AMD Radeon Pro 455 vs NVIDIA GeForce GTX 965M Comparison

Head-to-Head Benchmarks

The recorded data shows a clear pattern in the head-to-head comparison between the NVIDIA GeForce GTX 965M and the AMD Radeon Pro 455. Across the two shared benchmark tests, the NVIDIA part secures both wins, but the magnitude of those victories differs substantially depending on the API workload.

In Geekbench OpenCL, the NVIDIA GeForce GTX 965M scores 14509 points against the AMD Radeon Pro 455's 10336 points. That is a delta of 40.4% in NVIDIA's favor. This is a dominant result, suggesting the Maxwell 2.0 architecture's compute throughput scales far better under OpenCL in this configuration. The AMD part's score here is notably lower than its own Vulkan result, which hints at a driver or hardware inefficiency in OpenCL dispatch for the Radeon Pro 455.

The Vulkan test narrows the gap considerably. NVIDIA still wins, scoring 14299 versus AMD's 12240, but the delta shrinks to 16.8%. While NVIDIA remains ahead, the Radeon Pro 455 closes more than half of the percentage-point gap it faced in OpenCL. This could indicate that GCN 4.0's asynchronous compute or command processor handles Vulkan's low-level overhead more effectively than its OpenCL path.

Looking at the broader averages, the NVIDIA card's average benchmark score is 14404, while the AMD card sits at 12831. That places the GTX 965M in the 56th percentile of all GPUs, versus the Radeon Pro 455's 52nd percentile. The nearest rivals for the NVIDIA part include the AMD Radeon RX Vega 11 at 14385 (0.1% delta) and the NVIDIA GeForce GTX TITAN at 14373 (0.2% delta), showing that the 965M sits in a tightly packed cluster. For the AMD side, the closest competitor is the NVIDIA GeForce GTX 590 at 12830 (0% delta), followed by the AMD FirePro W5100 at 12847 (-0.1% delta). These margins are razor-thin, meaning the Radeon Pro 455 is essentially tied with several older high-end parts in aggregate performance.

Architecture Differences

The two GPUs come from fundamentally different design philosophies and manufacturing nodes. The NVIDIA GeForce GTX 965M uses the GM204 chip based on Maxwell 2.0 architecture, fabricated by TSMC on a 28 nm process. The die measures 398 mm² and packs 5,200 million transistors, yielding a transistor density of 13.1M per mm². In contrast, the AMD Radeon Pro 455 uses the Baffin chip based on GCN 4.0, built by GlobalFoundries on a 14 nm process. Its die is far smaller at 123 mm², containing 3,000 million transistors, which results in a much higher density of 24.4M per mm². The shift to 14 nm allows AMD to pack more logic per area despite fewer total transistors.

The compute resources differ significantly. The GTX 965M carries 1024 shading units, 64 texture mapping units, and 32 raster operations pipelines. The Radeon Pro 455 has 768 shading units, 48 TMUs, and only 16 ROPs. These raw counts explain part of the performance gap: NVIDIA has 33% more shading units, 33% more TMUs, and double the ROPs. The pixel rate reflects this: NVIDIA achieves 30.40 GPixel/s while AMD manages 13.68 GPixel/s. Texture rate follows the same pattern: 60.80 GTexel/s for NVIDIA versus 41.04 GTexel/s for AMD. In FP32 compute, the GTX 965M reaches 1.946 TFLOPS, while the Radeon Pro 455 outputs 1,313.3 GFLOPS. Interestingly, the AMD card offers FP16 at 1,313.3 GFLOPS with a 1:1 ratio, whereas the NVIDIA part lists no FP16 capability, meaning it likely relies on FP32 emulation for half-precision work.

Clock behavior also diverges. The NVIDIA chip has explicit base and boost clocks of 924 MHz and 950 MHz, respectively. The AMD part lists no base or boost clock in the data, only its memory clock of 1270 MHz with 5.1 Gbps effective. Memory bandwidth is nearly identical: the GTX 965M provides 80.19 GB/s, the Radeon Pro 455 provides 81.28 GB/s, both over a 128-bit GDDR5 bus with 2 GB capacity. The slight edge in bandwidth goes to AMD, but it is too small to offset the compute and ROP disadvantages.

API support shows another split. The NVIDIA card supports DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.4. The AMD part supports DirectX 12 (12_0), OpenGL 4.6, and Vulkan 1.3. The higher DirectX feature level and newer Vulkan version on the NVIDIA side could matter for certain workloads, though the benchmark data does not isolate these specific effects. The AMD card draws 35 W, while the NVIDIA card's TDP is not listed in the database. Both use MXM modules with no external power connectors, and both have display outputs described as "Portable Device Dependent."

The Verdict

The benchmark data points to a decisive overall winner: the NVIDIA GeForce GTX 965M. In OpenCL, it leads by 40.4% (14509 versus 10336). In Vulkan, it leads by 16.8% (14299 versus 12240). Its average score of 14404 sits 12.3% above the AMD part's 12831. The percentile rankings reinforce this: 56th for NVIDIA versus 52nd for AMD. If the task is purely about raw compute performance in shared benchmarks, the GTX 965M is the stronger choice.

