AMD Radeon Pro 450 vs NVIDIA GeForce GTX 960M Comparison

Head-to-Head Benchmarks

The recorded database contains two direct comparison points between the AMD Radeon Pro 450 and the NVIDIA GeForce GTX 960M, and the results split cleanly along API lines. In Geekbench OpenCL, the NVIDIA GeForce GTX 960M posts a score of 11045 against the AMD Radeon Pro 450's 9002, a delta of 18.5 percent in NVIDIA's favor. That is a substantial margin, indicating the Maxwell-based part holds a clear advantage in this particular compute workload. In Geekbench Vulkan, the tables turn completely: the AMD Radeon Pro 450 scores 10518 while the GTX 960M manages only 8245, yielding a 27.6 percent lead for AMD. This is a larger swing than NVIDIA's OpenCL win, both in raw points and in percentage terms.

Looking at overall averages, the AMD part records an average benchmark score of 10804 across its three tested workloads (Geekbench Metal, OpenCL, and Vulkan), while the NVIDIA part averages 9645 from its two tests (OpenCL and Vulkan). That puts the AMD Radeon Pro 450 roughly 12 percent ahead on aggregate, despite losing the OpenCL head-to-head. The difference is explained by the Vulkan result, which is the stronger of the two AMD scores and the weaker of the two NVIDIA scores.

The percentile rankings reinforce this split. The AMD Radeon Pro 450 sits at the 49th percentile among all GPUs in the database, while the GTX 960M sits at the 46th. Both are mid-pack parts, but the AMD chip edges ahead in overall standing. When placed against their nearest rivals, the AMD part's average of 10804 lands within 0.4 percent of the NVIDIA Quadro K2200 (10761), within 0.7 percent of the GeForce MX350 (10883), and 1.1 percent above the GeForce GTX 560 Ti (10690). The GTX 960M's 9645 average is nearly identical to the Quadro K5000 (9637, 0.1 percent difference) and the Radeon Pro WX 2100 (9653, 0.1 percent difference in the opposite direction). In short, both GPUs cluster tightly with their direct contemporaries, but the AMD part's Vulkan strength pushes it into a slightly higher tier.

Where Each One Wins

The AMD Radeon Pro 450 wins decisively in Vulkan compute. Its 10518 score versus 8245 for the GTX 960M is a 27.6 percent advantage, which is the single largest margin recorded in either direction for this pairing. That Vulkan result also exceeds the AMD part's OpenCL score by over 1500 points, suggesting the GCN 4.0 architecture handles Vulkan's explicit low-level API particularly well. For workloads that leverage Vulkan, the AMD part is the clear choice.

The NVIDIA GeForce GTX 960M wins in OpenCL. Its 11045 score is 18.5 percent ahead of the AMD part's 9002, and it is also the highest single score recorded for either GPU in any shared test. Interestingly, the GTX 960M's OpenCL result is higher than its own Vulkan score by nearly 2800 points, indicating a pronounced weakness in Vulkan relative to OpenCL on the Maxwell architecture. For OpenCL-based compute tasks, the NVIDIA part holds a meaningful edge.

Beyond those two shared tests, the AMD Radeon Pro 450 also has a Geekbench Metal score of 12893, which is not available for the GTX 960M. That Metal result is the AMD part's best across all three recorded benchmarks, and it suggests Apple's Metal API is particularly well served by this GPU. The GTX 960M has no corresponding Metal score in the database, so no direct comparison is possible, but the AMD part's Metal output comfortably exceeds its own OpenCL and Vulkan numbers.

Architecture Differences

The two GPUs come from different foundries and process nodes. The AMD Radeon Pro 450 uses a 14 nm process at GlobalFoundries, while the NVIDIA GeForce GTX 960M uses a 28 nm process at TSMC. This node advantage shows up in transistor density: the AMD chip packs 3,000 million transistors into a 123 mm² die, yielding 24.4 million transistors per square millimeter. The NVIDIA chip, by contrast, has 1,870 million transistors on a 148 mm² die, for a density of 12.6 million per square millimeter. The AMD part achieves nearly double the transistor density on a smaller die.

The architectures themselves are from different generations. The AMD Radeon Pro 450 is built on GCN 4.0 with the Baffin chip, part of the Radeon Pro Mac (400 Series) generation. The NVIDIA GeForce GTX 960M uses the Maxwell architecture with the GM107 chip, from the GeForce 900M series. Both have the same shading unit count (640), texture mapping units (40), and render output units (16), so the raw shader throughput hardware is structurally similar. The difference lies in how those units are clocked and scheduled.

Clock speeds favor NVIDIA. The GTX 960M has a base clock of 1097 MHz and a boost clock of 1176 MHz, while the Radeon Pro 450's base and boost clocks are not recorded in the database. Memory clocks are close: the AMD part runs at 1270 MHz with 5.1 Gbps effective, while the NVIDIA part runs at 1253 MHz with 5 Gbps effective. Memory bandwidth is nearly identical, 81.28 GB/s for AMD versus 80.19 GB/s for NVIDIA, reflecting the same 128-bit bus width and similar GDDR5 memory clocks.

Despite similar shader counts, the resulting throughput numbers diverge. The NVIDIA part records a pixel rate of 18.82 GPixel/s and a texture rate of 47.04 GTexel/s, while the AMD part manages 12.80 GPixel/s and 32.00 GTexel/s. In FP32 compute, the GTX 960M delivers 1.505 TFLOPS against the AMD part's 1,024.0 GFLOPS. The AMD part does have FP16 support at a 1:1 ratio (1,024.0 GFLOPS), while the NVIDIA part has no recorded FP16 capability. Power consumption also differs: the AMD Radeon Pro 450 is rated at 35 W, while the GTX 960M draws 75 W. The AMD part delivers its compute results at less than half the power envelope.

