GPU Comparison

The NVIDIA GeForce RTX 2070 and Intel Arc A770M represent two very different approaches to GPU design, separated by process node and target platform. The data shows a clear split: the RTX 2070, a desktop card from the Turing generation, dominates in legacy and general-purpose workloads, while the Arc A770M, a mobile-first part built on Xe-HPG, takes the lead in modern DirectX 12 and compute APIs. Across the ten head-to-head benchmarks, the RTX 2070 wins seven and the Arc A770M wins three, yet their overall average benchmark scores are remarkably close, with the RTX 2070 at 18789 and the Arc A770M at 18383, a gap of just 2.2%.

Head-to-Head Benchmarks

The most lopsided results come from the older DirectX APIs, where the RTX 2070 is decisively ahead. In Passmark DirectX 10, the RTX 2070 scores 111 against the Arc A770M's 56, a delta of 98.2%, essentially double the performance. DirectX 11 shows a similar story: 129 versus 69, a 87% advantage for the NVIDIA card. Even DirectX 9, a legacy API, favors the RTX 2070 with a score of 211 versus 178, an 18.5% lead. These results indicate that the Turing architecture retains strong driver optimization and hardware scheduling for older graphics pipelines, while Intel's newer part appears to deprioritize these paths.

The general-purpose and compute-oriented Passmark tests also favor the RTX 2070. In Passmark G3D, the NVIDIA card scores 16094 against 11774, a 36.7% victory. Passmark GPU Compute follows with 6411 versus 4778, a 34.2% edge. The 2D performance test, Passmark G2D, goes to the RTX 2070 at 819 versus 711, a 15.2% win. These are not narrow margins; they suggest that in rasterization-heavy and compute-light workloads, the RTX 2070's 2304 shading units and 64 ROPs, combined with its 175 W TDP, translate into higher sustained throughput than the Arc A770M's 4096 shading units and 128 ROPs, which are constrained by a 120 W mobile power envelope.

The Arc A770M's wins are concentrated in newer, more parallel-friendly APIs. The largest is in 3DMark Steel Nomad DX12, where the Intel card scores 2278 against 1569, a 31.1% advantage. This is a significant margin in a modern DirectX 12 test, indicating that the Arc A770M's architecture, with its 256 TMUs and 524.8 GTexel/s texture rate, scales better with the draw calls and async compute typical of DX12 titles. In Geekbench OpenCL, the Arc A770M scores 89494 versus 79966, a 10.6% lead, showing strength in general compute workloads that leverage its 16.79 TFLOPS FP32 throughput. The third win is in Passmark DirectX 12, where the Arc A770M scores 70 against 60, a 14.3% margin, reinforcing the pattern of modern API superiority.

The Vulkan test is the only one where the two cards are close, but it goes to the RTX 2070. Geekbench Vulkan shows the NVIDIA card at 82521 versus 74422, a 10.9% lead. This is notable because it shows the RTX 2070 holding its own in a modern, low-overhead API, countering the narrative that the Arc A770M is universally better in new APIs. The overall picture is one of specialization: the RTX 2070 is a versatile workhorse that excels in established workloads, while the Arc A770M is a forward-looking design that trades legacy compatibility for raw throughput in DX12 and compute.

Where Each One Wins

The RTX 2070 is the clear choice for anyone running DirectX 10 or DirectX 11 applications, where its leads of 98.2% and 87% respectively are insurmountable. It is also the better card for general 3D rendering, as evidenced by the 36.7% margin in Passmark G3D, and for 2D desktop workloads, where the 15.2% win in Passmark G2D indicates faster window composition and image blitting. The 34.2% advantage in Passmark GPU Compute makes it the better option for older compute frameworks or applications that rely on OpenCL or CUDA-like execution patterns, even though the Arc A770M has a higher theoretical FP32 peak. For Vulkan-based titles, the RTX 2070's 10.9% lead in Geekbench Vulkan suggests it can handle modern low-level APIs without compromise.

The Arc A770M wins where the workload is modern and highly parallel. The 31.1% margin in 3DMark Steel Nomad DX12 is the headline result, indicating superior performance in modern DirectX 12 games that utilize mesh shaders and variable rate shading. The 10.6% lead in Geekbench OpenCL shows that its 16.79 TFLOPS FP32 throughput is not just theoretical; it translates to real compute wins in OpenCL-heavy applications like physics simulations or image processing. The Passmark DirectX 12 win, though smaller at 14.3%, confirms that the Arc A770M's architecture is better suited to the resource binding and multi-threaded command submission of DX12. The data also shows that the Arc A770M does not win any test by a narrow margin, its three wins are all decisive, suggesting that when the workload matches its strengths, it is clearly superior.

The Verdict

From the data, the RTX 2070 is the more consistent and broadly capable card. It wins seven of ten head-to-head tests, including all the Passmark DirectX legacy tests, G2D, G3D, and GPU Compute, and it also takes the Vulkan test. Its overall average benchmark score of 18789 is higher than the Arc A770M's 18383, and it sits at the 63rd percentile of all GPUs compared to the Intel card's 62nd. For users with a library of DirectX 9, 10, or 11 games, or who rely on OpenCL compute, the RTX 2070 is the safer pick. Its nearest rivals in the database, the NVIDIA RTX 2000 Ada Generation at 18954 (0.9% ahead) and the AMD Radeon Pro 5700 XT at 18685 (0.6% behind), confirm that it sits in a tight, well-rounded performance band.

