AMD Radeon 860M vs NVIDIA GeForce RTX 3080 Comparison

The AMD Radeon 860M and NVIDIA GeForce RTX 3080 occupy entirely different tiers of the graphics hardware spectrum, yet their benchmark data reveals a more nuanced relationship than their specifications might suggest. The Radeon 860M is an integrated graphics processor built into AMD's Krackan Point mobile chips, while the RTX 3080 is a high-end discrete desktop GPU from the GeForce 30-series. The database places the Radeon 860M at the 72nd percentile of all GPUs, while the RTX 3080 sits at the 68th percentile, a surprising inversion that stems from the different benchmark suites recorded for each product. The two share only two direct head-to-head measurements: Geekbench OpenCL and Geekbench Vulkan, where the RTX 3080 wins both, but by vastly different margins.

Where Each One Wins

The Radeon 860M wins in no direct head-to-head benchmark, as the recorded data shows zero wins for the integrated part. However, its average benchmark score of 26401 across its two recorded tests exceeds the RTX 3080's average of 23172, which is heavily weighted by a large set of Passmark tests where the 860M has no recorded entries. This discrepancy matters: the 860M's average is pulled upward by a strong Vulkan score of 30043, while the RTX 3080's average is dragged down by low Passmark DirectX scores that range from 100 to 258. In the shared Geekbench suite, the RTX 3080 demonstrates its dominance in raw compute, delivering a Geekbench OpenCL score of 152423 versus 22759 for the 860M, a delta of -85.1% for the AMD part. In Geekbench Vulkan, the gap narrows dramatically, with the RTX 3080 scoring 33620 against the 860M's 30043, a delta of -10.6%. This suggests the 860M is highly competitive in Vulkan-based workloads, likely because its RDNA 3.5 architecture is well-optimized for this API, whereas the RTX 3080's enormous advantage in OpenCL reflects its sheer shading power and memory bandwidth.

The RTX 3080 wins decisively in compute-heavy tasks, particularly those using OpenCL, where its 8704 shading units and 29.77 TFLOPS FP32 throughput dwarf the 860M's 512 shading units and 3.072 TFLOPS. The 860M, by contrast, shows its best relative performance in Vulkan, trailing by only 10.6%, which positions it as a capable option for lighter gaming or graphics workloads that lean on modern APIs. For users focused on integrated graphics in a 15 W envelope, the 860M's Vulkan score is a standout, but for any task requiring sustained compute, the RTX 3080 is the clear winner.

Architecture Differences

The architectural gulf between these two GPUs is vast. The Radeon 860M is built on TSMC's 4 nm process and uses AMD's RDNA 3.5 architecture, part of the Navi III IGP generation for Strix Point Mobile. It packs 512 shading units, 32 texture mapping units, 16 ROPs, and 8 ray tracing cores. Its base clock is 600 MHz with a boost clock of 3000 MHz, producing a pixel rate of 48.00 GPixel/s and a texture rate of 96.00 GTexel/s. FP32 performance is 3.072 TFLOPS, with FP16 at the same level (1:1). Memory is system shared, meaning size, type, and bus width are all system dependent, and bandwidth is listed as system dependent as well. The chip has no dedicated tensor cores, and its TDP is just 15 W. It interfaces via PCIe 4.0 x8 and supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. It is an IGP with no power connectors and no slot width beyond its integrated nature, and it was released on February 28, 2025, with a predecessor of Navi II IGP.

The NVIDIA GeForce RTX 3080, in contrast, is a discrete dual-slot card built on Samsung's 8 nm process using the Ampere architecture, with the GA102 chip. It contains 28,300 million transistors on a 628 mm² die, yielding a transistor density of 45.1M / mm². It has 8704 shading units, 272 TMUs, 96 ROPs, 68 RT cores, and 272 tensor cores. Base clock is 1440 MHz with a boost of 1710 MHz, producing a pixel rate of 164.2 GPixel/s and a texture rate of 465.1 GTexel/s. FP32 performance is 29.77 TFLOPS, with FP16 at the same level (1:1). Memory is 10 GB of GDDR6X on a 320-bit bus, with a bandwidth of 760.3 GB/s and an effective memory clock of 19 Gbps. TDP is 320 W, requiring a 700 W suggested PSU and a single 12-pin power connector. It uses PCIe 4.0 x16, offers 1x HDMI 2.1 and 3x DisplayPort 1.4a outputs, and measures 285 mm in length. It was released on August 31, 2020, is now end-of-life, and had a launch MSRP of 699 USD. Its predecessor is GeForce 20 and successor is GeForce 40.

The key differences are process node (4 nm versus 8 nm), transistor count (unknown versus 28,300 million), die size (unknown versus 628 mm²), memory architecture (system shared versus 10 GB GDDR6X), and compute resources (512 versus 8704 shading units). The 860M relies on system memory, which is a fundamental bottleneck for bandwidth, whereas the RTX 3080's 760.3 GB/s dedicated bandwidth is a massive advantage. The 860M's higher boost clock of 3000 MHz versus 1710 MHz partially compensates for fewer cores, but not enough to close the compute gap. The 860M also lacks tensor cores, while the RTX 3080 has 272, making the latter far more capable for AI and machine learning workloads.

