GPU Comparison

The Intel Arc A750 and NVIDIA GeForce RTX 2070 SUPER represent two very different approaches to the same performance class, separated by three years of GPU architecture evolution. On average benchmark score, the Arc A750 edges ahead at 20582 versus 20282 for the RTX 2070 SUPER, a margin of roughly 1.5%. Both cards sit in the 65th-66th percentile of all GPUs, firmly in mid-range territory. The head-to-head results, however, tell a more nuanced story: the RTX 2070 SUPER wins 7 of 10 benchmarks, but the Arc A750 takes the most demanding modern test by a landslide. This is not a simple "one is faster" situation, it depends entirely on which workloads you prioritize.

Head-to-Head Benchmarks

The biggest single victory belongs to the Intel Arc A750 in 3DMark Steel Nomad DX12. Here, the Arc scores 2612 against the RTX 2070 SUPER's 1651, a commanding 58.2% lead. This is a massive gap in a modern DirectX 12 test, suggesting the Arc's architecture is far better optimized for current-generation rendering workloads. The Arc also wins Geekbench OpenCL with 98554 versus 83358, an 18.2% advantage that points to strong general-purpose compute performance. In Passmark DirectX 12, the Arc takes a narrower 4.5% win (70 vs 67), barely edging out the NVIDIA card in the latest API version.

The RTX 2070 SUPER dominates the older and legacy benchmarks. In Passmark DirectX 11, it scores 151 versus the Arc's 72, a 52.3% blowout. DirectX 10 shows a similar story: 132 vs 65, a 50.8% lead for NVIDIA. Passmark DirectX 9 gives the RTX 2070 SUPER a 223 to 181 win, an 18.8% margin. These older API results suggest NVIDIA's driver maturity and architecture have a significant advantage in legacy titles. The RTX 2070 SUPER also wins Passmark G3D (18169 vs 12534, a 31% lead), Passmark GPU Compute (7557 vs 5368, a 29% lead), and Passmark G2D (878 vs 732, a 16.6% lead). In Geekbench Vulkan, the NVIDIA card wins 90637 to 85631, a 5.5% margin that shows its strength in Vulkan workloads.

Architecture Differences

The fundamental architectural split is stark. The Intel Arc A750 uses the DG2-512 chip built on Xe-HPG architecture at TSMC's 6 nm process, packing 21,700 million transistors into a 406 mm² die. The NVIDIA GeForce RTX 2070 SUPER uses the TU104 chip with Turing architecture on a 12 nm TSMC process, containing 13,600 million transistors on a larger 545 mm² die. This translates to a transistor density of 53.4M per mm² for Intel versus just 25.0M per mm² for NVIDIA, the Arc packs more than twice the transistor density.

The compute resources differ dramatically. The Arc A750 has 3584 shading units, 224 texture mapping units, and 112 ROPs, while the RTX 2070 SUPER has 2560 shading units, 160 TMUs, and only 64 ROPs. The Arc's pixel rate of 268.8 GPixel/s and texture rate of 537.6 GTexel/s far exceed the RTX 2070 SUPER's 113.3 GPixel/s and 283.2 GTexel/s. In raw FP32 throughput, the Arc hits 17.20 TFLOPS versus 9.062 TFLOPS for the NVIDIA card, a 90% advantage. FP16 performance shows 34.41 TFLOPS for the Arc versus 18.12 TFLOPS for the RTX 2070 SUPER.

Ray tracing and tensor hardware tell a different story. The RTX 2070 SUPER has 40 RT cores and 320 tensor cores, while the Arc A750 has 28 RT cores and no tensor cores listed. This gives NVIDIA a structural advantage in ray-traced workloads and any tensor-based features. Clock speeds also differ: the Arc runs at a 2050 MHz base and 2400 MHz boost, while the RTX 2070 SUPER runs at 1605 MHz base and 1770 MHz boost. Memory configurations are similar in capacity (8 GB GDDR6 on both) and bus width (256 bit on both), but the Arc's memory runs at 2000 MHz with 16 Gbps effective speed for 512.0 GB/s bandwidth, versus 1750 MHz at 14 Gbps effective for 448.0 GB/s on the RTX 2070 SUPER.

The Verdict

The data supports a clear split: the Intel Arc A750 is the modern-performance pick, while the NVIDIA GeForce RTX 2070 SUPER is the legacy-compatibility and driver-maturity choice. If your workload is primarily current-generation DirectX 12 titles or compute-heavy OpenCL tasks, the Arc A750's 58.2% lead in 3DMark Steel Nomad and 18.2% lead in Geekbench OpenCL make it the stronger card. The Arc also has superior raw specifications: more shading units, higher clock speeds, more memory bandwidth, and nearly double the FP32 throughput.

However, the RTX 2070 SUPER's wins in DirectX 9, 10, and 11 by margins of 18.8% to 52.3% indicate that older games and applications will run significantly better on NVIDIA's hardware. Its 31% lead in Passmark G3D and 29% lead in GPU compute further suggests that general-purpose GPU compute tasks and DirectX 11-era games favor the Turing architecture. The RTX 2070 SUPER also wins in Vulkan by 5.5%, which matters for Vulkan-based titles and emulators.