API support shows a notable difference in DirectX and Vulkan versions. The AMD part supports DirectX 12 (12_0) and Vulkan 1.3, while the NVIDIA part supports DirectX 12 (11_0) and Vulkan 1.4. Both support OpenGL 4.6. The higher Vulkan version on NVIDIA does not translate into better Vulkan performance in the recorded data, as the AMD part wins that test by a wide margin. Both use MXM modules and have portable-device-dependent display outputs. The NVIDIA part lists no power connectors, while the AMD part's power connector field is null. The GTX 960M uses an MXM-B (3.0) bus interface, while the AMD part uses PCIe 3.0 x8.

FAQ

Q: Which GPU has the higher average benchmark score?

A: The AMD Radeon Pro 450 has an average benchmark score of 10804, compared to 9645 for the NVIDIA GeForce GTX 960M, a difference of roughly 12 percent.

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

A: The NVIDIA GeForce GTX 960M wins the OpenCL test with a score of 11045, which is 18.5 percent higher than the AMD Radeon Pro 450's 9002.

Q: Which GPU performs better in Vulkan?

A: The AMD Radeon Pro 450 wins Vulkan decisively, scoring 10518 against the GTX 960M's 8245, a 27.6 percent advantage.

Q: What is the process node difference between the two chips?

A: The AMD Radeon Pro 450 is built on a 14 nm process at GlobalFoundries, while the NVIDIA GeForce GTX 960M uses a 28 nm process at TSMC.

Q: Do both GPUs have the same number of shading units?

A: Yes, both have 640 shading units, along with 40 texture mapping units and 16 render output units.

Q: Which GPU has higher power consumption?

A: The NVIDIA GeForce GTX 960M is rated at 75 W, while the AMD Radeon Pro 450 is rated at 35 W.

The Verdict

The data supports a clear split based on workload type. For Vulkan-based applications, the AMD Radeon Pro 450 is the stronger part. Its 10518 Vulkan score beats the GTX 960M by 27.6 percent, and that Vulkan result alone lifts the AMD part's average score above the NVIDIA part's despite the OpenCL loss. The AMD part also has a Metal score of 12893, which is the highest single result recorded for either GPU across any test, making it the better option for Metal-centric environments.

For OpenCL-heavy workloads, the NVIDIA GeForce GTX 960M is the better choice. Its 11045 OpenCL score is 18.5 percent ahead of the AMD part, and it also has higher pixel and texture rates (18.82 GPixel/s and 47.04 GTexel/s versus 12.80 and 32.00, respectively) as well as higher FP32 throughput (1.505 TFLOPS versus 1,024.0 GFLOPS). The GTX 960M also offers 4 GB of VRAM versus 2 GB on the AMD part, which could matter for memory-capacity-sensitive tasks.

The power story favors AMD. The Radeon Pro 450 draws 35 W against the GTX 960M's 75 W, and it still manages to win the Vulkan test and post a higher average score. The 14 nm process node and GCN 4.0 architecture deliver better efficiency per watt, even if the NVIDIA part has higher raw fill rates.

The recorded data shows one win each in the head-to-head tests, but the magnitudes differ. NVIDIA's OpenCL victory is 18.5 percent, while AMD's Vulkan victory is 27.6 percent. That larger margin, combined with the Metal score and the lower power draw, gives the AMD Radeon Pro 450 the edge in overall assessment. Users needing OpenCL performance or maximum VRAM should choose the GTX 960M; users prioritizing Vulkan, Metal, or power efficiency should choose the Radeon Pro 450.

Specification Differences

| Specification | AMD Radeon Pro 450 | NVIDIA GeForce GTX 960M |

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

| Architecture | GCN 4.0 | Maxwell |

| Chip | Baffin | GM107 |

| Generation | Radeon Pro Mac (400 Series) | GeForce 900M |

| Process Node | 14 nm | 28 nm |

| Foundry | GlobalFoundries | TSMC |

| Transistors | 3,000 million | 1,870 million |

| Die Size | 123 mm² | 148 mm² |

| Transistor Density | 24.4M / mm² | 12.6M / mm² |

| Base Clock | Not recorded | 1097 MHz |

| Boost Clock | Not recorded | 1176 MHz |

| Memory Clock | 1270 MHz (5.1 Gbps effective) | 1253 MHz (5 Gbps effective) |

| Memory Size | 2 GB | 4 GB |

| Memory Type | GDDR5 | GDDR5 |

| Memory Bus Width | 128 bit | 128 bit |

| Memory Bandwidth | 81.28 GB/s | 80.19 GB/s |

| Shading Units | 640 | 640 |

| TMUs | 40 | 40 |

| ROPs | 16 | 16 |

| Pixel Rate | 12.80 GPixel/s | 18.82 GPixel/s |

| Texture Rate | 32.00 GTexel/s | 47.04 GTexel/s |

| FP32 Performance | 1,024.0 GFLOPS | 1.505 TFLOPS |

| FP16 Performance | 1,024.0 GFLOPS (1:1) | Not recorded |

| TDP | 35 W | 75 W |

| Bus Interface | PCIe 3.0 x8 | MXM-B (3.0) |

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

| OpenGL Support | 4.6 | 4.6 |

| Vulkan Support | 1.3 | 1.4 |

| Release Date | 2016-10-29 | 2015-03-12 |

| Predecessor | Not recorded | GeForce 800M |

| Successor | Not recorded | GeForce 10 Mobile |

| Power Connectors | Not recorded | None |