Intel Arc B390 vs NVIDIA GeForce RTX 4070 GDDR6 Comparison

The Intel Arc B390 and NVIDIA GeForce RTX 4070 GDDR6 share one recorded head-to-head benchmark in the database. The RTX 4070 GDDR6 posts 4334.5 points in 3DMark Steel Nomad DX12, and the Arc B390 posts 1482 points. The delta is -65.8 percent, meaning the Intel part trails by that margin in the only direct comparison available.

Head-to-Head Benchmarks

The head-to-head record contains a single test: 3dmark_3dmark_steel_nomad_dx12. The winner is the NVIDIA GeForce RTX 4070 GDDR6, with 4334.5 points against 1482 points for the Intel Arc B390. The database records a delta of -65.8 percent, and the win count is 1 for the NVIDIA part and 0 for the Intel part.

The average benchmark scores follow the same order. The RTX 4070 GDDR6 has an average benchmark score of 4335, while the Arc B390 has an average of 1482. Percentile placement also separates them: the RTX 4070 GDDR6 sits at the 25th percentile of all GPUs, and the Arc B390 sits at the 9th percentile.

Nearest-rival data gives context for the Arc B390's result. The Arc B390 is 1.3% above the NVIDIA GeForce GT 520MX, 1.5% above the GeForce 800M, 2.5% above the GeForce GT 625 OEM, and 2.7% above the GeForce GT 710. All four margins are small, so the Arc B390's score is closely grouped with those entries in this benchmark.

For the RTX 4070 GDDR6, the nearest-rival cluster is also tight. The NVIDIA part is 0.6% below the Intel Iris Pro 5200, 0.9% above the AMD FirePro W2100, 1.2% below the GeForce 930M, and 1.2% above the GeForce GTX 460M. The score of 4334.5 therefore lands between 4282 and 4388 in the nearest-rival group.

The Arc B390's boost clock of 2500 MHz is higher than the RTX 4070 GDDR6's 2475 MHz, yet its measured score is 1482 against 4334.5. That suggests clock rate alone does not explain the performance gap; resource counts and memory configuration differ sharply.

Where Each One Wins

The benchmark split is one-sided. The RTX 4070 GDDR6 wins the only head-to-head test, with a -65.8 percent delta against the Arc B390. The Arc B390 has zero benchmark wins in the database. The data does not support any direct performance win for the Intel part.

Behind the RTX 4070 GDDR6's win are larger resource counts. It has 5888 shading units, 184 TMUs, 64 ROPs, 46 ray tracing cores, and 184 tensor cores. Its FP32 throughput is 29.15 TFLOPS, its texture rate is 455.4 GTexel/s, and its pixel rate is 158.4 GPixel/s. The Arc B390 has 1536 shading units, 48 TMUs, 24 ROPs, and 12 ray tracing cores, with 7.680 TFLOPS FP32, 120.0 GTexel/s texture rate, and 60.00 GPixel/s pixel rate. These counts align with the benchmark ordering.

The Arc B390's advantages are structural. It has an 80W TDP, an IGP slot width, no power connectors, and an IGP bus interface. It uses system-shared memory with system-dependent bandwidth. The RTX 4070 GDDR6 has a 200W TDP, a dual-slot width, one 16-pin power connector, a PCIe 4.0 x16 interface, and 12 GB of GDDR6. For a use case that can accept a dual-slot card with external power, the RTX 4070 GDDR6 is the benchmark winner. For a use case that must use an IGP configuration, the Arc B390 is the only one of the two that fits.

The Verdict

Strictly from the recorded data, the RTX 4070 GDDR6 is the higher-performance part in this comparison. It wins the only benchmark, has the higher average benchmark score (4335 versus 1482), and sits at the higher percentile (25 versus 9).

The Arc B390 belongs in a different physical class. It is an IGP part with an 80W TDP, no power connectors, and system-shared memory. The RTX 4070 GDDR6 is a dual-slot card with a 200W TDP, one 16-pin power connector, and a suggested PSU of 550W. The data implies that a system requiring the highest 3DMark Steel Nomad DX12 score should use the RTX 4070 GDDR6, while a system requiring an IGP configuration without external power should use the Arc B390.

FAQ

Q: What is the only head-to-head benchmark result?

A: The only head-to-head test is 3DMark Steel Nomad DX12. The RTX 4070 GDDR6 scores 4334.5, the Arc B390 scores 1482, and the delta is -65.8 percent.

Q: How does the Arc B390 compare with its nearest database rivals?

A: The Arc B390 is 1.3% above the GeForce GT 520MX, 1.5% above the GeForce 800M, 2.5% above the GeForce GT 625 OEM, and 2.7% above the GeForce GT 710.

Q: How does the RTX 4070 GDDR6 compare with its nearest database rivals?

A: It is 0.6% below the Intel Iris Pro 5200, 0.9% above the AMD FirePro W2100, 1.2% below the GeForce 930M, and 1.2% above the GeForce GTX 460M.