Specification Differences

| Specification | Intel Arc A750 | NVIDIA GeForce RTX 2070 SUPER |

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

| Process Node | 6 nm | 12 nm |

| Transistors | 21,700 million | 13,600 million |

| Die Size | 406 mm² | 545 mm² |

| Transistor Density | 53.4M / mm² | 25.0M / mm² |

| Base Clock | 2050 MHz | 1605 MHz |

| Boost Clock | 2400 MHz | 1770 MHz |

| Memory Clock | 2000 MHz / 16 Gbps effective | 1750 MHz / 14 Gbps effective |

| Memory Bandwidth | 512.0 GB/s | 448.0 GB/s |

| Shading Units | 3584 | 2560 |

| TMUs | 224 | 160 |

| ROPs | 112 | 64 |

| RT Cores | 28 | 40 |

| Tensor Cores | N/A | 320 |

| Pixel Rate | 268.8 GPixel/s | 113.3 GPixel/s |

| Texture Rate | 537.6 GTexel/s | 283.2 GTexel/s |

| FP32 | 17.20 TFLOPS | 9.062 TFLOPS |

| FP16 | 34.41 TFLOPS (2:1) | 18.12 TFLOPS (2:1) |

| TDP | 225 W | 215 W |

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

| Display Outputs | 1x HDMI 2.1, 3x DisplayPort 2.0 | 1x HDMI 2.0, 3x DisplayPort 1.4a, 1x USB Type-C |

| Dimensions | N/A | 267 mm / 10.5 inches long, 116 mm / 4.6 inches high, 35 mm / 1.4 inches wide |

| Release Date | 2022-10-11 | 2019-07-08 |

| Launch MSRP | 289 USD | 499 USD |

FAQ

Q: Which card has higher raw compute performance?

A: The Intel Arc A750 has significantly higher raw compute: 17.20 TFLOPS FP32 versus 9.062 TFLOPS for the RTX 2070 SUPER, and 34.41 TFLOPS FP16 versus 18.12 TFLOPS. This is reflected in the Arc's 18.2% Geekbench OpenCL win.

Q: Why does the RTX 2070 SUPER win so many Passmark tests?

A: The RTX 2070 SUPER wins Passmark DirectX 9, 10, and 11 by margins of 18.8%, 50.8%, and 52.3% respectively. This indicates NVIDIA's Turing architecture has much stronger support for legacy DirectX APIs, likely due to driver maturity and hardware design optimized for those older interfaces.

Q: Which card handles modern DirectX 12 games better?

A: The Intel Arc A750 wins 3DMark Steel Nomad DX12 by 58.2% (2612 vs 1651) and Passmark DirectX 12 by 4.5% (70 vs 67). The Arc's architecture appears substantially better optimized for current DirectX 12 workloads.

Q: Do both cards have the same memory configuration?

A: Both have 8 GB GDDR6 on a 256-bit bus, but the Arc A750 runs faster memory at 2000 MHz (16 Gbps effective) yielding 512.0 GB/s bandwidth, versus 1750 MHz (14 Gbps effective) for 448.0 GB/s on the RTX 2070 SUPER.

Q: Which card has more ray tracing hardware?

A: The RTX 2070 SUPER has 40 RT cores and 320 tensor cores, while the Arc A750 has 28 RT cores and no tensor cores listed. NVIDIA has a 43% advantage in RT core count.

Q: What is the average benchmark score difference?

A: The Arc A750 averages 20582 versus 20282 for the RTX 2070 SUPER, a 1.5% difference. Both cards sit in the 65th-66th percentile of all GPUs, placing them in the same performance tier overall.

Where Each One Wins

Intel Arc A750 wins for: Modern DirectX 12 gaming, where its 58.2% 3DMark Steel Nomad lead is decisive. OpenCL compute workloads, with an 18.2% advantage in Geekbench OpenCL. Any application that leverages its higher FP32 or FP16 throughput, since it has roughly 90% more raw compute than the RTX 2070 SUPER. Users who want PCIe 4.0 x16 support and DisplayPort 2.0 outputs, which the NVIDIA card lacks.

NVIDIA GeForce RTX 2070 SUPER wins for: Legacy DirectX gaming, especially DirectX 11 (52.3% lead) and DirectX 10 (50.8% lead). Vulkan workloads, with a 5.5% edge in Geekbench Vulkan. General GPU compute as measured by Passmark (29% lead in GPU compute). Overall 3D rendering in the Passmark G3D test (31% lead). Ray-traced workloads, given its 40 RT cores versus 28. Tensor-based features, thanks to 320 tensor cores that the Arc lacks entirely. Users who need a USB Type-C display output or value a physically smaller card, the RTX 2070 SUPER measures 267 mm long, 116 mm high, and 35 mm wide.