Intel Arc A380 vs NVIDIA GeForce GTX 970 Comparison

FAQ

Q: How does the Intel Arc A380 compare to the NVIDIA GeForce GTX 970 in overall average benchmark score?

A: The Arc A380 records an average benchmark score of 8558, while the GTX 970 sits at 7157. That places the Intel card about 19.6% higher on average, and its percentile ranking among all GPUs is 44 versus 39 for the NVIDIA card.

Q: Which card wins the most head-to-head benchmark tests?

A: The GTX 970 wins 7 of the 10 recorded head-to-head tests, while the Arc A380 wins 3. However, the Intel card's wins come in modern API and compute workloads, including a 119% lead in 3DMark Steel Nomad DX12.

Q: What is the biggest single benchmark margin between these two cards?

A: In 3DMark Steel Nomad DX12, the Arc A380 scores 808 versus 369 for the GTX 970, a delta of 119%. The largest NVIDIA win is in Passmark DirectX 9, where the GTX 970 scores 143 versus 73, a 49% advantage.

Q: How do the memory subsystems differ in capacity and bandwidth?

A: The Arc A380 has 6 GB of GDDR6 on a 96-bit bus, yielding 186.0 GB/s of bandwidth. The GTX 970 has 4 GB of GDDR5 on a 256-bit bus, delivering 224.4 GB/s. Despite having less VRAM, the NVIDIA card has higher memory bandwidth.

Q: Which card has a higher transistor density despite being on a smaller process?

A: The Arc A380 is fabricated on TSMC's 6 nm process and packs 7,200 million transistors into a 157 mm² die, giving a density of 45.9M transistors per mm². The GTX 970 uses TSMC's 28 nm process with 5,200 million transistors on a 398 mm² die, for a density of 13.1M per mm².

Q: What is the difference in power connector requirements?

A: The Arc A380 uses a single 8-pin power connector with a 75 W TDP and a suggested 250 W PSU. The GTX 970 requires two 6-pin connectors, has a 148 W TDP, and a suggested 300 W PSU.

Architecture Differences

The two cards represent fundamentally different eras of GPU design. The Intel Arc A380 is built on the Xe-HPG architecture, fabricated at TSMC's 6 nm node. It uses the DG2-128 chip and belongs to the Alchemist generation, specifically the Arc 3 family. The NVIDIA GeForce GTX 970, in contrast, uses the Maxwell 2.0 architecture on TSMC's 28 nm process, with the GM204 chip from the GeForce 900 series.

The most striking architectural divergence is in feature support. The Arc A380 includes 8 dedicated ray tracing cores, a capability entirely absent from the GTX 970. The Intel card also supports DirectX 12 Ultimate (12_2), whereas the NVIDIA card is limited to DirectX 12 (12_1). Both cards support OpenGL 4.6 and Vulkan 1.4, but the Intel card's API set reflects a more modern feature baseline.

Transistor counts and die sizes tell a story of process technology progression. The Arc A380 integrates 7,200 million transistors into a 157 mm² die, while the GTX 970 manages 5,200 million transistors across a much larger 398 mm² die. This translates to a transistor density of 45.9M per mm² for Intel versus 13.1M per mm² for NVIDIA, a 3.5x density advantage enabled by the 6 nm versus 28 nm process gap.

Clock behavior also differs significantly. The Arc A380 runs a base clock of 2000 MHz with a boost of 2050 MHz, while the GTX 970 operates at 1050 MHz base and 1178 MHz boost. Higher clocks on the Intel card partially compensate for its smaller shader array. The Arc A380 has 1024 shading units, 64 TMUs, and 32 ROPs, while the GTX 970 fields 1664 shading units, 104 TMUs, and 56 ROPs. The NVIDIA card's larger execution resource pool helps it maintain leadership in several rasterization-focused tests.

