AMD Radeon 660M
vs
AMD Radeon 680M

vs
AMD Radeon 660M vs AMD Radeon 680M graphics card comparison

GPU Comparison Result

AMD Radeon 660M vs Radeon 680M: How Much Faster is the 680M in Games and 3DMark

The AMD Radeon 660M and Radeon 680M utilize the same RDNA 2 architecture but belong to different levels of integrated graphics. The Radeon 680M has twice as many compute units and a higher clock speed. This does not result in double the performance, but in 3DMark, the advantage of the higher model typically reaches 45-55%. The specific result depends on RAM speed, power limits, and laptop cooling.

Parameter AMD Radeon 660M AMD Radeon 680M
Architecture RDNA 2 RDNA 2
Compute Units 6 CU 12 CU
Stream Processors 384 768
Max Clock Speed up to 1900 MHz up to 2200 MHz
Typical Processors Ryzen 5 6600U, 6600H, 7535U Ryzen 7 6800U, 6800H, 7735U
Main Class of Devices Budget and office laptops Performance laptops without discrete graphics

Both models use system memory instead of their own video memory. Therefore, a laptop with fast LPDDR5-6400 can significantly outperform a device with DDR5-4800, even if they both have the same GPU.

Why the Radeon 680M is Not Twice as Fast

The Radeon 680M has double the compute units: 12 compared to 6. The clock speed is also higher-up to 2200 MHz instead of 1900 MHz. However, the performance of integrated graphics does not scale linearly with the number of compute units.

The CPU and GPU are housed in the same chip package and share a common power limit. Under simultaneous load, the system redistributes the available energy between CPU cores and graphics. In thin laptops, cooling becomes an additional limitation.

The bandwidth of the RAM does not increase with the number of compute units either. The Radeon 680M quickly hits the limits of DDR5 or LPDDR5, so some of its resources are not always fully utilized.

According to average test results, the Radeon 680M outperforms the Radeon 660M by about 53% in 3DMark Time Spy Graphics and by 46% in Fire Strike Graphics. This is a significant gap, but far from double.

Results in Specific Laptops

The table below shows the Graphics Score, not the overall 3DMark score. This reduces the influence of CPU core performance on the comparison.

Integrated Graphics Laptop and Processor CPU Package Power Limits Fire Strike Graphics Time Spy Graphics
Radeon 660M Lenovo ThinkPad L14 G5, Ryzen 5 7535U 30 / 25 W 3527 1016
Radeon 660M Lenovo ThinkPad Z13, Ryzen 5 PRO 6650U up to 28 W 4487 1588
Radeon 660M Lenovo IdeaPad 5 Pro 16ARH7, Ryzen 5 6600HS 60 / 42 W 4993 1580
Radeon 680M Asus Zenbook S 13 OLED, Ryzen 7 6800U 30 / 25 W 6683 2287
Radeon 680M Lenovo ThinkPad T14 G3, Ryzen 7 PRO 6850U 30 / 25 W 7050 2289
Radeon 680M HP Dragonfly Pro, Ryzen 7 7736U 51 / 40 W 7549 2607

The first value of the limit indicates short-term power, while the second indicates sustained power. This is the total power of the CPU package, within which the CPU and integrated graphics share the available watts.

The table shows that even a fast implementation of the Radeon 660M does not catch up to a typical Radeon 680M. The IdeaPad 5 Pro with the Ryzen 5 6600HS and a high power limit scored 4993 points in Fire Strike Graphics. The Asus Zenbook S 13 OLED with the Radeon 680M achieved 6683 points with significantly lower limits.

There is also considerable variation within a single model. The Radeon 660M in the ThinkPad L14 G5 scored only 1016 points in Time Spy Graphics, while in the ThinkPad Z13, the score reached 1588 points. The difference exceeds 50%, even though both graphics belong to the same model.

