AMD Radeon 860M vs NVIDIA GB10 Comparison

Head-to-Head Benchmarks

The recorded data shows a decisive performance gap between these two parts, with the NVIDIA GB10 winning both benchmark tests outright. The Geekbench OpenCL score for the NVIDIA GB10 is 120,137, which is 81.1% higher than the AMD Radeon 860M's score of 22,759. In the Vulkan test, the NVIDIA GB10 records 114,648 points against the AMD part's 30,043, a delta of 73.8% in favor of NVIDIA.

The magnitude of these differences is substantial. The NVIDIA GB10's OpenCL result is more than five times the AMD Radeon 860M's score, while the Vulkan gap is nearly four times. These are not marginal wins; they represent a fundamental difference in compute capability. The AMD Radeon 860M does show a relative strength in Vulkan compared to its own OpenCL result, scoring roughly 32% higher in Vulkan (30,043 versus 22,759). The NVIDIA GB10, by contrast, scores slightly lower in Vulkan (114,648) than in OpenCL (120,137), suggesting the AMD architecture handles Vulkan workloads relatively better than OpenCL, though it still trails by a wide margin.

The database places the AMD Radeon 860M in the 72nd percentile of all GPUs, while the NVIDIA GB10 sits in the 95th percentile. This percentile gap reflects the benchmark data: the NVIDIA part operates in a different performance tier entirely. The AMD Radeon 860M's nearest rivals include the NVIDIA GeForce MX550 (average score 26,421, delta -0.1%), the NVIDIA GeForce RTX 5060 (26,331, delta 0.3%), and the AMD Radeon RX 5700 XT 50th Anniversary (26,553, delta -0.6%). These are close competitors, all within roughly 1% of the AMD part's average score of 26,401. The NVIDIA GB10's nearest rivals are substantially faster, including the NVIDIA RTX 4000 SFF Ada Generation (117,088, delta 0.3%), the AMD Radeon PRO W7700 (118,976, delta -1.3%), and the NVIDIA Tesla V100 SXM2 16 GB (114,395, delta 2.6%).

What the data implies is that the AMD Radeon 860M competes with entry-level discrete GPUs from several generations ago, while the NVIDIA GB10 matches modern workstation-class accelerators. The average benchmark score for the AMD part is 26,401, while the NVIDIA GB10 averages 117,393. That is a 4.4x difference in average score, which is consistent with the individual test deltas.

Architecture Differences

The two GPUs belong to different architectural generations and target completely different market segments. The AMD Radeon 860M uses the RDNA 3.5 architecture, built on a 4 nm process at TSMC. It is part of the Navi III IGP generation for Strix Point Mobile, using the Krackan Point chip. The NVIDIA GB10 uses Blackwell 2.0, fabricated on a 5 nm TSMC process, and belongs to the Server Blackwell generation with the GB20B chip. The die size for the NVIDIA part is 382 mm², while the AMD part's die size is unknown in the database.

The compute resources differ sharply. The AMD Radeon 860M has 512 shading units, 32 texture mapping units, 16 render output units, and 8 ray tracing cores. The NVIDIA GB10 has 6,144 shading units, 384 TMUs, 48 ROPs, and 48 ray tracing cores. The NVIDIA part also includes 384 tensor cores, a feature entirely absent from the AMD Radeon 860M's specification sheet. The FP32 throughput is 3.072 TFLOPS for the AMD part and 29.71 TFLOPS for the NVIDIA GB10, a ratio of nearly 10:1. Both parts list FP16 at the same rate as FP32 (1:1), which indicates no dedicated half-precision acceleration on either GPU.

Clock speeds show an interesting inversion. The AMD Radeon 860M has a base clock of 600 MHz and a boost clock of 3,000 MHz, so it relies on aggressive boosting to reach its peak performance. The NVIDIA GB10 has a base clock of 1,665 MHz and a boost clock of 2,418 MHz, a much narrower boost range. The AMD part's boost clock is actually higher than the NVIDIA part's boost clock (3,000 MHz versus 2,418 MHz), but the NVIDIA part's massive shading unit count and higher base clock overwhelm the AMD part's frequency advantage.

