AMD Radeon HD 8950M vs NVIDIA GeForce GTX 1660 SUPER Comparison

# AMD Radeon HD 8950M vs NVIDIA GeForce GTX 1660 SUPER

The data presents a clear generational divide: the NVIDIA GeForce GTX 1660 SUPER dominates the AMD Radeon HD 8950M across every shared benchmark, yet both cards sit at nearly identical overall percentile rankings (54th vs 53rd percentile). This paradox — one card being dramatically faster while both occupy similar positions in the global GPU distribution — suggests the HD 8950M's ranking is buoyed by its mobile context, while the GTX 1660 SUPER's ranking reflects a far more demanding competitive landscape.

Where Each One Wins

The AMD Radeon HD 8950M does not win a single benchmark in the head-to-head comparison. The data shows zero wins for the AMD card and two wins for the NVIDIA card. However, the HD 8950M's average benchmark score of 13,376 is actually higher than the GTX 1660 SUPER's average of 12,986. This apparent contradiction stems from the different benchmark suites each card was tested with — the HD 8950M ran Geekbench OpenCL and Vulkan tests, while the GTX 1660 SUPER ran those same tests plus a broader array of 3DMark and Passmark workloads.

The GTX 1660 SUPER's wins are comprehensive in compute performance. In Geekbench OpenCL, it scores 52,490 versus the HD 8950M's 13,772 — a 73.8% lead. In Geekbench Vulkan, the NVIDIA card scores 57,102 versus 12,979, a 77.3% advantage. These are not marginal victories; they represent a 3.8x and 4.4x raw performance gap respectively.

Where the HD 8950M retains relevance is in its efficiency profile and form factor. As a 100W MXM module, it fits into mobile workstations from 2013. The GTX 1660 SUPER, by contrast, is a dual-slot desktop card with a 125W TDP, a 229mm length, and a 300W suggested PSU. The AMD card's portability advantage is its only clear win — the benchmark data offers nothing else.

Architecture Differences

The architectural gap between these two GPUs spans six years of silicon evolution. The AMD Radeon HD 8950M uses the Saturn chip built on GCN 2.0 architecture, fabricated by TSMC on a 28nm process. It packs 2,080 million transistors into a 160 mm² die, yielding a transistor density of 13.0M per mm². The NVIDIA GeForce GTX 1660 SUPER uses the TU116 chip with Turing architecture, also from TSMC but on a 12nm process. This newer node allows 6,600 million transistors in a 284 mm² die — a density of 23.2M per mm².

The compute resources differ massively. The HD 8950M has 768 shading units, 48 texture mapping units, and 16 ROPs. The GTX 1660 SUPER nearly doubles those counts: 1,408 shading units, 88 TMUs, and 48 ROPs. Pixel rate jumps from 17.20 GPixel/s to 85.68 GPixel/s, and texture rate from 51.60 GTexel/s to 157.1 GTexel/s. Floating-point performance tells the same story: 1.651 TFLOPS versus 5.027 TFLOPS, a 3x difference.

Memory subsystems are equally divergent. The HD 8950M uses 2GB of GDDR5 on a 128-bit bus, producing 88.00 GB/s of bandwidth. The GTX 1660 SUPER has 6GB of GDDR6 on a 192-bit bus, yielding 336.0 GB/s — nearly four times the bandwidth. The NVIDIA card also supports FP16 at 10.05 TFLOPS with a 2:1 ratio, while the HD 8950M has no listed FP16 capability.

Head-to-Head Benchmarks

The only direct comparisons available are two Geekbench tests, and both are decisive victories for the NVIDIA card. In Geekbench OpenCL, the GTX 1660 SUPER scores 52,490 against the HD 8950M's 13,772. This 73.8% delta means the NVIDIA card delivers roughly 3.8 times the compute throughput. The Vulkan test shows a similar picture: 57,102 versus 12,979, a 77.3% gap that represents a 4.4x performance advantage.

These results align perfectly with the architectural differences. The GTX 1660 SUPER's 1,408 shading units operating at 1,785 MHz boost clocks (versus the HD 8950M's 768 units with no listed boost clock) explain the OpenCL gap. The Vulkan advantage is even larger, which may reflect the Turing architecture's more mature Vulkan driver stack — the GTX 1660 SUPER supports Vulkan 1.4, while the HD 8950M is limited to Vulkan 1.2.170.

The broader benchmark suite for the GTX 1660 SUPER provides additional context. Its Passmark G3D score of 12,699 and GPU Compute score of 5,076 show solid DirectX performance across generations (DirectX 9: 189, DirectX 11: 104, DirectX 12: 50). The HD 8950M has no comparable DirectX benchmark data in this pack, so those numbers cannot be directly compared.