The HP Dragonfly Pro shows the upper limits of the Radeon 680M's capabilities. The high score was achieved not only due to LPDDR5-6400 but also due to aggressive CPU package limits of 51/40 W. Such graphics would perform slower in a thinner or quieter configured laptop.

Performance in Games

The Radeon 660M is primarily designed for less demanding games. It works best at resolutions of 1280 × 720 or 1600 × 900 with low settings. In esports titles, Full HD can be used, but the outcome heavily depends on the specific game and the laptop's memory.

The Radeon 680M operates much more confidently at 1920 × 1080. In older, networked, and moderately demanding games, low or medium settings are often available. In heavy modern titles, it still requires lowering graphics quality or enabling FSR.

The practical difference looks like this:

  • where the Radeon 660M requires 720p and minimum preset, the Radeon 680M often allows for switching to 900p or 1080p;
  • if the Radeon 660M delivers borderline frame rates, the Radeon 680M is more likely to provide a more stable result;
  • in the most demanding games, both graphics remain a limited solution and do not replace a discrete graphics card.

The Radeon 680M is also better suited for working with graphic applications, hardware acceleration, and connecting high-resolution monitors. However, the performance gain depends on the software: not all tasks can utilize the additional compute units.

What to Check Before Buying a Laptop

The name of the integrated graphics does not guarantee a certain level of performance. Before purchasing, it is important to check:

  • the type and speed of the RAM;
  • whether the memory operates in dual-channel mode;
  • power limits of the CPU;
  • 3DMark results specifically for the chosen laptop model;
  • device behavior under prolonged load;
  • the possibility of upgrading the memory.

Particular caution should be taken with laptops that have a single memory stick or limited power limits. In such configurations, the Radeon 680M can lose a significant part of its advantages, while the Radeon 660M can drop to the level of much weaker integrated graphics.

Which Model to Choose

The Radeon 660M is suitable for office tasks, browsing, video watching, programming, and occasional undemanding gaming. It is justified in a budget laptop if graphical performance is not a primary criterion.

The Radeon 680M should be chosen for a laptop without a discrete graphics card that is intended for gaming or graphic applications. The difference of 45-55% is large enough to be felt not only in tests but also under real workloads.

Conclusion

The Radeon 680M is significantly faster than the Radeon 660M due to the doubled number of compute units and higher clock speed. On average, the higher graphics outperform the lower one by about 46% in Fire Strike and 53% in Time Spy.

There is no doubling of performance due to limited memory bandwidth, the shared power limit with the CPU, and cooling capabilities. Nevertheless, the Radeon 680M is a full step above: it handles Full HD better and provides more headroom in games and graphic applications.

The Radeon 660M remains a basic solution for business laptops. The Radeon 680M is a more suitable configuration for a device without a discrete graphics card.

Advantages

  • Higher Boost Clock: 2200MHz (1900MHz vs 2200MHz)
  • More Shading Units: 768 (384 vs 768)

Basic

AMD
Label Name
AMD
January 2022
Launch Date
January 2022
Integrated
Platform
Integrated
Radeon 660M
Model Name
Radeon 680M
Rembrandt
Generation
Navi II IGP
1500MHz
Base Clock
2000MHz
1900MHz
Boost Clock
2200MHz
PCIe 4.0 x8
Bus Interface
PCIe 4.0 x8
13,100 million
Transistors
13,100 million
6
RT Cores
12
6
Compute Units
12
24
TMUs
?
Texture Mapping Units (TMUs) serve as components of the GPU, which are capable of rotating, scaling, and distorting binary images, and then placing them as textures onto any plane of a given 3D model. This process is called texture mapping.
48
TSMC
Foundry
TSMC
6 nm
Process Size
6 nm
RDNA 2.0
Architecture
RDNA 2.0