Memory architecture is another major divider. The AMD Radeon 860M uses system-shared memory with no dedicated VRAM, a bus width listed as system shared, and bandwidth described as system dependent. The NVIDIA GB10 has 128 GB of LPDDR5X memory on a 256-bit bus, delivering 273.2 GB/s of bandwidth. The memory clock is listed as 1,067 MHz with 8.5 Gbps effective throughput. This dedicated memory allocation is critical for the NVIDIA part's performance, especially in Vulkan and OpenCL workloads that can utilize large datasets.

The power envelope is also vastly different. The AMD Radeon 860M carries a 15 W TDP, while the NVIDIA GB10 has a 140 W TDP. The database lists no power connectors for either part, and the NVIDIA GB10 suggests a 300 W power supply. The AMD part is an integrated GPU (IGP) with a PCIe 4.0 x8 interface, while the NVIDIA GB10 is also classified as an IGP but uses a PCIe 5.0 x16 interface. The NVIDIA part has physical dimensions of 150 mm by 51 mm by 150 mm (5.9 inches by 2 inches by 5.9 inches), while the AMD part's dimensions are not recorded.

API support differs completely. The AMD Radeon 860M supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The NVIDIA GB10 lists N/A for DirectX, OpenGL, and Vulkan, which indicates it is not designed for conventional graphics API workloads. This is a server-class part aimed at compute tasks rather than rendering. The display outputs also reflect this: the AMD part is listed as portable device dependent, while the NVIDIA GB10 has a single HDMI output.

The Verdict

The benchmark data supports a clear split. The NVIDIA GB10 is overwhelmingly faster in both recorded tests, with an 81.1% lead in OpenCL and a 73.8% lead in Vulkan. Its 95th percentile ranking versus the AMD Radeon 860M's 72nd percentile confirms that these parts serve different purposes. The NVIDIA GB10's average score of 117,393 puts it in the same class as the NVIDIA RTX 4000 SFF Ada Generation and AMD Radeon PRO W7700, both of which are workstation-grade accelerators. The AMD Radeon 860M's average score of 26,401 places it alongside the NVIDIA GeForce MX550 and RTX 5060, which are entry-level discrete parts.

The AMD Radeon 860M offers a 15 W TDP, making it suitable for battery-constrained portable devices. Its boost clock of 3,000 MHz is the highest frequency recorded for either part, and it supports modern graphics APIs including DirectX 12 Ultimate. The NVIDIA GB10 requires 140 W and a 300 W power supply suggestion, which rules out thin-and-light mobile integration. It also lacks conventional graphics API support, so it would not function as a general-purpose gaming GPU.

The data indicates the NVIDIA GB10 is a compute-first accelerator for server workloads, while the AMD Radeon 860M is a graphics-first integrated solution for mobile platforms. The 4.4x difference in average benchmark score (117,393 versus 26,401) is the dominant factor in any comparison. The AMD part's only advantages are lower power consumption, a smaller footprint (as an integrated solution), and support for standard graphics APIs. Neither part is a substitute for the other.

Specification Differences

The following fields differ between the two parts:

