AMD Radeon 740M vs NVIDIA GeForce GTX 950 Comparison

The NVIDIA GeForce GTX 950 and AMD Radeon 740M represent two very different approaches to graphics hardware: a 2015 discrete desktop card built on a mature 28 nm process, and a 2024 integrated GPU (IGP) built on a advanced 4 nm node. While both land in the 53rd percentile of all GPUs, their benchmark profiles reveal a nuanced picture—the GTX 950 leads in raw Compute scores, but the Radeon 740M closes the gap significantly in Vulkan, suggesting architectural advantages that matter depending on the workload.

FAQ

Q: Which GPU has the higher average benchmark score?

A: The NVIDIA GeForce GTX 950 posts an average benchmark score of 13,189, while the AMD Radeon 740M scores 12,870. That puts the GTX 950 roughly 2.5% ahead in aggregate performance, though both sit at the 53rd percentile of all GPUs.

Q: How do the two compare in Geekbench OpenCL?

A: The GTX 950 wins decisively with a score of 15,662 versus 10,172 for the Radeon 740M, a 54% advantage. This is the largest single-benchmark gap between the two.

Q: Is the Radeon 740M competitive in Vulkan?

A: Yes. In Geekbench Vulkan, the GTX 950 scores 16,734 and the Radeon 740M scores 15,568, a margin of just 7.5%. Given the Radeon’s much lower OpenCL score, this near-parity in Vulkan is notable.

Q: What are the process node and transistor density differences?

A: The GTX 950 uses a 28 nm TSMC process with 2,940 million transistors on a 228 mm² die (12.9M transistors per mm²). The Radeon 740M uses a 4 nm TSMC process with 20,900 million transistors on a 137 mm² die (152.6M per mm²)—a dramatic density advantage.

Q: Which GPU supports ray tracing?

A: Only the AMD Radeon 740M has dedicated ray tracing cores (4). The GTX 950 has none, as Maxwell 2.0 predates hardware ray tracing.

Q: What is the TDP and power connector situation?

A: The GTX 950 has a 90 W TDP and requires a 1x 6-pin power connector, with a suggested PSU of 250 W. The Radeon 740M is an IGP with a 45 W TDP and needs no power connectors, drawing power from the motherboard.

Architecture Differences

The architectural gulf between these two GPUs is vast, spanning nearly a decade of silicon evolution. The GTX 950 is built on NVIDIA’s Maxwell 2.0 architecture, using the GM206 chip fabricated on TSMC’s 28 nm process. It packs 2,940 million transistors into a 228 mm² die, yielding a transistor density of 12.9M per mm². The Radeon 740M, by contrast, uses AMD’s RDNA 3.0 architecture on the Phoenix2 chip, built on a 4 nm process. It integrates 20,900 million transistors into a much smaller 137 mm² die, achieving a density of 152.6M per mm²—more than ten times denser.

The compute layouts differ sharply. The GTX 950 fields 768 shading units, 48 texture mapping units, and 32 ROPs, while the Radeon 740M has just 256 shading units, 16 TMUs, and 8 ROPs. Yet the Radeon compensates with far higher boost clocks: 2800 MHz versus 1188 MHz for the GTX 950. This clock advantage drives the Radeon’s higher FP32 throughput of 2.867 TFLOPS versus 1.825 TFLOPS for the GTX 950, despite having a third of the shaders. The Radeon also supports FP16 at a 1:1 ratio (2.867 TFLOPS), while the GTX 950 lists no FP16 capability.

Memory architecture is fundamentally different. The GTX 950 has dedicated 2 GB of GDDR5 on a 128-bit bus, delivering 105.8 GB/s of bandwidth. The Radeon 740M uses System Shared memory, with its size, type, bus width, and bandwidth all listed as "System Dependent." This means the Radeon’s memory performance varies entirely with the host system, whereas the GTX 950 has predictable, fixed bandwidth.

Feature sets also diverge. The Radeon 740M includes 4 ray tracing cores and supports DirectX 12 Ultimate (12_2), while the GTX 950 maxes out at DirectX 12 (12_1) with no ray tracing hardware. Both support OpenGL 4.6 and Vulkan 1.4. The GTX 950 offers dedicated display outputs (1x DVI, 1x HDMI 2.0, 3x DisplayPort 1.2), while the Radeon’s outputs are motherboard-dependent. The GTX 950 is a dual-slot card using PCIe 3.0 x16; the Radeon is an IGP using PCIe 4.0 x8.

Where Each One Wins

The GTX 950 wins outright in two of the two head-to-head benchmarks, but the nature of those wins suggests a split in real-world usage. In OpenCL—a workload common in compute and some productivity applications—the GTX 950 is 54% ahead, a commanding margin. This makes it the clear choice for OpenCL-centric tasks that can leverage its 768 shaders and dedicated GDDR5 bandwidth. The Radeon 740M’s reliance on shared system memory likely hampers its OpenCL performance, as memory latency and bandwidth are not dedicated to the GPU.