Memory architecture further separates the two. The Arc A380 uses 6 GB of GDDR6 across a 96-bit bus for 186.0 GB/s bandwidth. The GTX 970 uses 4 GB of GDDR5 across a 256-bit bus for 224.4 GB/s. The NVIDIA card's wider bus provides more bandwidth despite slower memory clocks (1753 MHz versus 1937 MHz for the Intel card, with 7 Gbps effective versus 15.5 Gbps effective). The GTX 970 also offers a higher pixel rate at 65.97 GPixel/s versus 65.60 GPixel/s, though the Intel card counters with a higher texture rate of 131.2 GTexel/s versus 122.5 GTexel/s.

The bus interface differs as well: the Arc A380 uses PCIe 4.0 x8, while the GTX 970 uses PCIe 3.0 x16. Display outputs favor the Intel card with 1x HDMI 2.1 and 3x DisplayPort 2.0, versus the GTX 970's 1x DVI, 1x HDMI 2.0, and 3x DisplayPort 1.2.

Head-to-Head Benchmarks

The 3DMark Steel Nomad DX12 test delivers the most dramatic divergence. The Arc A380 scores 808 against the GTX 970's 369, a 119% advantage. This test likely leverages the Intel card's DirectX 12 Ultimate feature set and ray tracing hardware, revealing a generational gap in modern API efficiency.

Compute workloads also favor the Intel card. In Geekbench OpenCL, the Arc A380 scores 38224 versus 29982, a 27.5% lead. Geekbench Vulkan shows an even larger gap: 36736 versus 20005, an 83.6% advantage. These results suggest the Xe-HPG architecture's compute throughput, particularly in Vulkan, substantially outperforms Maxwell 2.0 in compute-oriented tasks.

The GTX 970, however, dominates in legacy DirectX tests. Passmark DirectX 10 shows the NVIDIA card ahead at 46 versus 37, a 19.6% margin. DirectX 11 widens the gap to 46.5% (71 versus 38), and DirectX 9 shows the largest NVIDIA win at 49% (143 versus 73). DirectX 12 in Passmark also goes to NVIDIA, but narrowly: 41 versus 35, a 14.6% margin. This pattern suggests driver maturity and architecture efficiency for older API paths remain NVIDIA strengths.

In overall 3D performance, Passmark G3D gives the GTX 970 a decisive 9638 score versus 6252 for the Arc A380, a 35.1% advantage. Passmark G2D also favors NVIDIA at 764 versus 610, a 20.2% margin. Passmark GPU Compute follows suit with 4073 versus 2762, a 32.2% lead for the GTX 970.

The wins break down unevenly: Intel takes 3 tests (3DMark Steel Nomad, Geekbench OpenCL, Geekbench Vulkan), while NVIDIA takes 7 (all Passmark tests). The average benchmark score, however, favors Intel at 8558 versus 7157, suggesting the compute-heavy Geekbench results carry significant weight in the aggregate metric.

The Verdict

The data presents a split personality. The Intel Arc A380 is the clear winner in modern API and compute workloads, with its 119% lead in 3DMark Steel Nomad DX12 being the single largest margin across all tests. Its Vulkan performance at 83.6% above the GTX 970 indicates superior compute architecture for contemporary workloads.

The NVIDIA GeForce GTX 970 remains dominant in legacy DirectX paths and traditional rasterization. Its Passmark G3D score of 9638 versus 6252 for the Arc A380 demonstrates a 35.1% advantage in general 3D rendering tests. The GTX 970 also holds leads in DirectX 9 through 12 in Passmark tests, with margins ranging from 14.6% to 49%.

For users running older titles or applications that rely on DirectX 9, 10, or 11, the GTX 970's 49%, 19.6%, and 46.5% wins respectively make it the more practical choice. The NVIDIA card also wins Passmark G2D and GPU Compute tests by 20.2% and 32.2%, respectively.