Head-to-Head Benchmarks

The shared benchmark suite consists of just two tests: Geekbench OpenCL and Geekbench Vulkan. In Geekbench OpenCL, the RTX 3080 scores 152423, versus 22759 for the 860M, a delta of -85.1% for the AMD part. This is a crushing defeat for the 860M, reflecting the RTX 3080's 9.7x advantage in raw FP32 throughput (29.77 versus 3.072 TFLOPS) and its vastly superior memory bandwidth. OpenCL workloads often scale with shading units and memory bandwidth, both of which the RTX 3080 has in abundance. The 860M's system-shared memory, with bandwidth dependent on the host system, cannot compete with 760.3 GB/s.

In Geekbench Vulkan, the results are far closer. The RTX 3080 scores 33620, while the 860M scores 30043, a delta of -10.6%. This 10.6% gap is remarkable given the difference in hardware resources. The 860M's RDNA 3.5 architecture appears to handle Vulkan efficiently, and its high boost clock of 3000 MHz likely contributes to closing the gap. The RTX 3080 still wins, but the margin is modest. For a user with an 860M, this means Vulkan-based games or applications could run at playable levels, whereas OpenCL-heavy tasks would expose the integrated GPU's limitations. The RTX 3080 also has a recorded 3DMark Steel Nomad DX12 score of 4407, but the 860M has no comparable entry, so no direct comparison is possible there. Passmark scores for the RTX 3080 include G3D at 25086, GPU Compute at 14397, and G2D at 1054, but again the 860M has no Passmark data in the database, limiting cross-referencing beyond the Geekbench pair.

The data shows the RTX 3080 wins both head-to-head tests, giving it a 2-0 win record, while the 860M has zero wins. The average benchmark scores, however, tell a different story in aggregate: 26401 for the 860M versus 23172 for the RTX 3080. This is because the RTX 3080's average includes many low Passmark DirectX scores (170 for DirectX 10, 207 for DirectX 11, 100 for DirectX 12, 258 for DirectX 9) that are not part of the 860M's suite, skewing the average downward. When compared to nearest rivals, the 860M is within 0.1% below the NVIDIA GeForce MX550, 0.3% above the NVIDIA GeForce RTX 5060, 0.6% below the AMD Radeon RX 5700 XT 50th Anniversary, and 1.1% below the NVIDIA RTX 4000. The RTX 3080, on the other hand, is 0.3% below the NVIDIA P106-100, 0.3% below the AMD Radeon Pro Vega 16, 0.4% below the AMD Radeon RX 6600M, and 0.4% below the AMD Radeon R9 M290X.

The Verdict

From the recorded data alone, the choice depends entirely on the user's workload context. The NVIDIA GeForce RTX 3080 is the superior product in raw compute performance in every shared benchmark. Its Geekbench OpenCL score of 152423 crushes the AMD Radeon 860M's 22759, a delta of -85.1%. For anyone needing heavy compute, ray tracing, or AI acceleration, the RTX 3080 is the only defensible selection. It offers 68 RT cores and 272 tensor cores, features the 860M simply lacks. Its 10 GB GDDR6X memory with 760.3 GB/s bandwidth also ensures high sustained performance in memory-intensive applications.

The AMD Radeon 860M, however, is a niche winner in efficiency and efficiency-adjacent metrics, though no direct head-to-head win is recorded, its 15 W TDP and IGP form factor make it suitable for thin-and-light mobile devices, where a discrete 320 W card is physically impossible. The Vulkan score of 30043 is only 10.6% behind the RTX 3080, which is a strong showing for an integrated part. Users who prioritize portability and run Vulkan workloads might find the 860M adequate, but the data cannot recommend it over the RTX 3080 for any performance-critical task. The RTX 3080 is end-of-life, so availability may be limited, whereas the 860M is active. The RTX 3080's launch MSRP was 699 USD, and its successor is GeForce 40, indicating a clear product lifecycle. The 860M has no successor listed.

If the use case demands the highest possible benchmark scores in OpenCL, choose the RTX 3080 without hesitation. If the use case demands integrated graphics in a low-power envelope, the 860M is a valid option, but with the understanding that it loses both head-to-head tests, the verdict from the database is unambiguously in favor of NVIDIA in every shared test scenario, leaving the 860M's relative strength in Vulkan as a consolation, not a victory, so the RTX 3080 wins.

FAQ

Q: Which GPU has a higher average benchmark score, the AMD Radeon 860M or the NVIDIA GeForce RTX 3080?

A: The AMD Radeon 860M has a higher average benchmark score of 26401, compared to 23172 for the NVIDIA GeForce RTX 3080, though this average is based on different benchmark suites for each card.

Q: How does the Radeon 860M compare to the RTX 3080 in Geekbench Vulkan?

A: The RTX 3080 wins the Geekbench Vulkan test with a score of 33620, while the Radeon 860M scores 30043, a delta of -10.6% for the AMD part.

Q: What is the biggest performance gap between the two GPUs in shared benchmarks?

A: The largest gap is in Geekbench OpenCL, where the RTX 3080 scores 152423 versus 22759 for the Radeon 860M, a delta of -85.1% in favor of NVIDIA.

Q: Does the Radeon 860M have any benchmark wins over the RTX 3080?

A: No, the recorded head-to-head data shows the Radeon 860M with zero wins and the RTX 3080 with two wins.

Q: What are the memory specifications of each GPU?

A: The Radeon 860M uses system shared memory with bandwidth listed as system dependent, while the RTX 3080 has 10 GB of GDDR6X on a 320-bit bus with a bandwidth of 760.3 GB/s.

Q: Which GPU has a higher percentile ranking among all GPUs?

A: The Radeon 860M ranks at the 72nd percentile, which is higher than the RTX 3080's 68th percentile, despite the RTX 3080 winning both direct head-to-head benchmarks.