Report an Error

NVIDIA Tesla K20m

Graphics Processor
GK110
Cores
2496
TMUs
208
ROPs
40
Memory Size
5 GB
Memory Type
GDDR5
Bus Width
320 bit
The Tesla K20m was a enthusiast-class professional graphics card by NVIDIA, launched in January 2013. Built on the 28 nm process, and based on the GK110 graphics processor, the card supports DirectX 12.0. The GK110 graphics processor is a large chip with a die area of 561 mm² and 7,080 million transistors. Unlike the fully unlocked GeForce GTX TITAN, which uses the same GPU but has all 2688 shaders enabled, NVIDIA has disabled some shading units on the Tesla K20m to reach the product's target shader count. It features 2496 shading units, 208 texture mapping units and 40 ROPs. NVIDIA has placed 5,120 MB GDDR5 memory on the card, which are connected using a 320-bit memory interface. The GPU is operating at a frequency of 706 MHz, memory is running at 1300 MHz.
Being a dual-slot card, the NVIDIA Tesla K20m draws power from 1x 6-pin + 1x 8-pin power connectors, with power draw rated at 225 W maximum. This device has no display connectivity, as it is not designed to have monitors connected to it. Tesla K20m is connected to the rest of the system using a PCI-Express 2.0 x16 interface. The card measures 267 mm in length, and features a dual-slot cooling solution. Its price at launch was 3199 US Dollars.

Graphics Processor

GPU Name
GK110
Architecture
Kepler
Foundry
TSMC
Process Size
28 nm
Transistors
7,080 million
Die Size
561 mm²

Graphics Card

Release Date
Jan 5th, 2013
Generation
Tesla
(Kxx)
Production
End-of-life
Launch Price
3,199 USD
Bus Interface
PCIe 2.0 x16

Relative Performance

3%
GeForce 210
4%
GeForce 9400 GT
5%
Radeon HD 4550
6%
Radeon HD 5450
8%
Radeon HD 6450
8%
GeForce GT 520
10%
GeForce GT 220
13%
GeForce GT 430
13%
Radeon HD 5570
14%
Radeon HD 4670
15%
GeForce GT 440
15%
GeForce GT 240
18%
GeForce 9600 GT
19%
Radeon HD 5670
20%
GeForce GT 640
21%
GeForce 9800 GT
22%
Radeon HD 6670
22%
Radeon HD 4830
24%
Radeon HD 4770
25%
Radeon HD 4850
26%
GeForce GTS 250
26%
GeForce GTS 450
27%
Radeon HD 5750
31%
Radeon HD 7750
31%
Radeon HD 5770
32%
GeForce GTX 260
32%
Radeon HD 4870
32%
Radeon Vega 8
33%
GeForce GTX 550 Ti
33%
GeForce GTX 650
35%
Radeon HD 5830
37%
Radeon HD 6790
37%
Radeon HD 4890
39%
GeForce GTX 460
39%
Radeon RX Vega 11
39%
GeForce GT 1030
40%
GeForce GTX 275
40%
Radeon HD 7770 GHz Edition
40%
GeForce GTX 280
41%
GeForce GTX 465
43%
Radeon HD 6850
44%
GeForce GTX 285
47%
Radeon HD 5850
48%
GeForce GTX 650 Ti
48%
Radeon HD 7790
50%
Radeon RX 550
50%
GeForce GTX 470
51%
Radeon HD 6870
55%
Radeon HD 5870
57%
GeForce GTX 560 Ti
57%
Radeon HD 4870 X2
58%
GeForce GTX 750 Ti
58%
Radeon HD 6950
60%
GeForce GTX 295
61%
GeForce GTX 650 Ti Boost
61%
Radeon HD 7850
63%
GeForce GTX 480
65%
Radeon HD 6970
67%
Radeon R7 265
68%
Radeon R7 370
68%
Radeon RX 560
68%
GeForce GTX 660
68%
GeForce GTX 570
74%
GeForce GTX 950
74%
Radeon HD 7870 GHz Edition
76%
Radeon R9 270X
77%
GeForce GTX 660 Ti
77%
GeForce GTX 580
78%
GeForce GTX 1050
78%
Radeon HD 7950
80%
Radeon RX 460
80%
Radeon HD 5970
82%
GeForce GTX 760
88%
GeForce GTX 670
91%
GeForce GTX 960
93%
Radeon R9 380
93%
Radeon R9 285
94%
Radeon HD 7970
95%
GeForce GTX 680
98%
GeForce GTX 1050 Ti
100%
Tesla K20m
100%
GeForce GTX 770
102%
Radeon HD 6990
105%
Radeon R9 280X
106%
GeForce GTX 590
106%
Radeon HD 7970 GHz Edition
118%
GeForce GTX 780
122%
GeForce GTX 1650
130%
Radeon RX 470
132%
Radeon R9 290
140%
Radeon R9 390
141%
Radeon R9 290X
143%
GeForce GTX 970
145%
GeForce GTX TITAN
146%
Radeon RX 570
147%
GeForce GTX 780 Ti
149%
Radeon R9 390X
150%
Radeon HD 7990
150%
Radeon RX 480
152%
GeForce GTX 690
160%
Radeon RX 580
163%
GeForce GTX 980
165%
GeForce GTX 1060 6 GB
165%
Radeon R9 FURY
166%
Radeon RX 590
182%
Radeon R9 295X2
185%
Radeon R9 FURY X
188%
GeForce GTX 980 Ti
202%
GeForce GTX TITAN X
212%
GeForce GTX 1660 Ti
213%
GeForce GTX 1070
222%
Radeon RX Vega 56
241%
GeForce GTX 1070 Ti
247%
Radeon RX Vega 64
247%
GeForce RTX 2060
254%
GeForce GTX 1080
291%
GeForce RTX 2070
291%
Radeon VII
305%
TITAN X Pascal
313%
GeForce GTX 1080 Ti
338%
GeForce RTX 2080
391%
GeForce RTX 2080 Ti
Based on TPU review data: "Performance Summary" at 1920x1080
Performance estimated based on architecture, shader count and clocks.

Clock Speeds

GPU Clock
706 MHz
Memory Clock
1300 MHz
5200 MHz effective

Memory

Memory Size
5 GB
Memory Type
GDDR5
Memory Bus
320 bit
Bandwidth
208.0 GB/s

Render Config

Shading Units
2496
TMUs
208
ROPs
40
SMX Count
13
L1 Cache
16 KB (per SMX)
L2 Cache
1280 KB

Theoretical Performance

Pixel Rate
36.71 GPixel/s
Texture Rate
146.8 GTexel/s
FP32 (float) performance
3.524 TFLOPS
FP64 (double) performance
1,175 GFLOPS (1:3)

Board Design

Slot Width
Dual-slot
Length
10.5 inches
267 mm
TDP
225 W
Outputs
No outputs
Power Connectors
1x 6-pin + 1x 8-pin
Board Number
P2081 SKU 208

Graphics Features

DirectX
12.0 (11_0)
OpenGL
4.6
OpenCL
1.2
Vulkan
1.1.103
CUDA
3.5
Shader Model
5.1

GK110 GPU Notes

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

GK110 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).