AMD Radeon 660M

AMD Radeon 660M
AMD Radeon 660M graphics card review

AMD Radeon 660M: Specifications, Game Benchmarks, and Comparison with Radeon 680M

The Radeon 660M brings RDNA 2 architecture to budget laptops but has only six compute units-half the number of the Radeon 680M. This is sufficient for Dota 2, GTA V, and other relatively light games. In modern AAA titles, however, users need to select low settings, enable FSR, and often lower the resolution below Full HD.

Architecture and Main Difference from Radeon 680M

The Radeon 660M contains six RDNA 2 compute units, or 384 shader units, and operates at a frequency of up to 1.9 GHz. It is found in Ryzen 5 6600U, 6600H, 6600HS, 7535U, 7535HS, and some later AMD processors.

This graphics card has no dedicated video memory. It utilizes system memory, either DDR5 or LPDDR5, so memory speed and the number of active channels directly impact performance.

Radeon 680M is built on the same architecture but has 12 compute units. The difference between the models is not merely a slight reduction in frequency: the lesser 660M has half as many computational resources. Thus, the gap between them is noticeable not only in benchmarks but also in games. AMD specifies six graphics units for Radeon 660M and 12 for Radeon 680M.

Radeon 660M in Specific Devices

Device Processor Long-Term Limit and Memory 3DMark Time Spy Graphics 3DMark Fire Strike Graphics
Lenovo ThinkPad Z13 Gen 1 Ryzen 5 PRO 6650U 22.5 W, 16 GB LPDDR5-6400 1588 4487
Ninkear S14 Ryzen 5 7535HS 25 W, 16 GB 1535 4929
Chuwi CoreBook Air Ryzen 5 6600H 28 W, 16 GB LPDDR5-6400 1514 4891
Morefine M600 Ryzen 5 6600U 28 W, 32 GB DDR5-4800 1430 4621
HP EliteBook 845 G9 Ryzen 5 PRO 6650U 16 GB DDR5-4800 1149 3447

The table does not include laptops with discrete graphics cards, so the results pertain specifically to the Radeon 660M. The values are taken from reviews of specific configurations, not random user-run tests.

The best result in Time Spy was delivered by the ThinkPad Z13, scoring 1588 points. The Ninkear S14 and Chuwi CoreBook Air slightly lagged behind, despite using processors with higher power consumption. This indicates that after a certain level, additional power does little to accelerate the Radeon 660M: the six compute units remain its primary limitation.

A noteworthy comparison is between the ThinkPad Z13 and HP EliteBook 845 G9. Both laptops use the Ryzen 5 PRO 6650U, but Lenovo scores 38% higher in Time Spy Graphics and 30% higher in Fire Strike Graphics. The ThinkPad is equipped with LPDDR5-6400, while the EliteBook uses DDR5-4800. However, the entire gap cannot be explained solely by memory differences: there are variations in power settings, BIOS configurations, and cooling.

The Morefine M600’s result falls between the faster laptops and the EliteBook. The mini-PC maintains the Ryzen 5 6600U at 28 W and uses two DDR5-4800 modules, but this does not allow it to outperform models with faster LPDDR5.

What Radeon 660M Can Handle in Games

The optimal scenario for Radeon 660M is running less demanding games at Full HD. Dota 2 on the Chuwi CoreBook Air delivers around 58 frames per second at high settings and 49 frames at maximum preset. In Strange Brigade, you can achieve about 40 frames per second at medium settings in Full HD.

In demanding games, performance noticeably lacks. On the same Chuwi CoreBook Air, results without FSR look like this:

  • Cyberpunk 2077 - 21 frames per second at low settings in Full HD;
  • Cyberpunk 2077 - 15 frames at medium preset;
  • Baldur’s Gate 3 - around 22 frames at low settings in Full HD;
  • Final Fantasy XV - about 23 frames at standard Full HD quality.

Switching to a resolution of 1280 × 720 significantly improves the situation. In Dota 2, the result exceeds 100 frames per second, and Strange Brigade approaches this level on minimum settings. However, scaling from 720p on a modern 1920 × 1200 or 2560 × 1600 display noticeably degrades sharpness.

The capabilities of the Radeon 660M can generally be divided into three groups:

  • Competitive and less demanding games - Full HD, medium or high settings;
  • Older AAA games - Full HD at low preset or 900p;
  • Modern heavy games - low settings, FSR, and resolution below Full HD.

Hardware ray tracing is supported, but enabling it in games is pointless: standard rendering already uses up almost all available performance.

How Much Faster is Radeon 680M

The difference is clearly illustrated by two versions of the Morefine M600 mini-PC. The model with Radeon 680M scored 2408 points in Time Spy Graphics, compared to 1430 for the Radeon 660M-an advantage of 68%. In Fire Strike Graphics, the gap reached 44%: 6644 vs. 4621 points.

This is not a completely clean comparison of the GPUs: the version with Radeon 680M uses a more powerful Ryzen 9 6900HX. However, the identical form factor and similar memory configuration highlight the difference between classes well. The Radeon 680M allows for more consistent experiences at Full HD and less frequent drops to minimum settings.

Therefore, for a small price difference, a laptop with Radeon 680M is generally more advantageous for gaming. The additional cost is definitely worth it for graphics, rather than just extra CPU cores.

Conclusion

The Radeon 660M is suitable for a work laptop or mini-PC that runs games occasionally. It confidently handles Dota 2 and other less demanding projects, but in modern AAA games, it fails to deliver stable 30 frames per second at native Full HD.

When choosing a device, fast multi-channel memory and adequate power limits are essential. However, even a favorable configuration does not turn the Radeon 660M into a Radeon 680M: the six compute units remain the key limitation.

Buying an expensive laptop solely for the Radeon 660M is not advisable. If gaming performance matters, it’s better to look for a model with Radeon 680M, Radeon 780M, or a discrete graphics card.

Basic

Label Name
AMD
Platform
Integrated
Launch Date
January 2022
Model Name
Radeon 660M
Generation
Rembrandt
Base Clock
1500MHz
Boost Clock
1900MHz
Bus Interface
PCIe 4.0 x8
Transistors
13,100 million
RT Cores
6
Compute Units
6
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.
24
Foundry
TSMC
Process Size
6 nm
Architecture
RDNA 2.0

Memory Specifications

Memory Size
System Shared
Memory Type
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
Memory Clock
SystemShared
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

Outputs
No outputs

Theoretical Performance

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.
30.40 GPixel/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.
45.60 GTexel/s
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.
2.918 TFLOPS
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.
91.20 GFLOPS
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.
1.43 TFLOPS

Miscellaneous

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.
384
L1 Cache
128 KB per Array
L2 Cache
2MB
TDP
15W
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.
1.2
OpenCL Version
2.0
OpenGL
4.6
DirectX
12 Ultimate (12_2)
Power Connectors
None
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.
16
Shader Model
6.5

Benchmarks

FP32 (float)
Score
1.43 TFLOPS
3DMark Time Spy
Score
1526
Blender
Score
92

Compared to Other GPU

FP32 (float) / TFLOPS
1.561 +9.2%
1.498 +4.8%
1.396 -2.4%
1.371 -4.1%
3DMark Time Spy
4682 +206.8%
3619 +137.2%
2329 +52.6%
371.8 -75.6%
Blender
1408.56 +1431%
802 +771.7%
391 +325%
191.62 +108.3%