AMD Radeon HD 6950 vs NVIDIA GeForce GTX 460 SE Comparison

The Verdict

The benchmark data tells a clear tale of two very different GPUs that nonetheless land in nearly the same performance bracket. The NVIDIA GeForce GTX 460 SE edges out the AMD Radeon HD 6950 in the sole OpenCL benchmark, scoring 6389 against 6210, a 2.9% margin. However, that narrow win comes from a card with drastically fewer resources. The GTX 460 SE is the pick for builders who want a compact, lower-power card that still delivers competitive compute scores. The HD 6950 is for those who need more memory and higher raw throughput in shading, texture, and pixel operations, accepting a larger physical footprint and greater power draw. If your workload is purely OpenCL compute, the GTX 460 SE wins on data. If you plan to push high resolutions with more texture detail, the HD 6950's specification sheet suggests it is the stronger candidate despite losing the benchmark.

Architecture Differences

The two cards come from opposing architectural philosophies. The GTX 460 SE uses the GF104 chip built on Fermi architecture, fabricated by TSMC on a 40 nm process. It packs 1,950 million transistors into a 332 mm² die, yielding a transistor density of 5.9M per mm². The HD 6950 uses the Cayman chip with TeraScale 3 architecture, also on TSMC's 40 nm node, but grows to 2,640 million transistors across a 389 mm² die, achieving 6.8M per mm². That is 35% more transistors than the GTX 460 SE, spread across a 17% larger die.

The shading unit counts are dramatically different. The GTX 460 SE carries 288 shading units, 48 texture mapping units, and 32 ROPs. The HD 6950 counters with 1,408 shading units, 88 TMUs, and the same 32 ROPs. That is nearly five times the shading units and nearly double the texture units on the AMD side. The HD 6950 also has higher memory clocks, running at 1250 MHz (5 Gbps effective) versus 850 MHz (3.4 Gbps effective) on the GTX 460 SE. Both use GDDR5 on a 256-bit bus, but the AMD card's higher clock yields 160.0 GB/s of bandwidth versus 108.8 GB/s on the NVIDIA card.

The HD 6950 doubles the memory capacity to 2 GB over the GTX 460 SE's 1024 MB. API support differs slightly: the GTX 460 SE lists DirectX 12 (11_0) and OpenGL 4.6, while the HD 6950 lists DirectX 11.2 (11_0) and OpenGL 4.4. Neither card has Vulkan support listed. The HD 6950 also offers a richer display output set with 2x DVI, 1x HDMI 1.4a, and 2x mini-DisplayPort 1.2, versus the GTX 460 SE's 2x DVI and 1x mini-HDMI 1.3a.

Where Each One Wins

The GTX 460 SE wins the only benchmark tested, the Geekbench OpenCL score, by 2.9%. It also wins on efficiency metrics per the specification sheet. Its 150 W TDP is 50 W lower than the HD 6950's 200 W, and its suggested PSU is 450 W versus 550 W. The GTX 460 SE is physically smaller at 210 mm (8.3 inches) in length, while the HD 6950 stretches to 286 mm (11.3 inches) with a height of 126 mm and width of 42 mm. Both are dual-slot cards with 2x 6-pin power connectors.

The HD 6950 wins on raw compute throughput. Its FP32 performance is 2.253 TFLOPS versus 748.8 GFLOPS on the GTX 460 SE, a threefold advantage. Pixel rate is 25.60 GPixel/s versus 7.800 GPixel/s, and texture rate is 70.40 GTexel/s versus 31.20 GTexel/s. The HD 6950 also has double the memory capacity and 47% more memory bandwidth. For any workload that scales with shading units, texture fill, or memory capacity, the HD 6950's specifications dominate. The GTX 460 SE's win in OpenCL suggests its Fermi architecture is better optimized for that specific compute workload despite the hardware deficit.

FAQ

Q: Which card has a higher OpenCL benchmark score?

A: The NVIDIA GeForce GTX 460 SE scores 6389, which is 2.9% higher than the AMD Radeon HD 6950's 6210 in the Geekbench OpenCL test.

Q: How much more memory does the HD 6950 have?

A: The HD 6950 has 2 GB of GDDR5 memory, exactly double the 1024 MB found on the GTX 460 SE.

Q: What is the transistor count difference between the two cards?

