GeForce GTX 660 Ti vs Radeon R7 384 Cores Kaveri

In this comparison between GeForce GTX 660 Ti and Radeon R7 384 Cores Kaveri you will find out which graphics card performs better in today's games. Bear in mind that third-party versions may have more efficient cooling and higher clock speeds. This will increase cards' performance, though not by much. In addition to raw power you should also take into account the dimensions. Thicker models simply will not fit into a small mini-ITX case. The resolution of your monitor also affects the choice, since 4K gameplay requires a more powerful GPU. And don't overspend on the graphics card. Other parts of your build may also need to be upgraded, save some money for the CPU or power supply. For some people GeForce GTX 660 Ti will be the best choice, for others Radeon R7 384 Cores Kaveri will be their preference. Study the comparison tables below and make your choice.
Radeon R7 384 Cores Kaveri

Main Specs

  GeForce GTX 660 Ti Radeon R7 384 Cores Kaveri
Power consumption (TDP) 150 Watt
Interface PCIe 3.0 x16
Supplementary power connectors Two 6-pin
Memory type GDDR5
Maximum RAM amount 2 GB
Display Connectors 2x DVI, 1x HDMI, 1x DisplayPort
  Check Price
  • Both cards are used in Desktops.
  • GeForce GTX 660 Ti is build with Kepler architecture, and Radeon R7 384 Cores Kaveri - with GCN.
  • Core clock speed of GeForce GTX 660 Ti is 195 MHz higher, than Radeon R7 384 Cores Kaveri.
  • GeForce GTX 660 Ti and Radeon R7 384 Cores Kaveri are manufactured by 28 nm process technology.

Game benchmarks

high / 1080p 21−24 7−8
ultra / 1080p 12−14 4−5
QHD / 1440p 8−9 0−1
4K / 2160p 5−6
low / 720p 40−45 18−20
medium / 1080p 27−30 9−10
The average gaming FPS of GeForce GTX 660 Ti in Assassin's Creed Odyssey is 160% more, than Radeon R7 384 Cores Kaveri.

Full Specs

  GeForce GTX 660 Ti Radeon R7 384 Cores Kaveri
Architecture Kepler GCN
Code name GK104 Kaveri Spectre
Type Desktop Desktop
Release date 16 August 2012 14 January 2014
Pipelines 1344 384
Core clock speed 915 MHz 720 MHz
Boost Clock 980 MHz
Transistor count 3,540 million
Manufacturing process technology 28 nm 28 nm
Texture fill rate 102.5 billion/sec
Floating-point performance 2,459.5 gflops
Length 9.5" (24.1 cm)
Memory bus width 192-bit GDDR5
Memory clock speed 6.0 GB/s
Memory bandwidth 144.2 GB/s
Shared memory - +
DirectX 12 (11_0) DirectX 12 (FL 12_0), Shader 5.2
Shader Model 5.1
OpenGL 4.3
OpenCL 1.2
Vulkan 1.1.126
CUDA +
CUDA cores 1344
Bus support PCI Express 3.0
Height 4.376" (11.1 cm)
SLI options +
Multi monitor support 4 displays
HDMI +
HDCP +
Maximum VGA resolution 2048x1536
Audio input for HDMI Internal
3D Blu-Ray +
3D Gaming +
3D Vision +
Bitcoin / BTC (SHA256) 99 Mh/s
  Check Price