Q: What memory configurations are recorded?

A: The Arc B390 uses system-shared memory, system-shared type, system-shared bus width, and system-dependent bandwidth. The RTX 4070 GDDR6 uses 12 GB GDDR6 on a 192-bit bus with 480.0 GB/s bandwidth.

Q: What are the power and physical differences?

A: The Arc B390 has an 80W TDP, an IGP slot width, and no power connectors. The RTX 4070 GDDR6 has a 200W TDP, a dual-slot width, one 16-pin power connector, and a suggested PSU of 550W.

Q: Do both parts support the same APIs?

A: Yes, both list DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4.

Architecture Differences

The two GPUs use different architectures, process nodes, and foundries. The Intel Arc B390 uses the Panther Lake chip with Xe3-LPG architecture, built on a 3 nm process at Intel. The NVIDIA GeForce RTX 4070 GDDR6 uses the AD104 chip with Ada Lovelace architecture, built on a 5 nm process at TSMC. The NVIDIA chip is recorded at 35,800 million transistors on a 294 mm² die, with a transistor density of 121.8M per mm². The Arc B390's transistor count and die size are recorded as unknown.

Core counts differ in every category. The Arc B390 has 1536 shading units, 48 TMUs, 24 ROPs, and 12 ray tracing cores. The RTX 4070 GDDR6 has 5888 shading units, 184 TMUs, 64 ROPs, 46 ray tracing cores, and 184 tensor cores. The database records no tensor core count for the Arc B390. Cache sizes are not recorded for either part.

Clock and memory architecture also differ. The Arc B390 has a 300 MHz base clock and a 2500 MHz boost clock, with system-shared memory. The RTX 4070 GDDR6 has a 1920 MHz base clock and a 2475 MHz boost clock, with memory clocked at 2500 MHz 20 Gbps effective. The NVIDIA part's memory is 12 GB GDDR6 on a 192-bit bus with 480.0 GB/s bandwidth; the Arc B390's memory bandwidth is system dependent.

Feature-level differences include FP16 throughput. The Arc B390 lists 15.36 TFLOPS FP16 with a 2:1 ratio, while the RTX 4070 GDDR6 lists 29.15 TFLOPS FP16 with a 1:1 ratio. The NVIDIA part also has a 16-pin power connector, dual-slot width, PCIe 4.0 x16 interface, and display outputs recorded as 1x HDMI 2.13x DisplayPort 1.4a. The Arc B390 has no power connectors, IGP slot width, IGP bus interface, and display outputs recorded as portable device dependent.

Specification Differences

Only the fields where the two parts differ are listed below.

| Field | Intel Arc B390 | NVIDIA GeForce RTX 4070 GDDR6 |

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

| Manufacturer | Intel | NVIDIA |

| Series | not recorded | GeForce 40-series |

| Chip | Panther Lake | AD104 |

| Architecture | Xe3-LPG | Ada Lovelace |

| Generation | Arc Graphics-M (Panther Lake) | GeForce 40 |

| Process node | 3 nm | 5 nm |

| Foundry | Intel | TSMC |

| Transistors | unknown | 35,800 million |

| Die size | unknown | 294 mm² |

| Transistor density | not recorded | 121.8M / mm² |

| Base clock | 300 MHz | 1920 MHz |

| Boost clock | 2500 MHz | 2475 MHz |

| Memory clock | System Shared | 2500 MHz 20 Gbps effective |

| Memory size | System Shared | 12 GB |

| Memory type | System Shared | GDDR6 |

| Memory bus width | System Shared | 192 bit |

| Memory bandwidth | System Dependent | 480.0 GB/s |

| Shading units | 1536 | 5888 |

| TMUs | 48 | 184 |

| ROPs | 24 | 64 |

| RT cores | 12 | 46 |

| Tensor cores | not recorded | 184 |

| Pixel rate | 60.00 GPixel/s | 158.4 GPixel/s |

| Texture rate | 120.0 GTexel/s | 455.4 GTexel/s |

| FP32 | 7.680 TFLOPS | 29.15 TFLOPS |

| FP16 | 15.36 TFLOPS (2:1) | 29.15 TFLOPS (1:1) |

| TDP | 80 W | 200 W |

| Slot width | IGP | Dual-slot |

| Power connectors | None | 1x 16-pin |

| Suggested PSU | not recorded | 550 W |

| Bus interface | IGP | PCIe 4.0 x16 |

| Display outputs | Portable Device Dependent | 1x HDMI 2.13x DisplayPort 1.4a |

| Dimensions | not recorded | 240 mm 9.4 inches, 110 mm 4.3 inches, 40 mm 1.6 inches |

| Production status | Active | End-of-life |

| Release date | 2026-01-26 | 2024-08-19 |

| Predecessor | not recorded | GeForce 30 |

| Successor | not recorded | GeForce 50 |

| Launch MSRP | not recorded | 599 USD |