The Arc A770M is a specialist. Its wins in 3DMark Steel Nomad, Geekbench OpenCL, and Passmark DirectX 12 show that it is the better choice for modern DirectX 12 games and compute-heavy OpenCL workloads, where its 31.1% and 10.6% leads are substantial. However, its average score of 18383 places it just 0.1% ahead of the AMD Radeon RX 460 and 1.2% ahead of the NVIDIA GeForce RTX 3060 Mobile, indicating that its overall performance is not exceptional outside its niche. The data suggests that the Arc A770M is not a general-purpose replacement for the RTX 2070; it is a part that wins specifically when the software is built for the newest APIs.

The choice is therefore straightforward. If the workload is dominated by DirectX 12 titles or OpenCL compute, the Arc A770M's wins are decisive enough to justify its selection. If the workload is mixed, older, or includes Vulkan, the RTX 2070's seven wins and higher average score make it the better all-rounder. The RTX 2070 also carries a launch MSRP of 499 USD, which is a data point for reference, but the Arc A770M has no listed launch MSRP. Ultimately, the RTX 2070 is the safer default, while the Arc A770M is the pick for those who know their software targets its strengths.

FAQ

Q: Which card wins more head-to-head benchmarks?

A: The NVIDIA GeForce RTX 2070 wins seven of the ten head-to-head tests, while the Intel Arc A770M wins three.

Q: What is the largest performance margin in any single test?

A: The largest margin is in Passmark DirectX 10, where the RTX 2070 scores 111 against the Arc A770M's 56, a 98.2% difference.

Q: Is the Intel Arc A770M better in any modern API?

A: Yes, the Arc A770M wins in 3DMark Steel Nomad DX12 by 31.1%, in Geekbench OpenCL by 10.6%, and in Passmark DirectX 12 by 14.3%.

Q: How do their overall average benchmark scores compare?

A: The RTX 2070 has an average benchmark score of 18789, while the Arc A770M has an average of 18383, making the RTX 2070 about 2.2% higher.

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

A: The RTX 2070 sits at the 63rd percentile, while the Arc A770M sits at the 62nd percentile.

Q: Does the RTX 2070 win in any modern low-level API?

A: Yes, the RTX 2070 wins in Geekbench Vulkan with a score of 82521 versus 74422, a 10.9% lead.

Architecture Differences

The two cards are built on fundamentally different architectures and process nodes. The RTX 2070 uses the TU106 chip on NVIDIA's Turing architecture, fabricated on a 12 nm process at TSMC. It contains 10,800 million transistors on a 445 mm² die, yielding a transistor density of 24.3M per mm². The Arc A770M uses the DG2-512 chip on Intel's Xe-HPG architecture, fabricated on a 6 nm process, also at TSMC. It packs 21,700 million transistors into a 406 mm² die, giving a density of 53.4M per mm², more than double the RTX 2070's density.

The memory subsystems differ in capacity but not bus width. Both cards use a 256-bit memory bus, but the RTX 2070 has 8 GB of GDDR6 at 1750 MHz (14 Gbps effective), yielding 448.0 GB/s bandwidth. The Arc A770M has 16 GB of GDDR6 at 2000 MHz (16 Gbps effective), yielding 512.0 GB/s. The Intel card has double the shading units (4096 versus 2304), double the TMUs (256 versus 144), and double the ROPs (128 versus 64). The RTX 2070 has 36 RT cores and 288 tensor cores, while the Arc A770M has 32 RT cores and no listed tensor cores. The Intel card has higher pixel and texture rates: 262.4 GPixel/s and 524.8 GTexel/s versus 103.7 GPixel/s and 233.3 GTexel/s. FP32 throughput is 16.79 TFLOPS for the Arc A770M versus 7.465 TFLOPS for the RTX 2070, and FP16 is 33.59 TFLOPS versus 14.93 TFLOPS, both at 2:1 ratios.

The cards also differ in their display output and interface. The RTX 2070 offers 1x DVI, 1x HDMI 2.0, 2x DisplayPort 1.4a, and 1x USB Type-C, while the Arc A770M's outputs are listed as "Portable Device Dependent." The RTX 2070 uses a PCIe 3.0 x16 interface, while the Arc A770M uses PCIe 4.0 x16. Both support DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4.

Specification Differences

The most direct specification differences between the two cards are in their clock speeds and power characteristics. The RTX 2070 has a base clock of 1410 MHz and a boost clock of 1620 MHz, while the Arc A770M has a base clock of 1650 MHz and a boost clock of 2050 MHz. The TDP differs significantly: the RTX 2070 is rated at 175 W, while the Arc A770M is rated at 120 W. The RTX 2070 is a dual-slot card with a 1x 8-pin power connector and a suggested PSU of 450 W, while the Arc A770M is an IGP (integrated graphics processor) with no power connector or suggested PSU listed.

Physical dimensions are only provided for the RTX 2070: 229 mm in length, 113 mm in height, and 35 mm in width. The Arc A770M has no listed dimensions. The RTX 2070 has a release date of 2018-10-16 and a launch MSRP of 499 USD, while the Arc A770M has no release date or launch MSRP. The RTX 2070 has a predecessor (GeForce 10) and a successor (GeForce 30), while the Arc A770M has neither listed. Both cards are marked as end-of-life in production status. The RTX 2070's memory clock is 1750 MHz (14 Gbps effective) versus the Arc A770M's 2000 MHz (16 Gbps effective). The RTX 2070's transistor count is 10,800 million versus 21,700 million for the Arc A770M, and its die size is 445 mm² versus 406 mm².