NVIDIA GeForce GTX 970 vs NVIDIA GeForce GTX 660

Compare NVIDIA GeForce GTX 970 4 GB vs NVIDIA GeForce GTX 660 3 GB, specs and GPU benchmark score. Which is the better graphics card for the money?

Benchmark Score
Other Benchmarks
Specifications
Comments (0)
Popular Comparisons

GPU Comparison

Quickly search and compare graphics cards

Benchmark Score

Overall Score

General gaming and workstation score

2262 points

177% significantly better overall score

816 points

Flux Core frame rate

Volumetric ray casting test, a computationally expensive method of rendering high-quality scenes

47 FPS

28 FPS significantly higher Flux Core frame rate

18 FPS

Electron frame rate

Randomly generated noise sphere test

35 FPS

24 FPS significantly higher Electron frame rate

11 FPS

City frame rate

Procedurally generated city scene with voxel rendering

31 FPS

19 FPS significantly higher City frame rate

13 FPS

Clouds frame rate

Real-time noise calculation and ray marching test

28 FPS

19 FPS significantly higher Clouds frame rate

9 FPS

Free GPU Benchmark

Want to compare your graphics card against the GeForce GTX 970 and the GeForce GTX 660? Download our free and quick PC Performance Test.

Download GPU Benchmark

Other Benchmarks

3DMark Graphics score

High-end graphics benchmark

3647 points

176% significantly higher 3DMark Graphics score

1317 points

Blender score

Cycles Render (Samples per minute)

289.94 points

136% significantly higher Blender score

122.56 points

Geekbench 5 (CUDA) score

CUDA compute benchmark

25908 points

201% significantly higher Geekbench 5 (CUDA) score

8583 points

Geekbench 5 (Metal) score

Metal compute benchmark

13595 points

149% significantly higher Geekbench 5 (Metal) score

5449 points

Geekbench 5 (OpenCL) score

OpenCL compute benchmark

29153 points

156% significantly higher Geekbench 5 (OpenCL) score

11381 points

Geekbench 5 (Vulkan) score

Vulkan compute benchmark

34107 points

218% significantly higher Geekbench 5 (Vulkan) score

10693 points

Geekbench 6 (Metal) score

Metal compute benchmark

13595 points

215% significantly higher Geekbench 6 (Metal) score

4311 points

Geekbench 6 (OpenCL) score

OpenCL compute benchmark

29277 points

157% significantly higher Geekbench 6 (OpenCL) score

11353 points

Geekbench 6 (Vulkan) score

Vulkan compute benchmark

32348 points

172% significantly higher Geekbench 6 (Vulkan) score

11858 points

Specifications

Core Clock Speed

Nominal chip frequency

1050 MHz

7% slightly higher core clock speed

980 MHz

Boost Clock Speed

Higher turbo frequency when power conditions are met

1178 MHz

14% slightly higher boost clock speed

1032 MHz

Memory

Onboard memory size for textures and vertices

4 GiB

33% slightly more memory

3 GiB

Memory Bus Width

Number of parallel lines to the memory chips

256 Bit

33% slightly larger memory bus width

192 Bit

Memory Bandwidth

Data transfer speed between GPU core and memory

224.4 GB/s

55% significantly higher memory bandwidth

144.2 GB/s

TDP

Thermal Design Power: Measure of heat generated by the GPU

145 W

140 W

3% slightly lower TDP

Pixel Rate

Number of pixels that can be rendered per second

65.97 Gigapixels/s

219% significantly higher pixel rate

20.64 Gigapixels/s

Texture Rate

Number of textured pixels that can be rendered per second

122.5 Gigatexels/s

48% significantly higher texture rate

82.56 Gigatexels/s

Floating Point Performance

Raw number of floating point operations per second

3.92 TFLOPS

97% significantly better floating point performance

1.98 TFLOPS

Shading Units

Number of processors dedicated to shader processing

1664

73% significantly more shading units

960

Texture Mapping Units

Number of processors dedicated to applying textures

104

30% slightly more texture mapping units

Render Output Processors

Number of processors dedicated to final pixel rendering

133% significantly more render output processors

Other details

Rank

Ranking in the hardwareDB database

193rd of 586

277th of 586

Release date

The official date of release of this chip

2014 September

Memory Type

The type of memory used by this chip

GDDR5

DirectX Support

Maximum version of DirectX supported

12.0

11.0

OpenGL Support

Maximum version of OpenGL supported

4.5

GeForce GTX 970 vs GeForce GTX 660 comparison

For gaming, the GeForce GTX 970 graphics card is better than the GeForce GTX 660 in our tests.

The GeForce GTX 970 has a slightly higher core clock speed. The core clock speed (or base speed) is the frequency at which the GPU core runs. This metric makes sense when comparing GPUs of a similar architecture or generation. In addition, the GeForce GTX 970 also has a slightly higher boost clock speed. This is a frequency that can be reached if the GPU is cooled sufficiently and has enough power.

As stated by the manufacturer, the GeForce GTX 970 has slightly more memory with 4 GiB of memory compared to 3 GiB. Memory size doesn't directly affect performance, but too little memory will certainly degrade gaming performance.

In addition, the GeForce GTX 660 has a slightly lower TDP at 140 W when compared to the GeForce GTX 970 at 145 W. TDP (Thermal Design Power) measures total heat output from the chip.

In terms of raw gaming performance in our GPU benchmark, the GeForce GTX 970 is better than the GeForce GTX 660.