AMD Radeon RX 550 vs NVIDIA GeForce GTX 660 Comparison

AMD
RADEON

AMD Radeon RX 550

CORE STATE Lexa
VRAM 2 GB
CLOCK SPEED 1183 MHz
TDP 50 W
BUS WIDTH 128 bit
ARCHITECTURE GCN 4.0
nm
PROCESS 14 nm
LAUNCH DATE 2017
VS
NVIDIA
GEFORCE

GeForce GTX 660

CORE STATE GK106
VRAM 2 GB
CLOCK SPEED 1032 MHz
TDP 140 W
BUS WIDTH 192 bit
ARCHITECTURE Kepler
nm
PROCESS 28 nm
LAUNCH DATE 2012

PERFORMANCE BENCHMARKS

3dmark_3dmark_steel_nomad_dx12
127
N/A
geekbench_metal
20,838
4,305
geekbench_opencl
11,063
11,347
geekbench_vulkan
12,270
11,415

Analysis: AMD Radeon RX 550 vs NVIDIA GeForce GTX 660

FAQ

Q: Which GPU has the higher average benchmark score?

A: The AMD Radeon RX 550 records an average benchmark score of 11,075, placing it in the 50th percentile of all GPUs. The NVIDIA GeForce GTX 660 averages 9,022, which puts it in the 45th percentile.

Q: How does the RX 550 compare to its nearest rivals in average score?

A: The RX 550 is effectively tied with the NVIDIA RTX PRO 6000D Blackwell Max-Q and RTX PRO 6000 Blackwell Max-Q, both at 11,088 (a delta of -0.1% against the RX 550). It edges out the GTX 1650 SUPER (11,047) by 0.3% and trails the AMD FirePro W4300 (11,225) by 1.3%.

Q: What is the most significant benchmark margin between these two cards?

A: In the Geekbench Metal test, the RX 550 scores 20,838 compared to the GTX 660's 4,305, a 384% advantage for the AMD card. This is the largest single margin recorded in the comparison.

Q: Does the GTX 660 win any benchmark outright?

A: Yes, in the Geekbench OpenCL test, the GTX 660 scores 11,347 versus the RX 550's 11,063, a 2.5% lead for the NVIDIA card.

Q: What are the launch MSRP values for these two cards?

A: The AMD Radeon RX 550 had a launch MSRP of 79 USD. The NVIDIA GeForce GTX 660 had a launch MSRP of 229 USD.

Q: What is the process node difference between the two chips?

A: The RX 550 is built on a 14 nm process at GlobalFoundries, while the GTX 660 uses a 28 nm process at TSMC. The RX 550 has a transistor density of 21.4M per mm², versus 11.5M per mm² for the GTX 660.

Where Each One Wins

The benchmark results create a clear split: the AMD Radeon RX 550 dominates in the newer graphics API workloads, while the NVIDIA GeForce GTX 660 holds a narrow lead in one compute-oriented test. The data shows a 2-1 win tally in favor of the RX 550, but the margins tell a more nuanced story.

The RX 550's decisive victory comes in Geekbench Metal, where it scores 20,838 against the GTX 660's 4,305. That 384% advantage suggests the AMD card is substantially stronger in Apple's Metal graphics framework, likely due to its more modern architecture and driver optimizations. This is the kind of result that matters for users running Metal-based applications or macOS-compatible workloads.

The AMD card also wins the Geekbench Vulkan test, scoring 12,270 versus the GTX 660's 11,415, a 7.5% margin. Vulkan is a cross-platform graphics API, so this win indicates the RX 550 has better modern API performance for current and upcoming games.

The GTX 660's sole victory is in Geekbench OpenCL, where it manages 11,347 against 11,063, a 2.5% edge. OpenCL is a general-purpose compute API, and the NVIDIA card's higher raw shader count (960 vs 512) likely contributes to this result. However, the margin is small enough that it is within typical run-to-run variance for benchmark suites.

For practical use cases, the RX 550 is the better choice for users prioritizing modern graphics APIs like Vulkan and Metal. The GTX 660 retains a slight edge in OpenCL compute, but its much lower average score and older feature set make it less compelling overall.