However, the Radeon Pro 455 is not without merit. Its 14 nm node and higher transistor density suggest better power efficiency per transistor, and its FP16 support at 1:1 ratio could benefit specific half-precision workloads, though the database does not include a benchmark that isolates this. The AMD card also has a slightly higher memory bandwidth (81.28 GB/s versus 80.19 GB/s) and a lower TDP of 35 W, which may matter in thermally constrained portable chassis. But on the evidence of recorded scores, the GTX 965M wins every head-to-head test. For users who prioritize compute performance in OpenCL or Vulkan, the choice is clear. For those who need FP16 or lower power draw, the Radeon Pro 455 has niche advantages, but those are not reflected in the benchmark averages.

FAQ

Q: Which GPU has the higher average benchmark score?

A: The NVIDIA GeForce GTX 965M, with an average score of 14404 versus the AMD Radeon Pro 455's 12831.

Q: How large is the performance gap in OpenCL?

A: The GTX 965M scores 14509 in Geekbench OpenCL, which is 40.4% higher than the Radeon Pro 455's 10336.

Q: Does the AMD Radeon Pro 455 win any head-to-head test?

A: No. The database shows the GTX 965M winning both the OpenCL and Vulkan tests. The AMD card's wins tally is 0.

Q: What are the memory specifications of both cards?

A: Both have 2 GB of GDDR5 on a 128-bit bus. The GTX 965M has a bandwidth of 80.19 GB/s, while the Radeon Pro 455 has 81.28 GB/s.

Q: Which GPU has more shading units?

A: The NVIDIA GTX 965M has 1024 shading units, compared to 768 on the AMD Radeon Pro 455.

Q: What is the difference in process node and die size?

A: The GTX 965M uses a 28 nm process with a 398 mm² die, while the Radeon Pro 455 uses a 14 nm process with a 123 mm² die.

Where Each One Wins

The NVIDIA GeForce GTX 965M wins in every recorded benchmark, but the nature of those wins suggests specific strengths. Its 40.4% lead in OpenCL indicates strong general compute performance, likely driven by the higher shading unit count (1024 versus 768) and double the ROPs (32 versus 16). The 16.8% lead in Vulkan shows it also handles modern low-level APIs well, though the margin is tighter. For tasks that rely on pixel throughput or texture fill, the GTX 965M's 30.40 GPixel/s and 60.80 GTexel/s are far ahead of the Radeon Pro 455's 13.68 GPixel/s and 41.04 GTexel/s. This makes the NVIDIA part better suited for rasterization-heavy workloads, even in a mobile form factor.

The AMD Radeon Pro 455's wins are more subtle and not captured by the head-to-head tests. It offers FP16 compute at 1,313.3 GFLOPS with a 1:1 ratio, which the GTX 965M cannot match (no FP16 listed). This could benefit machine learning inference or image processing that uses half precision. Its 35 W TDP, if accurate, is a clear advantage for battery life and thermal management in thin laptops, though the GTX 965M's TDP is not listed for comparison. The AMD card also has a marginally higher memory bandwidth (81.28 GB/s versus 80.19 GB/s), which might help in bandwidth-bound scenarios, but the effect is negligible given the overall compute deficit. The Radeon Pro 455's smaller die (123 mm² versus 398 mm²) and higher transistor density (24.4M per mm² versus 13.1M per mm²) suggest better manufacturing efficiency, but that does not translate into benchmark wins.

In practical terms, the GTX 965M is the pick for anyone running OpenCL or Vulkan applications where raw performance matters. The Radeon Pro 455 makes sense only for specialized workloads that leverage FP16 or where the lower power draw is a hard requirement. The database does not include a direct FP16 benchmark, so that advantage remains potential rather than proven.

Specification Differences

| Specification | NVIDIA GeForce GTX 965M | AMD Radeon Pro 455 |

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

| Chip | GM204 | Baffin |

| Architecture | Maxwell 2.0 | GCN 4.0 |

| Process Node | 28 nm | 14 nm |

| Foundry | TSMC | GlobalFoundries |

| Transistors | 5,200 million | 3,000 million |

| Die Size | 398 mm² | 123 mm² |

| Transistor Density | 13.1M / mm² | 24.4M / mm² |

| Shading Units | 1024 | 768 |

| TMUs | 64 | 48 |

| ROPs | 32 | 16 |

| Pixel Rate | 30.40 GPixel/s | 13.68 GPixel/s |

| Texture Rate | 60.80 GTexel/s | 41.04 GTexel/s |

| FP32 | 1.946 TFLOPS | 1,313.3 GFLOPS |

| FP16 | Not listed | 1,313.3 GFLOPS (1:1) |

| TDP | Not listed | 35 W |

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

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

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

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

| Vulkan | 1.4 | 1.3 |

| Release Date | 2015-01-08 | 2016-10-29 |

| Avg Benchmark Score | 14404 | 12831 |

| Percentile | 56 | 52 |