Intel Arc B570 vs NVIDIA GeForce RTX 2070 Comparison

FAQ

Q: Which GPU wins in the 3DMark Steel Nomad DX12 benchmark?

A: The Intel Arc B570 wins decisively with a score of 2649 versus the NVIDIA GeForce RTX 2070's 1569, a delta of 68.8%. This is the largest performance gap in the head-to-head comparison.

Q: How does the RTX 2070 perform in DirectX 10 and DirectX 9 legacy tests?

A: The RTX 2070 leads in both. It scores 111 in Passmark DirectX 10 (vs. 65 for the Arc B570, a -41.4% delta) and 211 in Passmark DirectX 9 (vs. 164, a -22.3% delta). This suggests NVIDIA retains an edge in older API workloads.

Q: What is the average benchmark score difference between the two cards?

A: The Intel Arc B570 has an average benchmark score of 20,556, while the RTX 2070 sits at 18,789. The Arc B570's overall average is approximately 9.4% higher, placing it in the 65th percentile of all GPUs versus the RTX 2070's 63rd percentile.

Q: Which GPU has more ray tracing cores?

A: The NVIDIA GeForce RTX 2070 has 36 RT cores, double the 18 RT cores found in the Intel Arc B570. However, this architectural advantage does not translate into a win in the modern DX12 benchmark included in the data.

Q: What are the memory configurations of the two cards?

A: The Intel Arc B570 has 10 GB of GDDR6 memory on a 160-bit bus with 380.0 GB/s bandwidth. The RTX 2070 has 8 GB of GDDR6 on a 256-bit bus with 448.0 GB/s bandwidth. The RTX 2070 has higher bandwidth, but the Arc B570 has more capacity.

Q: Which GPU is still in active production?

A: The Intel Arc B570 is listed as "Active" production status, released on 2025-01-15. The NVIDIA GeForce RTX 2070 is marked as "End-of-life" production status, having been released on 2018-10-16.

Architecture Differences

The two GPUs represent fundamentally different design generations and manufacturing approaches. The Intel Arc B570 is built on TSMC's 5 nm process node, packing 19,600 million transistors into a 272 mm² die. This yields a transistor density of 72.1M per mm². By contrast, the NVIDIA GeForce RTX 2070 uses TSMC's 12 nm process, containing 10,800 million transistors on a much larger 445 mm² die, with a density of only 24.3M per mm². The Arc B570's architecture, Xe2-HPG from the Battlemage generation, achieves nearly three times the transistor density of the older Turing architecture in the RTX 2070.

Clock speeds diverge sharply. The Arc B570 runs at a fixed 2500 MHz for both base and boost, while the RTX 2070 has a 1410 MHz base clock and 1620 MHz boost. Memory clocks also differ: the Arc B570 operates at 2375 MHz (19 Gbps effective), while the RTX 2070 runs at 1750 MHz (14 Gbps effective). Despite the RTX 2070's wider 256-bit memory bus versus the Arc B570's 160-bit bus, the Intel card's higher memory clock partially compensates, though NVIDIA retains the bandwidth lead at 448.0 GB/s versus 380.0 GB/s.

Shader and texture resources are identical in count, with both GPUs featuring 2304 shading units and 144 texture mapping units. However, the RTX 2070 has 36 RT cores and 288 tensor cores, while the Arc B570 has 18 RT cores and no tensor core count listed. The Intel card has more ROPs: 80 versus 64 on the NVIDIA card. This ROP advantage contributes to the Arc B570's higher pixel rate of 200.0 GPixel/s, nearly double the RTX 2070's 103.7 GPixel/s. The texture rate also favors Intel at 360.0 GTexel/s versus 233.3 GTexel/s.

Compute throughput tells a similar story. The Arc B570 delivers 11.52 TFLOPS FP32 and 23.04 TFLOPS FP16, while the RTX 2070 manages 7.465 TFLOPS FP32 and 14.93 TFLOPS FP16. The Intel card is roughly 54% ahead in raw FP32 compute. Interface and connectivity also differ: the Arc B570 uses PCIe 4.0 x8, while the RTX 2070 uses PCIe 3.0 x16. Display outputs favor Intel with 1x HDMI 2.1a and 3x DisplayPort 2.1, versus NVIDIA's older 1x DVI, 1x HDMI 2.0, 2x DisplayPort 1.4a, and 1x USB Type-C.

Head-to-Head Benchmarks

The 3DMark Steel Nomad DX12 test delivers the most striking contrast. The Intel Arc B570 scores 2649, while the RTX 2070 trails at 1569. This 68.8% delta indicates the Arc B570's modern architecture is far more efficient in this contemporary DirectX 12 workload, likely benefiting from the newer Xe2-HPG design and higher clock speeds.

In compute-oriented benchmarks, Intel maintains a consistent lead. The Geekbench Vulkan test shows the Arc B570 at 96,844 versus 82,521 for the RTX 2070, a 17.4% advantage. Geekbench OpenCL results are closer: 83,514 for Intel versus 79,966 for NVIDIA, a modest 4.4% delta. The Passmark GPU Compute test also favors Intel at 7,281 versus 6,411, a 13.6% win. These results indicate the Arc B570's higher FP32 and FP16 throughput translates into real compute performance gains.

