-slot^*
Board Design TDP: 235^* W
Board Design Suggested PSU: 550^* W
Board Design Outputs: No outputs^*
Board Design Board Number: P2081^*
Graphics Features DirectX: 12 (11_0)^*
Graphics Features OpenGL: 4.6^*
Graphics Features OpenCL: 3^*
Graphics Features Vulkan: 1.2.175^*
source file: nvidia2012.json^*
Graphics Processor Foundry: TSMC^*
Graphics Card Predecessor: Tesla Fermi^*
Graphics Card Launch Price: 7,699 USD^*
Board Design Length: 267 mm 10.5 inches^*
Board Design Power Connectors: 1x 6-pin + 1x 8-pin^*
Graphics Processor Density: 12.6M / mm²^*
Top TMUS: 224^*
Graphics Processor Chip Package: BGA-2152^*
Top ROPS: 48^*
Graphics Features Shader Model: 6.5 (5.1)^*
Top CORES: 2,688^*
Render Config Shading Units: 2,688^*
Render Config L2 Cache: 1,536^* KB
Top MEMORY SIZE: 6^* GB
Theoretical Performance FP32 (float): 3.935^* TFLOPS
Theoretical Performance FP64 (double): 1,311.7 GFLOPS (1:3)
Render Config L1 Cache: 16 KB (per SMX)^*
Top MEMORY TYPE: GDDR5^*

Specifications 10 · 24 fields

model: Tesla K20X^*
base clock mhz: 732^* MHz
max boost clock mhz: 0^* MHz
bus type: GDDR5^*
bus width: 384^*
vram: 6^*
clock speed: 5,200^*
bandwith: 250^*
fp16: 0^*
fp32: 3,935^*
fp64: 1,312.00^*
price: 24.99^*
tdp: 235^*
nv link: No^*
bus interface: PCIe 3.0 x16^*
eol: Yes^*
eol date: 12/2019^*
market: Legacy^*
ecc equipped: Yes^*
architecture: Kepler^*
release date: 4700^*
tensor cores: 0^*
tensor core gen: No Tensor Cores^*
cuda cores: 2,688^*

Specifications 11 · 27 fields

manufacturer: NVIDIA^*
manufacturer: NVIDIA^*
rop: 64^*
rop: 48^*
igp: No^*
igp: No^*
bus: PCIe 3.0 x16^*
bus: PCIe 3.0 x16^*
memType: GDDR5^*
memType: GDDR6^*
gpuChip: GK110^*
gpuChip: TU106^*
productName: Tesla K20X^*
productName: Tesla K20X^*
releaseYear: 2,012^*
releaseYear: 2,012^*
memSize: 6^*
memSize: 6^*
memBusWidth: 384^*
gpuClock: 732^*
gpuClock: 885^*
memClock: 1,300^*
memClock: 1,500^*
unifiedShader: 2,304^*
unifiedShader: 2,688^*
tmu: 224^*
tmu: 144^*

* Provisional data