Intel Arc Pro B390 vs Intel Arc Pro B70 Comparison

Head-to-Head Benchmarks

The recorded data for the Intel Arc Pro B390 and the Intel Arc Pro B70 contains no direct head-to-head benchmark results. The database shows zero wins for either part in comparative testing, and neither has an average benchmark score recorded. The percentile ranking for both sits at 50, placing each squarely in the middle of the database's distribution of all GPUs, though this is a neutral position rather than a measured performance tier.

What the data does provide is a clear paper specification gap that translates into a massive theoretical compute advantage for the B70. The B70 delivers 22.94 TFLOPS of FP32 throughput, which is 2.99 times the B390's 7.680 TFLOPS. In FP16 compute, the B70 reaches 45.88 TFLOPS versus 15.36 TFLOPS for the B390, a 2.99x multiplier that holds consistently across both precisions. This is not a marginal difference; it is a threefold raw compute lead that will dominate any workload that scales with shader throughput.

Pixel throughput tells the same story. The B70 outputs 358.4 GPixel/s, while the B390 manages 60.00 GPixel/s, a 5.97x advantage for the discrete card. Texture rate follows with 716.8 GTexel/s against 120.0 GTexel/s, a 5.97x gap. These fillrate differences are far larger than the compute gap, indicating that the B70's render output and texture units are proportionally much stronger relative to its shader count.

Memory bandwidth is the most extreme divergence. The B70 has 608.0 GB/s of dedicated GDDR6 bandwidth across a 256-bit bus, while the B390 relies on System Shared memory with bandwidth listed as System Dependent. The B70's memory subsystem is entirely independent of the host, whereas the B390's performance will vary with the platform's memory configuration. The B70 also carries 32 GB of GDDR6, while the B390 has no dedicated VRAM allocation at all.

The clock speeds add another layer. The B70 boosts to 2800 MHz with a 2280 MHz base, while the B390 boosts to 2500 MHz from a 300 MHz base. The B70 runs 300 MHz higher at boost and 1980 MHz higher at base. The B390's very low base clock suggests aggressive power management for an integrated part, while the B70 sustains a much higher floor.

Where Each One Wins

The B70 wins in every measured specification category that affects standalone graphics performance. Its FP32 and FP16 compute are each roughly triple the B390's figures. Its pixel rate and texture rate are each nearly six times higher. It has 4096 shading units against 1536, 256 TMUs against 48, 128 ROPs against 24, and 32 ray tracing cores against 12. Every shader-focused workload, every rasterization-heavy scene, and every ray tracing task will favor the B70 by a wide margin.

The B390's wins are architectural and platform-based rather than performance-based. It is an IGP with a 80 W TDP, no power connectors, and a slot width listed as IGP, meaning it requires no additional power delivery and occupies no expansion slot. The B70 uses a Dual-slot cooler, requires a 1x 8-pin power connector, and lists a suggested PSU of 550 W. The B390 is built on a 3 nm process from Intel, while the B70 uses a 5 nm process from TSMC. The B390's smaller process node gives it a transistor density and efficiency profile that the larger, older node cannot match, though the database does not record transistor counts or die size for the B390. The B70 lists a die size of 368 mm², but no comparable figure exists for the B390.

The B390 is also earlier in the release sequence, with a launch date of 2026-01-26, while the B70 follows on 2026-03-25. The B390's generation is Arc Graphics-WM (Panther Lake) with a Panther Lake chip and Xe3-LPG architecture, while the B70 is Battlemage (Pro Series) with a BMG-G31 chip and Xe2-HPG architecture. The B390 lists a predecessor, HD Graphics-WM, while the B70 has no predecessor recorded.

Architecture Differences

The two parts come from different architectural families. The B390 uses Xe3-LPG, a low-power graphics architecture, built on a 3 nm process at Intel's own foundry. The B70 uses Xe2-HPG, a high-performance graphics architecture, built on a 5 nm process at TSMC. This is a fundamental split: the B390 targets integrated, power-constrained environments, while the B70 targets discrete, performance-oriented installations.

The B390's chip is Panther Lake, and its generation is Arc Graphics-WM (Panther Lake). The B70's chip is BMG-G31, and its generation is Battlemage (Pro Series). The B390 is the successor to HD Graphics-WM, while the B70 has no predecessor listed. The B390's process node advantage, 3 nm versus 5 nm, is offset by the B70's far larger implementation: 4096 shading units, 256 TMUs, 128 ROPs, and 32 ray tracing cores, versus 1536 shading units, 48 TMUs, 24 ROPs, and 12 ray tracing cores.

Memory architecture is entirely different. The B390 has System Shared memory, System Shared type, System Shared bus width, and System Dependent bandwidth. The B70 has 32 GB of GDDR6 on a 256-bit bus with 608.0 GB/s of bandwidth. The B390's memory clock is also listed as System Shared, while the B70 runs at 2375 MHz with 19 Gbps effective. This means the B390's memory performance is a function of the host platform, while the B70's is fixed and independent.

