AMD Radeon 660M vs AMD Radeon R9 M380 Comparison

Head-to-Head Benchmarks

The recorded data shows a single direct benchmark comparison between these two AMD GPUs, and the result is a narrow victory for the AMD Radeon 660M. In the Geekbench OpenCL test, the Radeon 660M scored 12,876 points, while the AMD Radeon R9 M380 scored 12,565 points. The delta percentage of 2.4% in favor of the Radeon 660M indicates a close contest, not a decisive gap. This is a modest edge, and the overall benchmark averages tell a slightly different story about the broader positioning of each part.

Looking at the aggregate average benchmark score, the Radeon R9 M380 actually holds a higher standing. Its average score is 15,521, placing it in the 58th percentile of all GPUs in the database. The Radeon 660M averages 13,812, which puts it in the 55th percentile. That is a difference of roughly 1,709 points in the R9 M380's favor, or about 12.4% higher than the 660M's average. This discrepancy between the single head-to-head OpenCL result and the overall averages suggests that the R9 M380 benefits from additional benchmark results beyond the one direct comparison, particularly its strong Geekbench Metal score of 18,476.

The Radeon 660M counters with a Geekbench Vulkan score of 14,748, a test that the R9 M380 does not have a recorded result for. This matters because Vulkan performance reflects a modern API path that the older GCN 2.0 architecture may not handle as efficiently. The R9 M380's only other benchmark, Geekbench OpenCL, is its weaker showing at 12,565, which is below its own average. In contrast, the 660M's OpenCL score of 12,876 is also below its average, but its Vulkan result pushes its overall profile higher.

The nearest rivals for each GPU further illustrate their performance neighborhoods. The R9 M380 sits within 1.6% of the NVIDIA GeForce GTX 580 (average score 15,283) and within 1.5% of the NVIDIA GeForce RTX 2060 (average score 15,290). It is essentially tied with the NVIDIA GeForce GTX 1080 Ti, which averages 15,548, a delta of only 0.2%. The Radeon 660M, by contrast, is nearly identical to the NVIDIA RTX A2000 Mobile, which averages 13,821, a delta of 0.1%, and trails the AMD Radeon RX 570X by only 0.4% (13,871). These rival comparisons show that the R9 M380, despite being older, competes in a higher performance bracket than the 660M when considering all available benchmark data.

FAQ

Q: Which GPU wins the direct OpenCL benchmark?

A: The AMD Radeon 660M wins the only head-to-head test, the Geekbench OpenCL run, with a score of 12,876 versus 12,565 for the AMD Radeon R9 M380. The margin is 2.4% in favor of the 660M.

Q: How do their overall average scores compare?

A: The Radeon R9 M380 has a higher average benchmark score of 15,521, compared to the Radeon 660M's 13,812. This places the R9 M380 in the 58th percentile of all GPUs, while the 660M sits in the 55th percentile.

Q: What benchmark results are available for each GPU?

A: The R9 M380 has recorded results in Geekbench Metal (18,476) and Geekbench OpenCL (12,565). The Radeon 660M has recorded results in Geekbench OpenCL (12,876) and Geekbench Vulkan (14,748). There is no shared test where both have a common benchmark other than OpenCL.

Q: How close are these GPUs to their nearest rivals?

A: The R9 M380 is within 0.2% of the NVIDIA GeForce GTX 1080 Ti (15,548) and 1.5% of the NVIDIA GeForce RTX 2060 (15,290). The 660M is within 0.1% of the NVIDIA RTX A2000 Mobile (13,821) and 0.4% of the AMD Radeon RX 570X (13,871).

Q: Which GPU has a higher pixel rate?

A: The Radeon 660M has a pixel rate of 30.40 GPixel/s, which is nearly double the R9 M380's 16.00 GPixel/s. This suggests the 660M has an advantage in fill-rate-bound scenarios.

Q: What is the production status of each GPU?

A: Both the AMD Radeon R9 M380 and the AMD Radeon 660M are listed as end-of-life products in the database.

The Verdict

The data supports a split decision depending on the workload. For raw compute throughput as measured by the direct comparison, the AMD Radeon 660M takes the edge in OpenCL. Its 2.4% lead over the R9 M380 in that specific test is real but small. However, the R9 M380's much higher average benchmark score of 15,521 versus 13,812 indicates that, across a broader set of metrics, the older GPU is the more capable part. The R9 M380 also delivers a strong Geekbench Metal result of 18,476, which is its best recorded score, suggesting particular strength in Metal-based applications. The 660M, meanwhile, shows a more balanced profile with Vulkan support and a higher pixel rate.

For users prioritizing modern API compatibility, the Radeon 660M is the clear choice. It supports DirectX 12 Ultimate (12_2) and Vulkan 1.4, whereas the R9 M380 is limited to DirectX 12 (12_0) and Vulkan 1.2.170. The 660M also has hardware ray tracing cores, six of them, which the R9 M380 lacks entirely. For legacy performance in Metal and OpenCL, the R9 M380 holds the higher average score and a better percentile ranking. The verdict is that the R9 M380 is the stronger overall GPU in the database's aggregate metrics, but the 660M is more future-proof and wins the one direct benchmark recorded.