| Specification | AMD Radeon 860M | NVIDIA GB10 |

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

| Manufacturer | AMD | NVIDIA |

| Chip | Krackan Point | GB20B |

| Architecture | RDNA 3.5 | Blackwell 2.0 |

| Generation | Navi III IGP (Strix Point Mobile) | Server Blackwell (Bxx) |

| Process node | 4 nm | 5 nm |

| Die size | unknown | 382 mm² |

| Base clock | 600 MHz | 1,665 MHz |

| Boost clock | 3,000 MHz | 2,418 MHz |

| Memory size | System Shared | 128 GB |

| Memory type | System Shared | LPDDR5X |

| Memory bus width | System Shared | 256 bit |

| Memory bandwidth | System Dependent | 273.2 GB/s |

| Memory clock | System Shared | 1,067 MHz 8.5 Gbps effective |

| Shading units | 512 | 6,144 |

| TMUs | 32 | 384 |

| ROPs | 16 | 48 |

| Ray tracing cores | 8 | 48 |

| Tensor cores | null | 384 |

| Pixel rate | 48.00 GPixel/s | 116.1 GPixel/s |

| Texture rate | 96.00 GTexel/s | 928.5 GTexel/s |

| FP32 | 3.072 TFLOPS | 29.71 TFLOPS |

| FP16 | 3.072 TFLOPS (1:1) | 29.71 TFLOPS (1:1) |

| TDP | 15 W | 140 W |

| Suggested PSU | null | 300 W |

| Bus interface | PCIe 4.0 x8 | PCIe 5.0 x16 |

| Display outputs | Portable Device Dependent | 1x HDMI |

| DirectX | 12 Ultimate (12_2) | N/A |

| OpenGL | 4.6 | N/A |

| Vulkan | 1.4 | N/A |

| Dimensions | null | 150 mm 5.9 inches, 51 mm 2 inches, 150 mm 5.9 inches |

| Release date | 2025-02-28 | 2025-10-14 |

| Predecessor | Navi II IGP | Server Hopper |

| Successor | null | Server Rubin |

| Launch MSRP | null | 3,999 USD |

The AMD Radeon 860M has a higher boost clock but lower base clock than the NVIDIA GB10. The NVIDIA part has 12 times the shading units, 12 times the TMUs, 3 times the ROPs, 6 times the ray tracing cores, and a 9.7x advantage in FP32 throughput. The NVIDIA GB10 also offers dedicated 128 GB memory versus the AMD part's shared memory model.

FAQ

Q: Which GPU has the higher benchmark score?

A: The NVIDIA GB10 wins both recorded tests. Its Geekbench OpenCL score is 120,137 versus 22,759 for the AMD Radeon 860M, and its Vulkan score is 114,648 versus 30,043.

Q: What is the performance difference in percentage terms?

A: The NVIDIA GB10 leads by 81.1% in Geekbench OpenCL and by 73.8% in Geekbench Vulkan. The average benchmark scores are 117,393 for the NVIDIA GB10 and 26,401 for the AMD Radeon 860M.

Q: Do both GPUs support ray tracing?

A: Yes, both have ray tracing cores. The AMD Radeon 860M has 8 ray tracing cores, while the NVIDIA GB10 has 48 ray tracing cores.

Q: What memory configurations do they use?

A: The AMD Radeon 860M uses system-shared memory with bandwidth dependent on the host system. The NVIDIA GB10 has 128 GB of LPDDR5X memory on a 256-bit bus with 273.2 GB/s bandwidth.

Q: Which GPU has tensor cores?

A: Only the NVIDIA GB10 has tensor cores, with 384 of them. The AMD Radeon 860M lists no tensor cores in the database.

Q: What are their power requirements?

A: The AMD Radeon 860M has a 15 W TDP. The NVIDIA GB10 has a 140 W TDP and a suggested power supply of 300 W.

Where Each One Wins

The NVIDIA GB10 wins in raw compute performance across both benchmark tests. Its OpenCL score of 120,137 and Vulkan score of 114,648 place it in the 95th percentile of all GPUs, matching the performance of workstation accelerators like the NVIDIA RTX 4000 SFF Ada Generation (117,088) and the AMD Radeon PRO W7700 (118,976). The NVIDIA part also wins on memory capacity and bandwidth, offering 128 GB of LPDDR5X at 273.2 GB/s, which is essential for large-scale compute tasks. Its 6,144 shading units, 384 TMUs, and 48 ROPs provide the hardware resources necessary for high-throughput workloads. The 384 tensor cores give it a capability the AMD part lacks entirely.

The AMD Radeon 860M wins on power efficiency and integration. Its 15 W TDP is less than one-tenth of the NVIDIA GB10's 140 W TDP. The AMD part operates as a true integrated GPU with no power connectors and a PCIe 4.0 x8 interface, while the NVIDIA GB10, despite also being labeled an IGP, requires a 300 W power supply suggestion and has physical dimensions of 150 mm by 51 mm by 150 mm. The AMD part's boost clock of 3,000 MHz exceeds the NVIDIA part's 2,418 MHz, which may help in latency-sensitive light workloads. The AMD Radeon 860M also supports DirectX 12 Ultimate, OpenGL 4.6, and Vulkan 1.4, whereas the NVIDIA GB10 lists no conventional graphics API support. For portable devices requiring graphics output and modern API compatibility, the AMD part is the only viable option. The NVIDIA GB10, with its 3,999 USD launch MSRP, is a server-oriented compute accelerator.