The bus interface reinforces the split. The B390 uses IGP, meaning it communicates through the integrated graphics path. The B70 uses PCIe 5.0 x16, a full-bandwidth discrete connection. The B390 is an IGP slot width with no power connectors and an 80 W TDP. The B70 is Dual-slot, requires a 1x 8-pin power connector, and has a 230 W TDP with a suggested PSU of 550 W.

API support is identical: both list DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. Display outputs differ. The B390 lists Portable Device Dependent outputs, while the B70 lists 1x HDMI 2.1 and 3x DisplayPort 2.1. The B70 has physical dimensions of 267 mm length, 110 mm height, and 39 mm width, while the B390 has no dimensions recorded.

The Verdict

The data supports a clear split. The B70 is the overwhelming performance choice. It triples the B390 in compute, multiplies fillrate by nearly six, provides 32 GB of dedicated GDDR6 with 608.0 GB/s of bandwidth, and adds 32 ray tracing cores. Any workload that depends on raw throughput, memory capacity, or bandwidth belongs to the B70. The B70 is also the only one of the two with a recorded launch MSRP of 949 USD, which can be stated once as a reference point.

The B390 is the efficiency and integration choice. It runs at 80 W versus 230 W, needs no power connector, occupies no slot, uses a smaller 3 nm Intel process, and ships earlier. It is built for systems where the GPU must share the host's memory and power budget, and where the discrete performance of the B70 is either unnecessary or physically impossible to install. The B390's System Dependent bandwidth and Portable Device Dependent display outputs indicate a part designed for mobile or compact platforms.

There is no benchmark data in the database to confirm real-world performance, so the verdict rests entirely on specification analysis. The compute, fillrate, memory, and feature gaps are so large that no software optimization could close them. The B70's 22.94 TFLOPS versus 7.680 TFLOPS, its 358.4 GPixel/s versus 60.00 GPixel/s, and its 608.0 GB/s versus System Dependent bandwidth all point in the same direction. The B70 is the performance part. The B390 is the integrated part.

FAQ

Q: Which GPU has higher FP32 compute performance?

A: The Intel Arc Pro B70 delivers 22.94 TFLOPS, which is 2.99 times the Intel Arc Pro B390's 7.680 TFLOPS.

Q: How much memory does each GPU have?

A: The Intel Arc Pro B70 has 32 GB of GDDR6 on a 256-bit bus with 608.0 GB/s bandwidth. The Intel Arc Pro B390 uses System Shared memory with System Dependent bandwidth.

Q: What are the TDP requirements?

A: The Intel Arc Pro B390 has an 80 W TDP with no power connectors. The Intel Arc Pro B70 has a 230 W TDP, requires a 1x 8-pin power connector, and lists a suggested PSU of 550 W.

Q: What are the process nodes?

A: The Intel Arc Pro B390 is built on a 3 nm process at Intel. The Intel Arc Pro B70 is built on a 5 nm process at TSMC.

Q: When were they released?

A: The Intel Arc Pro B390 was released on 2026-01-26. The Intel Arc Pro B70 was released on 2026-03-25.

Q: What is the slot and interface difference?

A: The Intel Arc Pro B390 is an IGP with a bus interface of IGP. The Intel Arc Pro B70 is a Dual-slot card using PCIe 5.0 x16.

Specification Differences

| Specification | Intel Arc Pro B390 | Intel Arc Pro B70 |

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

| Chip | Panther Lake | BMG-G31 |

| Architecture | Xe3-LPG | Xe2-HPG |

| Generation | Arc Graphics-WM (Panther Lake) | Battlemage (Pro Series) |

| Process Node | 3 nm | 5 nm |

| Foundry | Intel | TSMC |

| Die Size | unknown | 368 mm² |

| Base Clock | 300 MHz | 2280 MHz |

| Boost Clock | 2500 MHz | 2800 MHz |

| Memory Size | System Shared | 32 GB |

| Memory Type | System Shared | GDDR6 |

| Memory Bus Width | System Shared | 256 bit |

| Memory Bandwidth | System Dependent | 608.0 GB/s |

| Memory Clock | System Shared | 2375 MHz 19 Gbps effective |

| Shading Units | 1536 | 4096 |

| TMUs | 48 | 256 |

| ROPs | 24 | 128 |

| Ray Tracing Cores | 12 | 32 |

| Pixel Rate | 60.00 GPixel/s | 358.4 GPixel/s |

| Texture Rate | 120.0 GTexel/s | 716.8 GTexel/s |

| FP32 | 7.680 TFLOPS | 22.94 TFLOPS |

| FP16 | 15.36 TFLOPS (2:1) | 45.88 TFLOPS (2:1) |

| TDP | 80 W | 230 W |

| Slot Width | IGP | Dual-slot |

| Power Connectors | None | 1x 8-pin |

| Suggested PSU | null | 550 W |

| Bus Interface | IGP | PCIe 5.0 x16 |

| Display Outputs | Portable Device Dependent | 1x HDMI 2.1, 3x DisplayPort 2.1 |

| Dimensions | null | 267 mm, 110 mm, 39 mm |

| Release Date | 2026-01-26 | 2026-03-25 |

| Predecessor | HD Graphics-WM | null |

| Launch MSRP | null | 949 USD |