Architecture Differences

The architectural gap between these two GPUs spans four years and two major fabrication nodes. The AMD Radeon RX 550 uses the Lexa chip based on GCN 4.0 architecture, part of the Polaris (RX 500) generation. It is fabricated on a 14 nm process at GlobalFoundries with 2,200 million transistors on a 103 mm² die. The NVIDIA GeForce GTX 660 uses the GK106 chip based on Kepler architecture from the GeForce 600 generation, built on TSMC's 28 nm process with 2,540 million transistors on a 221 mm² die.

The transistor density difference is stark: the RX 550 packs 21.4 million transistors per mm², while the GTX 660 manages only 11.5 million per mm². This means the AMD chip achieves higher complexity in a much smaller physical area, which contributes to its dramatically lower power consumption.

In terms of compute resources, the GTX 660 has more raw hardware: 960 shading units, 80 texture mapping units, and 24 render output units. The RX 550 has 512 shading units, 32 TMUs, and 16 ROPs. Despite this 2x disadvantage in shading units, the RX 550 achieves comparable or better performance in most recorded benchmarks, indicating that its GCN 4.0 architecture extracts more efficiency per unit of silicon.

Memory subsystem design also differs. The RX 550 uses a 128-bit bus with 2 GB GDDR5 memory at 112.0 GB/s bandwidth. The GTX 660 uses a wider 192-bit bus with 2 GB GDDR5 at 144.2 GB/s, giving it 28.7% more bandwidth. However, the RX 550's faster effective memory clock (7 Gbps vs 6 Gbps) partially compensates for the narrower bus.

The API support reflects their respective eras. The RX 550 supports DirectX 12 (12_0), OpenGL 4.6, and Vulkan 1.3. The GTX 660 supports DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.2.175. The RX 550's Vulkan version is newer, which likely explains its 7.5% Vulkan benchmark win.

Specification Differences

The two cards differ across nearly every physical and electrical specification. The AMD Radeon RX 550 has a base clock of 1100 MHz and boost clock of 1183 MHz. The NVIDIA GeForce GTX 660 has a base clock of 980 MHz and boost of 1032 MHz. The RX 550 runs at higher frequencies, though it has fewer execution units.

Memory configurations differ: both have 2 GB of GDDR5, but the RX 550 uses a 128-bit bus with 112.0 GB/s bandwidth, while the GTX 660 uses a 192-bit bus with 144.2 GB/s bandwidth. The RX 550's memory runs at 1750 MHz (7 Gbps effective), while the GTX 660's runs at 1502 MHz (6 Gbps effective).

Power requirements show a major divergence. The RX 550 has a TDP of 50 W and requires no power connectors, with a suggested PSU of 250 W. The GTX 660 has a TDP of 140 W, requires a single 6-pin power connector, and suggests a 300 W PSU. This makes the RX 550 significantly easier to integrate into low-power systems.

Physical dimensions also differ: the RX 550 measures 145 mm (5.7 inches) in length, while the GTX 660 measures 241 mm (9.5 inches). Both are dual-slot cards. The RX 550 uses PCIe 3.0 x8, while the GTX 660 uses PCIe 3.0 x16.

Display outputs vary: the RX 550 offers 1x DVI, 1x HDMI 2.0b, and 1x DisplayPort 1.4a. The GTX 660 offers 2x DVI, 1x HDMI 1.4a, and 1x DisplayPort 1.2. The RX 550 supports newer display standards.

Compute throughput numbers favor the GTX 660: it delivers 1.981 TFLOPS FP32 and rates of 20.64 GPixel/s and 82.56 GTexel/s. The RX 550 delivers 1,211.4 GFLOPS FP32 (also 1,211.4 GFLOPS FP16 at 1:1 ratio) and rates of 18.93 GPixel/s and 37.86 GTexel/s. The GTX 660 has higher peak throughput, but the RX 550's architecture achieves better real-world results in most benchmarks.

Head-to-Head Benchmarks

The direct comparison between these two GPUs yields three recorded benchmark results, with the AMD Radeon RX 550 winning two and the NVIDIA GeForce GTX 660 winning one.

The Geekbench Metal test shows the most dramatic difference. The RX 550 posts 20,838 points against the GTX 660's 4,305 points, a 384% advantage. This result is remarkable given that the GTX 660 has nearly twice the shading units and 28.7% more memory bandwidth. The RX 550's GCN 4.0 architecture and Metal driver maturity clearly give it a decisive edge in this workload.