Memory Specifications

System Shared
Memory Size
System Shared
System Shared
Memory Type
System Shared
System Shared
Memory Bus
?
The memory bus width refers to the number of bits of data that the video memory can transfer within a single clock cycle. The larger the bus width, the greater the amount of data that can be transmitted instantaneously, making it one of the crucial parameters of video memory. The memory bandwidth is calculated as: Memory Bandwidth = Memory Frequency x Memory Bus Width / 8. Therefore, when the memory frequencies are similar, the memory bus width will determine the size of the memory bandwidth.
System Shared
SystemShared
Memory Clock
SystemShared
System Dependent
Bandwidth
?
Memory bandwidth refers to the data transfer rate between the graphics chip and the video memory. It is measured in bytes per second, and the formula to calculate it is: memory bandwidth = working frequency × memory bus width / 8 bits.
System Dependent

Display and Media

No outputs
Outputs
Portable Device Dependent

Theoretical Performance

30.40 GPixel/s
Pixel Rate
?
Pixel fill rate refers to the number of pixels a graphics processing unit (GPU) can render per second, measured in MPixels/s (million pixels per second) or GPixels/s (billion pixels per second). It is the most commonly used metric to evaluate the pixel processing performance of a graphics card.
70.40 GPixel/s
45.60 GTexel/s
Texture Rate
?
Texture fill rate refers to the number of texture map elements (texels) that a GPU can map to pixels in a single second.
105.6 GTexel/s
2.918 TFLOPS
FP16 (half)
?
An important metric for measuring GPU performance is floating-point computing capability. Half-precision floating-point numbers (16-bit) are used for applications like machine learning, where lower precision is acceptable. Single-precision floating-point numbers (32-bit) are used for common multimedia and graphics processing tasks, while double-precision floating-point numbers (64-bit) are required for scientific computing that demands a wide numeric range and high accuracy.
6.758 TFLOPS
91.20 GFLOPS
FP64 (double)
?
An important metric for measuring GPU performance is floating-point computing capability. Double-precision floating-point numbers (64-bit) are required for scientific computing that demands a wide numeric range and high accuracy, while single-precision floating-point numbers (32-bit) are used for common multimedia and graphics processing tasks. Half-precision floating-point numbers (16-bit) are used for applications like machine learning, where lower precision is acceptable.
211.2 GFLOPS
1.43 TFLOPS
FP32 (float)
?
An important metric for measuring GPU performance is floating-point computing capability. Single-precision floating-point numbers (32-bit) are used for common multimedia and graphics processing tasks, while double-precision floating-point numbers (64-bit) are required for scientific computing that demands a wide numeric range and high accuracy. Half-precision floating-point numbers (16-bit) are used for applications like machine learning, where lower precision is acceptable.
3.311 TFLOPS

Miscellaneous

384
Shading Units
?
The most fundamental processing unit is the Streaming Processor (SP), where specific instructions and tasks are executed. GPUs perform parallel computing, which means multiple SPs work simultaneously to process tasks.
768
128 KB per Array
L1 Cache
128 KB per Array
2MB
L2 Cache
2MB
15W
TDP
50W
1.2
Vulkan Version
?
Vulkan is a cross-platform graphics and compute API by Khronos Group, offering high performance and low CPU overhead. It lets developers control the GPU directly, reduces rendering overhead, and supports multi-threading and multi-core processors.
1.3
2.0
OpenCL Version
2.0
4.6
OpenGL
4.6
12 Ultimate (12_2)
DirectX
12 Ultimate (12_2)
None
Power Connectors
None
16
ROPs
?
The Raster Operations Pipeline (ROPs) is primarily responsible for handling lighting and reflection calculations in games, as well as managing effects like anti-aliasing (AA), high resolution, smoke, and fire. The more demanding the anti-aliasing and lighting effects in a game, the higher the performance requirements for the ROPs; otherwise, it may result in a sharp drop in frame rate.
32
6.5
Shader Model
6.7

Benchmarks

FP32 (float) / TFLOPS
Radeon 660M
1.43
Radeon 680M
3.311 +132%
3DMark Time Spy
Radeon 660M
1526
Radeon 680M
2399 +57%
Blender
Radeon 660M
92
Radeon 680M
249 +171%