AMD Radeon 660M vs AMD Radeon RX 580 Comparison

The benchmark data is unambiguous: the AMD Radeon RX 580 is the decisive victor over the AMD Radeon 660M in every head-to-head comparison available. The RX 580 delivers roughly triple the raw compute performance of the integrated Radeon 660M in both OpenCL and Vulkan workloads, making it the far more capable option for demanding graphics tasks, despite its older architecture and higher power draw.

Head-to-Head Benchmarks

The RX 580 dominates the Radeon 660M in the two shared benchmark tests. In Geekbench OpenCL, the RX 580 scores 37,453 against the 660M’s 12,876, a delta of -65.6% from the 660M’s perspective. In Geekbench Vulkan, the gap is even wider: the RX 580 posts 45,173 while the 660M manages only 14,748, a -67.4% difference. The RX 580 wins both head-to-head matchups, with a clean 2-0 record.

Looking at the broader average benchmark scores, the RX 580’s advantage persists but narrows. The RX 580 holds an average benchmark score of 12,928, while the 660M averages 13,812. That places the 660M slightly ahead of the RX 580 in overall average, a counterintuitive result given the head-to-head losses. The explanation lies in the benchmark mix: the 660M’s two scores (OpenCL and Vulkan) are both strong for its class, while the RX 580’s average is pulled down by weaker results in some tests like Passmark DirectX 9 (124) and Passmark G2D (769). The RX 580’s Passmark G3D score of 8,813 and Geekbench Metal score of 45,235 show its peak strengths, but its average is diluted by those lower-scoring legacy tests.

Percentile rankings tell a similar story of near-parity. The 660M sits at the 55th percentile of all GPUs, while the RX 580 ranks at the 53rd percentile — a mere two-point gap. The 660M’s nearest rival is the NVIDIA RTX A2000 Mobile, with a delta of -0.1%, while the RX 580’s closest competitor is the NVIDIA GeForce RTX 3050 Ti Mobile, also at -0.1%. Both cards are firmly in the mid-range tier, but the RX 580’s raw performance ceiling is far higher.

Where Each One Wins

The RX 580 wins decisively in raw compute throughput. Its FP32 performance of 6.175 TFLOPS is more than four times the 660M’s 1,459.2 GFLOPS. Texture rate tells a similar story: the RX 580 delivers 193.0 GTexel/s versus the 660M’s 45.60 GTexel/s. Pixel rate favors the RX 580 as well, at 42.88 GPixel/s compared to 30.40 GPixel/s. For gaming, rendering, or any GPU-accelerated compute task that scales with shading units and memory bandwidth, the RX 580 is the clear choice.

The 660M wins in efficiency and integration. Its 40 W TDP is dramatically lower than the RX 580’s 185 W, and it requires no power connectors, drawing power from the system. The 660M is an IGP (integrated graphics processor) built into the Rembrandt mobile chip, meaning it takes up no expansion slot and produces no additional cooling burden. For thin-and-light laptops or compact systems where discrete graphics are impractical, the 660M is the only viable option. It also supports DirectX 12 Ultimate (12_2) and Vulkan 1.4, whereas the RX 580 is limited to DirectX 12 (12_0) and Vulkan 1.3.

The RX 580 also wins on memory, but with a caveat. It has a dedicated 8 GB of GDDR5 on a 256-bit bus, delivering 256.0 GB/s of bandwidth. The 660M uses system-shared memory with bandwidth described as "System Dependent." For workloads that hammer memory bandwidth, the RX 580’s dedicated VRAM is a major advantage. The 660M’s shared memory is flexible but slower in practice.

Architecture Differences

The architectural gap between these two GPUs is generational. The Radeon 660M is built on RDNA 2.0 architecture, fabricated on TSMC’s 6 nm process node. The RX 580 uses the older GCN 4.0 architecture on GlobalFoundries’ 14 nm node. This process advantage is significant: the 660M packs 13,100 million transistors into a 208 mm² die, yielding a transistor density of 63.0M per mm². The RX 580 has only 5,700 million transistors on a larger 232 mm² die, giving it a density of just 24.6M per mm².

The 660M’s RDNA 2.0 architecture brings modern features that the GCN 4.0 RX 580 lacks entirely. Most notably, the 660M includes 6 dedicated ray tracing cores, while the RX 580 has none. The 660M also supports DirectX 12 Ultimate (12_2), which includes hardware ray tracing and other advanced features; the RX 580 only reaches DirectX 12 (12_0). The 660M’s FP16 throughput is 2.918 TFLOPS (2:1 ratio), while the RX 580’s FP16 is identical to its FP32 at 6.175 TFLOPS (1:1 ratio) — an architectural difference in how each card handles half-precision math.

