AMD Radeon Vega 8 vs NVIDIA GeForce GTX 950M Comparison

Head-to-Head Benchmarks

The direct comparison between these two mobile graphics solutions is unusually close, with each side claiming one decisive victory in the two recorded OpenCL and Vulkan tests. The database shows a split decision, and the magnitude of each win tells a different story about the strengths of each architecture.

In the Geekbench OpenCL test, the NVIDIA GeForce GTX 950M posts a score of 9,745 against 8,822 for the AMD Radeon Vega 8. That is a 9.5% advantage for the NVIDIA part, a solid but not overwhelming margin. The OpenCL result aligns with the GTX 950M's higher raw throughput figures, which the specification data confirms: the NVIDIA chip carries 640 shading units, 40 texture mapping units, and 16 ROPs, compared to 512 shading units, 32 TMUs, and 8 ROPs on the AMD side. The GTX 950M also achieves a higher FP32 compute rate of 1,438.7 GFLOPS versus 1,126.4 GFLOPS for the Vega 8. In raw compute workloads that scale well with shader count and fill rate, the NVIDIA part is the clear leader.

The Vulkan test flips the script dramatically. The AMD Radeon Vega 8 scores 8,134, while the GeForce GTX 950M manages only 6,525. That is a 24.7% advantage for the AMD part, a far larger margin than NVIDIA's OpenCL lead. Vulkan is a low-level API that exposes hardware directly to developers, and the data suggests the Vega 8's architecture handles this workload far more efficiently. The Vega 8's GCN 5.0 design, paired with its 2.253 TFLOPS of FP16 compute (using a 2:1 ratio), appears to give it a significant edge in scenarios that leverage modern API features and mixed-precision math. The GTX 950M, based on the older Maxwell architecture, has no recorded FP16 capability in the database, which may explain its weaker Vulkan showing.

Looking at the average benchmark scores across all recorded tests, the AMD Radeon Vega 8 sits at 9,221, while the GeForce GTX 950M averages 8,135. That flips the individual OpenCL result: the Vega 8's strong Vulkan performance lifts its overall average above the GTX 950M by 13.3%. However, the percentile rankings tell a slightly different story. The Vega 8 lands at the 45th percentile among all GPUs in the database, while the GTX 950M sits at the 42nd percentile. Both are mid-pack parts, and the percentile gap of three points is modest. The nearest rivals for each card reinforce this positioning: the Vega 8's closest competitor is the AMD Radeon 890M (average score 9,210, a 0.1% delta), while the GTX 950M's nearest rival is the AMD Radeon R9 M360 (average score 8,129, a 0.1% delta). Interestingly, the Vega 8 also sits within 0.9% of the GeForce GTX 960 and GTX 850M, while the GTX 950M is within 0.7% of the NVIDIA GRID K2. The data positions both cards in a tight cluster of mid-range mobile and entry-level desktop parts, with no single dominant outlier.

Architecture Differences

The two chips come from fundamentally different design philosophies and fabrication eras. The AMD Radeon Vega 8 uses the Raven chip built on GlobalFoundries' 14 nm process, with a die size of 210 mm² and 4,940 million transistors. That yields a transistor density of 23.5 million per square millimeter, a figure that reflects the more modern manufacturing node. The NVIDIA GeForce GTX 950M uses the GM107 chip on TSMC's 28 nm process, with 1,870 million transistors on a 148 mm² die, giving a density of 12.6 million per square millimeter. The Vega 8 packs nearly 2.6 times more transistors into a die that is only 42% larger, a direct consequence of the 14 nm versus 28 nm process gap.

The memory subsystem is another major divergence. The Vega 8 uses system shared memory for both capacity and bandwidth, with the exact figures dependent on the host system's configuration. The GTX 950M has a dedicated 4 GB DDR3 pool on a 128-bit bus, delivering 28.80 GB/s of bandwidth. This is a classic trade-off: the integrated AMD part relies on fast system memory (which can be a double-edged sword depending on the platform), while the discrete NVIDIA part has a fixed, predictable memory pipeline. The Vega 8's memory clock is listed as "System Shared" with no fixed speed, while the GTX 950M runs at a base memory clock of 900 MHz with 1800 Mbps effective.

Clock speeds also favor the NVIDIA part on paper. The GTX 950M runs at a 993 MHz base clock and boosts to 1124 MHz. The Vega 8 starts at a lowly 300 MHz base but boosts to 1100 MHz. Those numbers, taken in isolation, suggest the NVIDIA chip should be faster, and indeed the OpenCL result confirms that for compute-heavy workloads. But the Vulkan result shows that clock speed alone does not tell the whole story, as the Vega 8's architectural efficiency with modern APIs compensates for its lower base clock.

The API support lists also differ in a notable way. Both cards support DirectX 12 and OpenGL 4.6, but the Vega 8 lists DirectX 12 (12_1) while the GTX 950M lists DirectX 12 (11_0). The Vega 8 also supports Vulkan 1.3, while the GTX 950M supports Vulkan 1.4. The higher Vulkan version on the NVIDIA part makes its 24.7% loss in the Vulkan benchmark even more striking, as the software layer is not the limiting factor.

The power situation is dramatically different. The Vega 8 has a TDP of 25 W, while the GTX 950M draws 75 W. That is a 3x difference in power consumption, which matters for laptop battery life and thermal management. Both are listed as IGP (integrated graphics processor) slot width with no power connectors, but the Vega 8 is a true integrated part (bus interface: IGP), while the GTX 950M connects via PCIe 3.0 x8. The Vega 8's display outputs are motherboard dependent, while the GTX 950M's are portable device dependent, reflecting their intended platforms.