In Vulkan, however, the Radeon 740M is far more competitive, trailing by only 7.5%. Vulkan’s lower-level API overhead allows the Radeon’s higher clock speeds and modern architecture to shine, nearly offsetting its smaller shader count. For gaming or applications that use Vulkan, the Radeon 740M is effectively on par with the GTX 950, despite being an integrated part. This suggests that in modern, well-optimized titles, the Radeon could match or even edge out the older discrete card.

The Radeon 740M also wins on efficiency and integration. Its 45 W TDP is half the GTX 950’s 90 W, and it requires no external power connector. For thin-and-light laptops or compact systems where power and space are constrained, the Radeon 740M is the only viable option—the GTX 950’s dual-slot design and 202 mm length simply cannot fit. The Radeon also brings ray tracing support, which the GTX 950 lacks entirely, making it future-proof for DirectX 12 Ultimate titles.

Specification Differences

| Specification | NVIDIA GeForce GTX 950 | AMD Radeon 740M |

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

| Architecture | Maxwell 2.0 | RDNA 3.0 |

| Process Node | 28 nm | 4 nm |

| Transistors | 2,940 million | 20,900 million |

| Die Size | 228 mm² | 137 mm² |

| Transistor Density | 12.9M / mm² | 152.6M / mm² |

| Base Clock | 1024 MHz | 800 MHz |

| Boost Clock | 1188 MHz | 2800 MHz |

| Memory | 2 GB GDDR5 | System Shared |

| Memory Bus | 128 bit | System Shared |

| Memory Bandwidth | 105.8 GB/s | System Dependent |

| Shading Units | 768 | 256 |

| TMUs | 48 | 16 |

| ROPs | 32 | 8 |

| Ray Tracing Cores | None | 4 |

| FP32 Performance | 1.825 TFLOPS | 2.867 TFLOPS |

| FP16 Performance | None listed | 2.867 TFLOPS (1:1) |

| Pixel Rate | 38.02 GPixel/s | 22.40 GPixel/s |

| Texture Rate | 57.02 GTexel/s | 44.80 GTexel/s |

| TDP | 90 W | 45 W |

| Slot Width | Dual-slot | IGP |

| Power Connectors | 1x 6-pin | None |

| Suggested PSU | 250 W | None |

| Bus Interface | PCIe 3.0 x16 | PCIe 4.0 x8 |

| DirectX Support | 12 (12_1) | 12 Ultimate (12_2) |

| Display Outputs | 1x DVI, 1x HDMI 2.0, 3x DisplayPort 1.2 | Motherboard Dependent |

| Release Date | 2015-08-19 | 2024-01-30 |

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

| Launch MSRP | 159 USD | None |

Head-to-Head Benchmarks

The benchmark data shows a clear, but not total, victory for the GTX 950. In Geekbench OpenCL, the GTX 950 scores 15,662 against the Radeon 740M’s 10,172. That is a 54% delta—the single largest performance gap in this comparison. The GTX 950’s dedicated GDDR5 memory and 768 shading units give it a massive advantage in this compute-heavy test. The Radeon 740M’s shared memory architecture appears to be a severe bottleneck here, as its raw FP32 throughput (2.867 TFLOPS) is actually higher than the GTX 950’s (1.825 TFLOPS), yet it scores far lower. This indicates that OpenCL performance is heavily memory-bound for the Radeon, and the lack of dedicated VRAM hurts it disproportionately.

The Vulkan test paints a different picture. The GTX 950 still wins, scoring 16,734 versus 15,568, but the margin shrinks to just 7.5%. This narrower gap suggests that Vulkan’s more efficient API reduces the memory bottleneck, allowing the Radeon 740M’s superior clocks and modern architecture to express themselves. The Radeon’s 2800 MHz boost clock, compared to the GTX 950’s 1188 MHz, likely drives much of this recovery. In absolute terms, the GTX 950 improves from OpenCL to Vulkan by only 6.8%, while the Radeon 740M improves by a remarkable 53% between the two tests. This volatility in the Radeon’s scores suggests that its performance is highly workload-dependent, whereas the GTX 950 is more consistent.

Given that the GTX 950 wins both head-to-head tests, it claims 2 wins to the Radeon 740M’s 0. However, the 7.5% Vulkan margin is close enough that in real-world gaming scenarios—where Vulkan usage is increasing—the Radeon 740M could be considered roughly equivalent. The GTX 950’s 54% OpenCL lead, meanwhile, makes it the undisputed choice for compute tasks. The Radeon 740M’s nearest rivals include the AMD FirePro W5100 (0.2% ahead) and the NVIDIA GeForce GTX 590 (0.3% behind), while the GTX 950 sits near the NVIDIA GeForce GTX 480 (0.8% behind) and AMD FirePro M6100 (1.2% behind). Both GPUs occupy a similar performance tier, but they achieve it through radically different means.