A: The HD 6950 has 2,640 million transistors, while the GTX 460 SE has 1,950 million, making the AMD card 35% denser in transistor count.

Q: Which card consumes less power?

A: The GTX 460 SE has a 150 W TDP and a 450 W suggested PSU, while the HD 6950 has a 200 W TDP and a 550 W suggested PSU.

Q: Do both cards use the same memory bus width?

A: Yes, both cards use a 256-bit memory bus, but the HD 6950's higher memory clock gives it 160.0 GB/s bandwidth versus 108.8 GB/s on the GTX 460 SE.

Q: What are the physical size differences?

A: The GTX 460 SE is 210 mm long, while the HD 6950 is 286 mm long, 126 mm tall, and 42 mm wide. Both are dual-slot cards.

Head-to-Head Benchmarks

The single available comparison is the Geekbench OpenCL test. The NVIDIA GeForce GTX 460 SE scores 6389, and the AMD Radeon HD 6950 scores 6210. The GTX 460 SE wins by 179 points, a 2.9% delta. This is a narrow margin, and the nearest rival data puts both cards in the same performance tier. The GTX 460 SE sits at the 37th percentile of all GPUs, with its closest rival being the NVIDIA GeForce GTX 580M at an identical 6389 score. The HD 6950 sits at the 36th percentile, with its nearest rival being the AMD FirePro W600 at 6223, a 0.2% difference.

What makes this result interesting is the hardware disparity. The HD 6950 has 1,408 shading units versus 288, and its FP32 throughput is 2.253 TFLOPS versus 748.8 GFLOPS. Yet the GTX 460 SE still wins the OpenCL test. That suggests the Fermi architecture's compute scheduling or driver optimization for OpenCL is more effective than the raw throughput numbers of TeraScale 3 would predict. The GTX 460 SE also matches exactly with the GTX 580M score of 6389, meaning the two NVIDIA cards perform identically in this test, while the HD 6950 trails its nearest rival by a hair.

The delta between the two cards is small enough that in real-world usage, other factors like memory capacity and power draw would likely decide the purchase. The HD 6950's 2 GB frame buffer and higher bandwidth make it better suited for high-resolution textures, while the GTX 460 SE's lower TDP and shorter length make it easier to fit in smaller cases with less demanding PSUs.

Specification Differences

The two cards diverge on nearly every internal specification except memory bus width and ROP count. Both use 32 ROPs and a 256-bit bus, but that is where the similarities end.

| Specification | GTX 460 SE | HD 6950 |

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

| Chip | GF104 | Cayman |

| Architecture | Fermi | TeraScale 3 |

| Process Node | 40 nm | 40 nm |

| Transistors | 1,950 million | 2,640 million |

| Die Size | 332 mm² | 389 mm² |

| Transistor Density | 5.9M / mm² | 6.8M / mm² |

| Memory Size | 1024 MB | 2 GB |

| Memory Clock | 850 MHz (3.4 Gbps) | 1250 MHz (5 Gbps) |

| Memory Bandwidth | 108.8 GB/s | 160.0 GB/s |

| Shading Units | 288 | 1,408 |

| TMUs | 48 | 88 |

| Pixel Rate | 7.800 GPixel/s | 25.60 GPixel/s |

| Texture Rate | 31.20 GTexel/s | 70.40 GTexel/s |

| FP32 | 748.8 GFLOPS | 2.253 TFLOPS |

| TDP | 150 W | 200 W |

| Suggested PSU | 450 W | 550 W |

| Length | 210 mm (8.3 in) | 286 mm (11.3 in) |

| Height | Not listed | 126 mm (5 in) |

| Width | Not listed | 42 mm (1.7 in) |

| Display Outputs | 2x DVI, 1x mini-HDMI 1.3a | 2x DVI, 1x HDMI 1.4a, 2x mini-DisplayPort 1.2 |

| DirectX | 12 (11_0) | 11.2 (11_0) |

| OpenGL | 4.6 | 4.4 |

| Launch MSRP | 160 USD | 299 USD |

The HD 6950 was released roughly one month after the GTX 460 SE, with the NVIDIA card launching on November 14, 2010, and the AMD card on December 13, 2010. Both are end-of-life products. The GTX 460 SE's predecessors is the GeForce 200 series, and its successor is the GeForce 500 series. The HD 6950's predecessor is Evergreen, and its successor is Southern Islands.