Shader resources heavily favor the RX 580. The RX 580 has 2,304 shading units, 144 texture mapping units, and 32 ROPs. The 660M has just 384 shading units, 24 TMUs, and 16 ROPs. This 6:1 ratio in shading units explains the RX 580’s massive compute advantage, despite the 660M’s newer architecture. The RX 580’s GCN design emphasized raw throughput with many simple cores; RDNA 2.0 focuses on efficiency and features per core.

Specification Differences

The most striking specification difference is power consumption. The 660M draws just 40 W, while the RX 580 is rated at 185 W TDP. The RX 580 requires a 450 W suggested PSU and a single 8-pin power connector; the 660M needs no external power. Physical form factors reflect this: the 660M is an IGP with no slot width, while the RX 580 is a dual-slot card measuring 241 mm (9.5 inches) in length.

Memory specifications diverge completely. The RX 580 has 8 GB of GDDR5 with a 256-bit bus and 256.0 GB/s bandwidth. The 660M uses system-shared memory with a system-shared bus width and system-dependent bandwidth. Clock speeds are interesting: the 660M boosts to 1900 MHz, while the RX 580 boosts to only 1340 MHz. The 660M’s higher clocks partially compensate for its fewer cores, but not enough to close the performance gap.

Bus interface and display outputs also differ. The 660M uses PCIe 4.0 x8, while the RX 580 uses PCIe 3.0 x16. The RX 580 offers fixed display outputs: 1x HDMI 2.0b and 3x DisplayPort 1.4a. The 660M’s display outputs are described as "Portable Device Dependent," reflecting its mobile IGP nature. The RX 580 has a launch MSRP of 229 USD. Release dates are far apart: the RX 580 launched on 2017-04-17, while the 660M arrived on 2022-01-03. The RX 580’s predecessor is Arctic Islands and its successor is Vega; the 660M’s predecessor is Vega II IGP and its successor is Navi III IGP.

FAQ

Q: Which GPU is faster in raw compute performance?

A: The AMD Radeon RX 580 is dramatically faster. It delivers 6.175 TFLOPS of FP32 performance versus the Radeon 660M’s 1,459.2 GFLOPS, and it wins both head-to-head benchmarks: 37,453 to 12,876 in Geekbench OpenCL and 45,173 to 14,748 in Geekbench Vulkan.

Q: Does the Radeon 660M support ray tracing?

A: Yes. The Radeon 660M includes 6 dedicated ray tracing cores and supports DirectX 12 Ultimate (12_2), which enables hardware-accelerated ray tracing. The RX 580 has no ray tracing cores and only supports DirectX 12 (12_0).

Q: How do their power requirements compare?

A: The Radeon 660M consumes only 40 W and requires no power connectors. The RX 580 draws 185 W TDP, needs a single 8-pin power connector, and requires a 450 W suggested PSU.

Q: Which GPU has more memory bandwidth?

A: The RX 580 has far more. It features 8 GB of GDDR5 on a 256-bit bus delivering 256.0 GB/s. The 660M uses system-shared memory with system-dependent bandwidth, which is typically much slower.

Q: Which GPU is better for a compact or mobile system?

A: The Radeon 660M is the clear choice for compact systems. It is an integrated GPU with no slot width, no power connectors, and a 40 W TDP. The RX 580 is a 241 mm dual-slot card requiring external power and substantial cooling.

Q: What are the average benchmark scores for each?

A: The Radeon 660M has an average benchmark score of 13,812, placing it at the 55th percentile of all GPUs. The RX 580 averages 12,928, at the 53rd percentile.

The Verdict

The data supports a clear split decision based on use case. For anyone building or upgrading a desktop system with space and power budget for a discrete card, the AMD Radeon RX 580 is the superior performer. It wins every head-to-head benchmark by a wide margin, offers dedicated 8 GB GDDR5 memory with 256.0 GB/s bandwidth, and delivers over four times the FP32 compute of the 660M. Its GCN 4.0 architecture is older and lacks ray tracing, but its raw throughput in traditional rasterization and compute workloads is unmatched by the 660M.

The AMD Radeon 660M is the right choice for ultraportable laptops or compact systems where a discrete GPU is impossible. Its 40 W TDP, integrated form factor, and system-shared memory make it the only practical option in those scenarios. It also offers modern features the RX 580 cannot match, including ray tracing cores, DirectX 12 Ultimate support, and a smaller 6 nm process. Its higher average benchmark score (13,812 vs 12,928) and better percentile ranking (55th vs 53rd) reflect its efficiency and feature set, but those numbers are skewed by the RX 580’s weaker legacy test results.

The verdict is straightforward: if you need performance, choose the RX 580. If you need integration and efficiency, choose the 660M. There is no scenario where both are equally suited, and the benchmark data provides no reason to compromise.