The Verdict

The data is unambiguous: the NVIDIA GeForce GTX 1660 SUPER is the superior performer by a wide margin in every measurable comparison. Anyone choosing between these two based purely on compute benchmarks should select the NVIDIA card without hesitation. The 73.8% OpenCL lead and 77.3% Vulkan lead are overwhelming, and the 6GB GDDR6 memory with 336.0 GB/s bandwidth dwarfs the 2GB GDDR5 with 88.00 GB/s.

However, the context matters. The HD 8950M is a mobile MXM module from 2013, designed for portable workstations. Its 100W TDP and compact form factor serve a completely different use case than the dual-slot, 229mm-long GTX 1660 SUPER. If the requirement is a legacy laptop upgrade, the HD 8950M remains the only option in this comparison. If the requirement is desktop performance, the GTX 1660 SUPER wins every category.

The percentile rankings add nuance. Both cards sit near the 54th percentile globally, but this means different things. The HD 8950M achieves its rank among 2013-era mobile GPUs; the GTX 1660 SUPER achieves a similar rank among 2019-era desktop GPUs. The latter is facing far stiffer competition, making its equal percentile more impressive. The GTX 1660 SUPER's nearest rivals include the RTX 3050 Ti Mobile (0.4% lower), RX 580 (0.4% lower), and Tesla M2090 (0.7% higher) — a tight cluster. The HD 8950M's rivals are all Radeon mobile or Pro cards within 0.4% of its score.

FAQ

Q: Which GPU has a higher average benchmark score?

A: The AMD Radeon HD 8950M has an average benchmark score of 13,376, while the NVIDIA GeForce GTX 1660 SUPER averages 12,986. However, the HD 8950M was tested on only two Geekbench workloads, while the GTX 1660 SUPER was tested on ten workloads including Passmark and 3DMark.

Q: How much faster is the GTX 1660 SUPER in Geekbench Vulkan?

A: The GTX 1660 SUPER scores 57,102 versus the HD 8950M's 12,979, a 77.3% advantage — meaning it delivers roughly 4.4 times the Vulkan performance.

Q: What is the memory bandwidth difference?

A: The GTX 1660 SUPER provides 336.0 GB/s from 6GB of GDDR6 on a 192-bit bus, while the HD 8950M offers 88.00 GB/s from 2GB of GDDR5 on a 128-bit bus — a 3.8x bandwidth advantage for NVIDIA.

Q: Do both cards support the same API levels?

A: No. The GTX 1660 SUPER supports DirectX 12 (12_1) and Vulkan 1.4, while the HD 8950M supports DirectX 12 (12_0) and Vulkan 1.2.170. Both support OpenGL 4.6.

Q: Which card has higher transistor density?

A: The GTX 1660 SUPER has 23.2M transistors per mm² from its 12nm process, versus the HD 8950M's 13.0M per mm² from its 28nm process. The NVIDIA chip also has more total transistors: 6,600 million versus 2,080 million.

Q: What form factors do these cards use?

A: The HD 8950M is an MXM Module with portable-device-dependent display outputs. The GTX 1660 SUPER is a dual-slot desktop card measuring 229mm in length, 111mm in height, and 35mm in width, with a DVI, HDMI 2.0, and DisplayPort 1.4a outputs.

Specification Differences

| Specification | AMD Radeon HD 8950M | NVIDIA GeForce GTX 1660 SUPER |

|---|---|---|

| Architecture | GCN 2.0 | Turing |

| Process Node | 28 nm | 12 nm |

| Transistors | 2,080 million | 6,600 million |

| Die Size | 160 mm² | 284 mm² |

| Transistor Density | 13.0M / mm² | 23.2M / mm² |

| Memory Size | 2 GB GDDR5 | 6 GB GDDR6 |

| Memory Bus | 128 bit | 192 bit |

| Memory Bandwidth | 88.00 GB/s | 336.0 GB/s |

| Shading Units | 768 | 1,408 |

| TMUs | 48 | 88 |

| ROPs | 16 | 48 |

| Pixel Rate | 17.20 GPixel/s | 85.68 GPixel/s |

| Texture Rate | 51.60 GTexel/s | 157.1 GTexel/s |

| FP32 Performance | 1.651 TFLOPS | 5.027 TFLOPS |

| FP16 Performance | Not listed | 10.05 TFLOPS (2:1) |

| TDP | 100 W | 125 W |

| Slot Width | MXM Module | Dual-slot |

| Power Connectors | Not listed | 1x 8-pin |

| Suggested PSU | Not listed | 300 W |

| DirectX Support | 12 (12_0) | 12 (12_1) |

| Vulkan Support | 1.2.170 | 1.4 |

| Release Date | 2013-05-13 | 2019-10-28 |

| Production Status | End-of-life | End-of-life |