For those working with modern APIs, particularly Vulkan or DirectX 12 Ultimate workloads, the Arc A380's advantages are compelling. The 83.6% Vulkan lead and 119% Steel Nomad advantage indicate the Intel card is built for current-generation software. Its 6 GB VRAM capacity also provides more headroom for modern textures than the GTX 970's 4 GB.

The percentile rankings place the Arc A380 at 44 versus 39 for the GTX 970, meaning the Intel card sits above a larger fraction of all GPUs. The average benchmark score of 8558 for Intel versus 7157 for NVIDIA reinforces this positioning. However, the GTX 970's seven wins in head-to-head tests show it remains highly competitive in specific workloads.

Specification Differences

| Specification | Intel Arc A380 | NVIDIA GeForce GTX 970 |

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

| Architecture | Xe-HPG | Maxwell 2.0 |

| Process Node | 6 nm | 28 nm |

| Transistors | 7,200 million | 5,200 million |

| Die Size | 157 mm² | 398 mm² |

| Transistor Density | 45.9M / mm² | 13.1M / mm² |

| Base Clock | 2000 MHz | 1050 MHz |

| Boost Clock | 2050 MHz | 1178 MHz |

| Memory Size | 6 GB | 4 GB |

| Memory Type | GDDR6 | GDDR5 |

| Memory Bus | 96 bit | 256 bit |

| Memory Bandwidth | 186.0 GB/s | 224.4 GB/s |

| Shading Units | 1024 | 1664 |

| TMUs | 64 | 104 |

| ROPs | 32 | 56 |

| RT Cores | 8 | None |

| Pixel Rate | 65.60 GPixel/s | 65.97 GPixel/s |

| Texture Rate | 131.2 GTexel/s | 122.5 GTexel/s |

| FP32 | 4.198 TFLOPS | 3.920 TFLOPS |

| FP16 | 8.397 TFLOPS (2:1) | None |

| TDP | 75 W | 148 W |

| Power Connectors | 1x 8-pin | 2x 6-pin |

| Suggested PSU | 250 W | 300 W |

| Bus Interface | PCIe 4.0 x8 | PCIe 3.0 x16 |

| Display Outputs | 1x HDMI 2.1, 3x DisplayPort 2.0 | 1x DVI, 1x HDMI 2.0, 3x DisplayPort 1.2 |

| DirectX Support | 12 Ultimate (12_2) | 12 (12_1) |

Where Each One Wins

The Intel Arc A380 wins decisively in modern API compute tasks. The 3DMark Steel Nomad DX12 test shows a 119% advantage, indicating the card's DirectX 12 Ultimate implementation and ray tracing cores deliver substantial performance in next-generation workloads. Geekbench Vulkan results reinforce this, with an 83.6% lead that points to strong driver optimization for the Vulkan API. OpenCL performance follows at 27.5% ahead, suggesting the Xe-HPG compute architecture scales well across different compute frameworks.

The GTX 970 wins across the legacy DirectX spectrum. Its 49% advantage in DirectX 9, 46.5% in DirectX 11, and 19.6% in DirectX 10 demonstrate mature driver support for older APIs. The Passmark DirectX 12 test also goes to NVIDIA, albeit narrowly at 14.6%, which is notable given the Arc A380's architectural focus on modern APIs.

For general 3D performance, the GTX 970's 35.1% Passmark G3D lead shows it handles traditional rasterization workloads more effectively. The 2D performance gap of 20.2% in Passmark G2D further favors NVIDIA. Compute performance in Passmark GPU Compute also goes to the GTX 970 by 32.2%, though this contrasts with the Intel card's Geekbench compute wins, suggesting different test methodologies capture different aspects of compute performance.

Users should choose based on software priorities. The Arc A380 suits Vulkan-heavy applications, DirectX 12 Ultimate titles, and workloads leveraging ray tracing or FP16 compute. The GTX 970 suits DirectX 9-11 legacy games, 2D-heavy desktop tasks, and general 3D rendering where its higher shading unit count and memory bandwidth provide advantages.