In Geekbench Vulkan, the RX 550 again leads, scoring 12,270 versus 11,415 for the GTX 660, a 7.5% margin. While not as lopsided as the Metal result, this still demonstrates the AMD card's superior performance in modern cross-platform graphics APIs. The RX 550's Vulkan 1.3 support versus the GTX 660's Vulkan 1.2.175 likely plays a role here.

The GTX 660's only win comes in Geekbench OpenCL, where it scores 11,347 against the RX 550's 11,063, a 2.5% advantage. The NVIDIA card's higher raw throughput (1.981 TFLOPS vs 1,211.4 GFLOPS FP32) and wider memory bus (192-bit vs 128-bit) contribute to this result. However, the margin is modest, and the RX 550's 14 nm efficiency helps it stay competitive despite the spec disadvantage.

The average benchmark scores reinforce the RX 550's overall superiority: 11,075 versus 9,022, a difference of 22.8% in favor of the AMD card. The percentile rankings also favor the RX 550 (50th vs 45th percentile of all GPUs). The RX 550's nearest rivals include the RTX PRO 6000D Blackwell Max-Q and RTX PRO 6000 Blackwell Max-Q (both 11,088, -0.1%), while the GTX 660's nearest rivals include the TITAN V CEO Edition (9,037, -0.2%) and GeForce GTX 560 (9,058, -0.4%). This places the RX 550 in a higher performance tier overall, despite the GTX 660's advantage in raw compute resources.

DETAILED SPECIFICATIONS

SPECIFICATION
RX 550
GTX 660
Core Specs
Shading Units
512
960 +87.5%
Shaders
512
960 +87.5%
TMUs
32
80 +150.0%
ROPs
16
24 +50.0%
Compute Units
8
Clocks
Base Clock
1100 MHz
980 MHz
Boost Clock
1183 MHz
1032 MHz
Memory Clock
1750 MHz 7 Gbps effective
1502 MHz 6 Gbps effective
Memory
Memory Size
2 GB
2 GB
VRAM (MB)
2,048
2,048 0.0%
Memory Type
GDDR5
GDDR5
Memory Bus
128 bit
192 bit
Bandwidth
112.0 GB/s
144.2 GB/s
Cache
L1 Cache
16 KB (per CU)
16 KB (per SMX)
L2 Cache
512 KB
384 KB
Performance
Pixel Rate
18.93 GPixel/s
20.64 GPixel/s
Texture Rate
37.86 GTexel/s
82.56 GTexel/s
FP32 (TFLOPS)
1,211.4 GFLOPS
1.981 TFLOPS
FP64 (TFLOPS)
75.71 GFLOPS (1:16)
82.56 GFLOPS (1:24)
FP16 (TFLOPS)
1,211.4 GFLOPS (1:1)
Power
TDP
50 W
140 W
TDP (W)
50
140 +180.0%
Suggested PSU
250 W
300 W
Power Connectors
None
1x 6-pin
Architecture
Architecture
GCN 4.0
Kepler
GPU Name
Lexa
GK106
Generation
Polaris (RX 500)
GeForce 600
Process Size
14 nm
28 nm
Transistors
2,200 million
2,540 million
Die Size
103 mm²
221 mm²
Foundry
GlobalFoundries
TSMC
Density
21.4M / mm²
11.5M / mm²
API Support
DirectX
12 (12_0)
12 (11_0)
OpenGL
4.6
4.6
Vulkan
1.3
1.2.175
OpenCL
2.1
3.0
CUDA
3.0
Shader Model
6.7
6.5 (5.1)
Physical
Slot Width
Dual-slot
Dual-slot
Length
145 mm 5.7 inches
241 mm 9.5 inches
Outputs
1x DVI1x HDMI 2.0b1x DisplayPort 1.4a
2x DVI1x HDMI 1.4a1x DisplayPort 1.2
Bus Interface
PCIe 3.0 x8
PCIe 3.0 x16
Other
Launch Price
79 USD
229 USD
Production
End-of-life
End-of-life
Predecessor
Arctic Islands
GeForce 500
Successor
Vega
GeForce 700
View Radeon RX 550 Details View GeForce GTX 660 Details