The Passmark DirectX 12 test is another Intel victory, with a score of 72 versus 60, a 20% delta. This aligns with the 3DMark result, suggesting the Arc B570 handles modern graphics APIs more effectively. However, the RTX 2070 fights back in legacy and 2D workloads. The Passmark DirectX 10 test shows NVIDIA winning 111 to 65, a -41.4% delta in Intel's favor. DirectX 11 also goes to NVIDIA: 129 versus 118, an -8.5% delta. The DirectX 9 test continues this trend, with NVIDIA ahead at 211 versus 164, a -22.3% delta.

The Passmark G3D test, a general 3D performance metric, goes to NVIDIA at 16,094 versus 14,195 for Intel, an -11.8% delta. The Passmark G2D test also favors NVIDIA: 819 versus 661, a -19.3% delta. The head-to-head record ends exactly tied at 5 wins apiece, but the magnitude of Intel's wins in modern workloads is substantially larger than NVIDIA's wins in legacy tests. The 3DMark Steel Nomad delta of 68.8% dwarfs the RTX 2070's largest victory, which is the 41.4% delta in DirectX 10.

Specification Differences

The two cards differ across nearly every specification category. Process node: 5 nm (Intel) versus 12 nm (NVIDIA). Transistor count: 19,600 million versus 10,800 million. Die size: 272 mm² versus 445 mm². Transistor density: 72.1M / mm² versus 24.3M / mm². Base clock: 2500 MHz versus 1410 MHz. Boost clock: 2500 MHz versus 1620 MHz. Memory clock: 2375 MHz versus 1750 MHz. Memory size: 10 GB versus 8 GB. Memory bus width: 160 bit versus 256 bit. Memory bandwidth: 380.0 GB/s versus 448.0 GB/s.

ROPs: 80 versus 64. RT cores: 18 versus 36. Tensor cores: not listed versus 288. Pixel rate: 200.0 GPixel/s versus 103.7 GPixel/s. Texture rate: 360.0 GTexel/s versus 233.3 GTexel/s. FP32: 11.52 TFLOPS versus 7.465 TFLOPS. FP16: 23.04 TFLOPS versus 14.93 TFLOPS. TDP: 150 W versus 175 W. Power connectors: both use 1x 8-pin, but suggested PSU is identical at 450 W. Bus interface: PCIe 4.0 x8 versus PCIe 3.0 x16. Display outputs: 1x HDMI 2.1a, 3x DisplayPort 2.1 versus 1x DVI, 1x HDMI 2.0, 2x DisplayPort 1.4a, 1x USB Type-C. Dimensions: 272 mm length versus 229 mm length; 115 mm height versus 113 mm height; the RTX 2070 has a listed width of 35 mm while the Arc B570 has none listed.

Production status: Active versus End-of-life. Release date: 2025-01-15 versus 2018-10-16. Predecessor: Alchemist versus GeForce 10. Successor: none listed versus GeForce 30. Launch MSRP: 219 USD for the Arc B570, 499 USD for the RTX 2070. Shading units and TMUs are identical at 2304 and 144, respectively. Both support DirectX 12 Ultimate, OpenGL 4.6, and Vulkan 1.4.

The Verdict

The data presents a clear generational shift. The Intel Arc B570 wins in 3DMark Steel Nomad, Geekbench Vulkan, Geekbench OpenCL, Passmark DirectX 12, and Passmark GPU Compute. Its average benchmark score of 20,556 places it in the 65th percentile of all GPUs, compared to the RTX 2070's 18,789 and 63rd percentile. The Arc B570's nearest rival, the NVIDIA GeForce RTX 3070 Mobile, scores 20,534 with a delta of just 0.1%, suggesting the Intel card sits in competitive territory with a much newer mobile flagship. The RTX 2070's nearest rival, the NVIDIA RTX 2000 Ada Generation, scores 18,954, a -0.9% delta, indicating the aging card still holds its own against a modern workstation part.

For users prioritizing modern DirectX 12 gaming and compute workloads, the Intel Arc B570 is the clear choice. Its 68.8% lead in 3DMark Steel Nomad and 17.4% lead in Vulkan demonstrate superior execution of contemporary APIs. The 20% win in Passmark DirectX 12 reinforces this. The card also offers more VRAM at 10 GB versus 8 GB, which matters for modern textures and resolutions.

The RTX 2070 retains relevance only in legacy DirectX 9, 10, and 11 scenarios, where it leads by margins ranging from 8.5% to 41.4%. Its higher memory bandwidth of 448.0 GB/s and 36 RT cores do not translate into wins in the modern benchmarks available. The RTX 2070's Passmark G3D win at 16,094 versus 14,195 is notable, but this is offset by the Arc B570's superior compute scores.

The verdict is straightforward. The Intel Arc B570 is the superior modern GPU, with higher average scores, better efficiency on a 5 nm process, and wins in all contemporary benchmarks. The RTX 2070 is an end-of-life product that only excels in older API tests. Users running legacy applications may find the NVIDIA card acceptable, but for any current or future workload, the Arc B570's data-driven advantages make it the rational selection. The RTX 2070's launch MSRP of 499 USD versus the Arc B570's 219 USD further underscores the generational leap, though pricing should not be the sole consideration given the performance disparities.