NVIDIA GK110

GK110

GK110-300-A1

GK110-400-A1

GK110-885-KA-A1

Fritzchens Fritz

Die Shot

Block Diagram

SMX Diagram

NVIDIA's GK110 GPU uses the Kepler architecture and is made using a 28 nm production process at TSMC. With a die size of 561 mm² and a transistor count of 7,080 million it is a very big chip. GK110 supports DirectX 12 (Feature Level 11_0). For GPU compute applications, OpenCL version 3.0 and CUDA 3.5 can be used. It features 2880 shading units, 240 texture mapping units and 48 ROPs.

Further reading: GK110 Compute Architecture Whitepaper

Graphics Processor

GPU Name: GK110

Codename: NVF0

Architecture: Kepler

Foundry: TSMC

Process Size: 28 nm

Transistors: 7,080 million

Density: 12.6M / mm²

Die Size: 561 mm²

Released: Nov 12th, 2012

Graphics Features

DirectX: 12 (11_0)

OpenGL: 4.6

OpenCL: 3.0

Vulkan: 1.1

CUDA: 3.5

Shader Model: 5.1

PureVideo HD: VP5

VDPAU: Feature Set D

Render Config

Shading Units: 2880

TMUs: 240

ROPs: 48

SMX Count: 15

DPUs: 960

SFUs: 480

GPCs: 5

PPC: 8 per Rasterizer

Tex L1 Cache: 48 KB per SMX

L1 Cache: 16 KB per SMX

L2 Cache: 1536 KB

Max. TDP: 250 W

All Kepler GPUs

NVIDIA GPU Architecture History

1998-2000 Fahrenheit
1999-2005 Celsius
2001-2003 Kelvin
2003-2005 Rankine
2003-2013 Curie
2006-2010 Tesla
2007-2013 Tesla 2.0
2010-2016 Fermi
2010-2013 VLIW Vec4
2010-2016 Fermi 2.0
2012-2018 Kepler
2013-2015 Kepler 2.0
2014-2017 Maxwell
2014-2019 Maxwell 2.0
2016-2021 Pascal
2017-2020 Volta
2018-2022 Turing
2020-2022 Ampere
2022 Hopper
2022-2023 Lovelace

Graphics cards using the NVIDIA GK110 GPU

Name	Chip	Memory	Shaders	TMUs	ROPs	GPU Clock	Memory Clock
NVIDIA Tesla K20c	GK110-885-KA-A1	5 GB	2496	208	40	706 MHz	1300 MHz
NVIDIA GeForce GTX 780	GK110-300-A1	3 GB	2304	192	48	863 MHz	1502 MHz
NVIDIA Tesla K20Xm		6 GB	2688	224	48	732 MHz	1300 MHz
NVIDIA GeForce GTX TITAN	GK110-400-A1	6 GB	2688	224	48	836 MHz	1502 MHz
NVIDIA Tesla K20m		5 GB	2496	208	40	706 MHz	1300 MHz
NVIDIA Tesla K20s		5 GB	2496	208	40	575 MHz	1300 MHz
NVIDIA Tesla K20X		6 GB	2688	224	48	732 MHz	1300 MHz
NVIDIA Quadro K6000 SDI		12 GB	2880	240	48	902 MHz	1502 MHz

GK110 GPU Notes

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

GK110 has 5 GPCs each capable of 8 pixels per clock. This limits complete GPU 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.
Additional ROPs however can be used for MSAA (because it doesn't require additional data from rasterizers, while giving more work to ROPs.)

New Forum Posts

09:16 by Zareek
What causes driver corruption? (62)
09:08 by Tigger
Monitor recommendation 1440p 144hz below £300 (19)
09:07 by Chomiq
PerfCap Reason PWR / GPU load 99% (4 x 3090 RTX) - Workstation Rendering issues (7)
09:04 by Noirua
Help undervolting i7 8750h Throttlestop (248)
08:57 by phill
Now, time for some possible overkill.... Network upgrade... (8)
08:57 by Splinterdog
Windows 11 General Discussion (3163)
08:34 by CallandorWoT
What's your latest tech purchase? (11983)
08:34 by bobbybluz
Asus Sabretooth Z97 Mark one-strange booting behaviour, not booting (15)
08:33 by sneekypeet
Very easy to mount cpu cooler for a ryzen 5800x (29)
08:00 by btk2k2
CPU Limited games people would like in the next CPU rebench. (64)