Specification Differences

The two GPUs differ in nearly every core specification category. The R9 M380 uses a 28 nm process node, while the Radeon 660M uses a 6 nm node. The R9 M380 is built on the Strato chip with GCN 2.0 architecture, whereas the 660M uses the Rembrandt chip with RDNA 2.0 architecture. Clock speeds diverge significantly: the R9 M380 has a base clock of 900 MHz and a boost of 1000 MHz, while the 660M operates at 1500 MHz base and 1900 MHz boost. Memory configuration is another major split, with the R9 M380 featuring 4 GB of GDDR5 on a 128-bit bus delivering 96.00 GB/s bandwidth, while the 660M uses system shared memory with system dependent bandwidth.

The compute unit counts show a trade-off. The R9 M380 has 768 shading units, 48 texture mapping units, and 16 ROPs. The 660M has 384 shading units, 24 TMUs, and 16 ROPs, so the R9 M380 has double the shader and TMU count. Despite this, the 660M achieves a higher pixel rate of 30.40 GPixel/s versus 16.00 GPixel/s, and a comparable texture rate of 45.60 GTexel/s versus 48.00 GTexel/s. The FP32 performance is nearly identical, with the R9 M380 at 1.536 TFLOPS and the 660M at 1,459.2 GFLOPS, a difference of about 5%. The 660M also supports FP16 at 2.918 TFLOPS with a 2:1 ratio, while the R9 M380 has no recorded FP16 capability. The 660M includes 6 ray tracing cores, and the R9 M380 has none. Power characteristics differ as well, with the 660M rated at 40 W TDP and no power connectors, while the R9 M380 has no recorded TDP. The bus interface also changes from PCIe 3.0 x16 on the R9 M380 to PCIe 4.0 x8 on the 660M.

Architecture Differences

The architectural gap between these two GPUs is generational. The R9 M380 is based on GCN 2.0, AMD's Graphics Core Next architecture from the Gem System generation, specifically the R9 M300 lineup, released in 2015. It uses the Strato chip fabricated on a 28 nm process at TSMC, with 2,080 million transistors on a 160 mm² die. The Radeon 660M, by contrast, uses RDNA 2.0, AMD's newer graphics architecture, from the Navi II IGP generation, Rembrandt Mobile, released in 2022. Its Rembrandt chip is built on a 6 nm process at TSMC, packing 13,100 million transistors into a 208 mm² die. The transistor density tells the story of the process improvement: the R9 M380 has 13.0 million transistors per mm², while the 660M has 63.0 million per mm², roughly 4.8 times denser.

Cache and feature differences are not fully recorded, but the API support reveals the architectural evolution. The R9 M380 supports DirectX 12 (12_0), OpenGL 4.6, and Vulkan 1.2.170. The 660M supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The DirectX 12 Ultimate designation includes features like ray tracing and variable rate shading, which the older GCN 2.0 architecture cannot provide. The 660M's 6 ray tracing cores are a hardware confirmation of this capability. The R9 M380 has no ray tracing cores and no tensor cores, and the 660M also has no tensor cores, but the ray tracing support alone marks a distinct architectural advantage for the newer part. The R9 M380's predecessor is noted as Solar System, with a successor of Polaris Mobile, while the 660M's predecessor is Vega II IGP and its successor is Navi III IGP, illustrating two different product lineage tracks.

Where Each One Wins

The AMD Radeon R9 M380 wins in scenarios where average benchmark scores and Metal API performance matter most. Its average score of 15,521 places it in the 58th percentile, which is three points higher than the 660M's 55th percentile. The R9 M380's Geekbench Metal score of 18,476 is its standout result, indicating strong performance in Metal-based workloads, likely relevant for macOS or iOS development contexts. Its higher shading unit count of 768 versus 384 also suggests an advantage in compute-heavy tasks that scale with shader count, even if the 660M's newer architecture achieves similar FP32 output. The R9 M380 also has higher FP32 throughput at 1.536 TFLOPS compared to the 660M's 1,459.2 GFLOPS, a small but measurable win.

The AMD Radeon 660M wins in the direct OpenCL comparison, scoring 12,876 versus 12,565, and it also wins on architectural modernity. Its pixel rate of 30.40 GPixel/s is 90% higher than the R9 M380's 16.00 GPixel/s, which benefits rasterization-bound workloads. The 660M's Vulkan score of 14,748 demonstrates strong performance in that API, which the R9 M380 has no recorded equivalent for. The 660M's higher clock speeds, 1500 MHz base and 1900 MHz boost, contribute to its responsiveness. Its 40 W TDP and lack of power connectors make it suitable for integrated graphics scenarios, while the R9 M380 has no power consumption data recorded. The 660M's support for DirectX 12 Ultimate and Vulkan 1.4, along with 6 ray tracing cores, makes it the better choice for modern game features and future software. In summary, the R9 M380 wins on aggregate performance and Metal, while the 660M wins on the direct benchmark, modern API support, and fill-rate efficiency.