FAQ

Q: Which GPU has the higher average benchmark score?

A: The AMD Radeon Vega 8 has an average benchmark score of 9,221, compared to 8,135 for the NVIDIA GeForce GTX 950M, a difference of 13.3% in favor of the AMD part.

Q: How do the two GPUs compare in Vulkan performance?

A: The AMD Radeon Vega 8 scores 8,134 in the Vulkan test, while the NVIDIA GeForce GTX 950M scores 6,525. The Vega 8 leads by 24.7% in this workload.

Q: What is the power consumption difference?

A: The AMD Radeon Vega 8 has a TDP of 25 W, while the NVIDIA GeForce GTX 950M has a TDP of 75 W. The Vega 8 consumes one-third the power of the GTX 950M.

Q: Which GPU has more shading units?

A: The NVIDIA GeForce GTX 950M has 640 shading units, while the AMD Radeon Vega 8 has 512 shading units. The GTX 950M also has more TMUs (40 versus 32) and ROPs (16 versus 8).

Q: What is the memory configuration of each GPU?

A: The NVIDIA GeForce GTX 950M has 4 GB of DDR3 memory on a 128-bit bus with 28.80 GB/s bandwidth. The AMD Radeon Vega 8 uses system shared memory with system-dependent bandwidth.

Q: How do the nearest rivals compare for each card?

A: The AMD Radeon Vega 8's nearest rival is the AMD Radeon 890M (average score 9,210, 0.1% delta), while the NVIDIA GeForce GTX 950M's nearest rival is the AMD Radeon R9 M360 (average score 8,129, 0.1% delta).

The Verdict

The data paints a nuanced picture that resists a simple "winner" declaration. The AMD Radeon Vega 8 wins the average benchmark score battle (9,221 versus 8,135) and the Vulkan test by a wide 24.7% margin. The NVIDIA GeForce GTX 950M wins the OpenCL test by 9.5% and offers a dedicated 4 GB memory pool with a fixed 28.80 GB/s bandwidth, which removes the variable of system memory quality. The Vega 8 counters with a 25 W TDP versus 75 W, making it far more energy-efficient, and a 14 nm process that packs 4,940 million transistors versus 1,870 million.

For users who prioritize modern API performance (Vulkan) and power efficiency, the AMD Radeon Vega 8 is the better choice based on the recorded data. For users who need consistent OpenCL compute performance and a discrete memory buffer, the NVIDIA GeForce GTX 950M holds the advantage. The percentile rankings (45th versus 42nd) suggest both are similar in the broader GPU landscape, but the Vega 8's higher average score and lower power draw give it the edge in overall value from a pure performance-per-watt standpoint.

Specification Differences

| Specification | AMD Radeon Vega 8 | NVIDIA GeForce GTX 950M |

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

| Architecture | GCN 5.0 | Maxwell |

| Process Node | 14 nm | 28 nm |

| Foundry | GlobalFoundries | TSMC |

| Transistors | 4,940 million | 1,870 million |

| Die Size | 210 mm² | 148 mm² |

| Transistor Density | 23.5M / mm² | 12.6M / mm² |

| Base Clock | 300 MHz | 993 MHz |

| Boost Clock | 1100 MHz | 1124 MHz |

| Memory Size | System Shared | 4 GB |

| Memory Type | System Shared | DDR3 |

| Memory Bus Width | System Shared | 128 bit |

| Memory Bandwidth | System Dependent | 28.80 GB/s |

| Shading Units | 512 | 640 |

| TMUs | 32 | 40 |

| ROPs | 8 | 16 |

| Pixel Rate | 8.800 GPixel/s | 17.98 GPixel/s |

| Texture Rate | 35.20 GTexel/s | 44.96 GTexel/s |

| FP32 Performance | 1,126.4 GFLOPS | 1,438.7 GFLOPS |

| FP16 Performance | 2.253 TFLOPS (2:1) | Not listed |

| TDP | 25 W | 75 W |

| Bus Interface | IGP | PCIe 3.0 x8 |

| DirectX Support | 12 (12_1) | 12 (11_0) |

| Vulkan Support | 1.3 | 1.4 |

| Release Date | 2018-02-11 | 2015-03-12 |

Where Each One Wins

AMD Radeon Vega 8 wins on: Vulkan performance, where it leads by 24.7%. Average benchmark score, where it leads by 13.3%. Power efficiency, with a 25 W TDP versus 75 W. Transistor density, at 23.5M per mm² versus 12.6M per mm². FP16 compute capability, with 2.253 TFLOPS available versus none recorded on the NVIDIA part. DirectX 12 feature level, with 12_1 support versus 11_0. Overall GPU percentile ranking, at 45th versus 42nd.

NVIDIA GeForce GTX 950M wins on: OpenCL performance, where it leads by 9.5%. Raw shader throughput, with 640 shading units versus 512. Texture and pixel fill rates, at 44.96 GTexel/s and 17.98 GPixel/s versus 35.20 GTexel/s and 8.800 GPixel/s. FP32 compute, at 1,438.7 GFLOPS versus 1,126.4 GFLOPS. Dedicated memory bandwidth, with a fixed 28.80 GB/s versus system-dependent bandwidth. Higher base and boost clocks, at 993 MHz and 1124 MHz versus 300 MHz and 1100 MHz. Vulkan API version, with 1.4 support versus 1.3.