GeForce GTX 650 vs Radeon HD 6870

If you are going to buy a new graphics card and are choosing between GeForce GTX 650 and Radeon HD 6870, there are a couple of things to consider. Cards with more VRAM in general perform better and allow you to play on higher graphics settings. Size also makes a difference. A model with a large heatsink can occupy up to three expansion slots on a motherboard. Be sure you have enough room in your PC case. When comparing GPUs with different architectures, more processing cores and even higher TFLOPS will not always translate to better performance. To help you decide which GPU you need, we have measured frame rates in a number of popular games. For more on how the GeForce GTX 650 stacks up against Radeon HD 6870, check out specs charts below.

Main Specs

  GeForce GTX 650 Radeon HD 6870
Power consumption (TDP) 64 Watt 151 Watt
Interface PCIe 3.0 x16 PCIe 2.0 x16
Supplementary power connectors One 6-pin 2x 6-pin
Memory type GDDR5 GDDR5
Maximum RAM amount 2 GB 1 GB
Display Connectors 2x DVI, 1x mini-HDMI 2x DVI, 1x HDMI, 2x mini-DisplayPort
  Check Price Check Price
  • Radeon HD 6870 has 135% more power consumption, than GeForce GTX 650.
  • GeForce GTX 650 is connected by PCIe 3.0 x16, and Radeon HD 6870 uses PCIe 2.0 x16 interface.
  • GeForce GTX 650 has 1 GB more memory, than Radeon HD 6870.
  • Both cards are used in Desktops.
  • GeForce GTX 650 is build with Kepler architecture, and Radeon HD 6870 - with Terascale 2.
  • GeForce GTX 650 is manufactured by 28 nm process technology, and Radeon HD 6870 - by 40 nm process technology.
  • Radeon HD 6870 is 215 mm longer, than GeForce GTX 650.
  • Memory clock speed of Radeon HD 6870 is 1045 MHz higher, than GeForce GTX 650.

Game benchmarks

high / 1080p 8−9 10−12
ultra / 1080p 5−6 6−7
QHD / 1440p 0−1 1−2
low / 720p 18−20 24−27
medium / 1080p 10−11 12−14
The average gaming FPS of Radeon HD 6870 in Assassin's Creed Odyssey is 40% more, than GeForce GTX 650.

Full Specs

  GeForce GTX 650 Radeon HD 6870
Architecture Kepler Terascale 2
Code name GK107 RV940 Barts
Type Desktop Desktop
Release date 6 September 2012 22 October 2010
Pipelines 384 1120
Core clock speed 1058 MHz
Boost Clock 900 MHz
Transistor count 1,270 million 1,700 million
Manufacturing process technology 28 nm 40 nm
Texture fill rate 33.9 billion/sec 50.40
Floating-point performance 812.5 gflops 2,016.0 gflops
Length 5.70" (14.5 cm) 220 mm
Memory bus width 128-bit GDDR5 256 Bit
Memory clock speed 5.0 GB/s 1050 MHz
Memory bandwidth 80.0 GB/s 134.4 GB/s
Shared memory -
DirectX 12 (11_0)
Shader Model 5.1 5.0
OpenGL 4.3 4.4
OpenCL 1.2 1.2
Vulkan 1.1.126
CUDA +
CUDA cores 384
Bus support PCI Express 3.0 PCIe 2.0 x16
Height 4.38" (11.1 cm)
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) 26 Mh/s 213 Mh/s
Eyefinity +
Design reference
CrossFire +
  Check Price Check Price