Intel Arc A570M vs Intel Arc B370 Comparison

FAQ

Q: How do the Intel Arc A570M and Intel Arc B370 compare in raw compute performance?

A: The Arc A570M delivers 5.325 TFLOPS FP32 and 10.65 TFLOPS FP16, while the Arc B370 delivers 6.144 TFLOPS FP32 and 12.29 TFLOPS FP16. The B370 holds a 15.4% advantage in FP32 throughput and a 15.4% advantage in FP16 throughput.

Q: Which GPU has the higher memory bandwidth?

A: The Arc A570M uses 8 GB GDDR6 on a 128-bit bus with 224.0 GB/s bandwidth. The Arc B370 uses system shared memory with system-dependent bandwidth, so it cannot match the dedicated memory bandwidth of the A570M.

Q: How do their benchmark scores compare?

A: The Arc A570M records a Geekbench OpenCL score of 58239, placing it in the 88th percentile of all GPUs. The Arc B370 records a 3DMark Steel Nomad DX12 score of 1184, placing it in the 5th percentile. These are different benchmark suites, so direct numerical comparison is not meaningful.

Q: What are the architectural generations of these two GPUs?

A: The Arc A570M uses the DG2-256 chip with Xe-HPG architecture, part of the Alchemist generation (Arc 5 Mobile). The Arc B370 uses the Panther Lake chip with Xe3-LPG architecture, part of the Arc Graphics-M (Panther Lake) generation.

Q: Which GPU has more shading units and texture units?

A: The Arc A570M has 2048 shading units, 128 TMUs, and 64 ROPs. The Arc B370 has 1280 shading units, 40 TMUs, and 20 ROPs. The A570M has 60% more shading units, 220% more TMUs, and 220% more ROPs.

Q: What is the thermal design power difference?

A: The Arc A570M has a TDP of 75 W, while the Arc B370 has a TDP of 25 W. The B370 consumes one-third the power of the A570M.

Where Each One Wins

The Arc A570M wins decisively in memory-centric workloads. Its dedicated 8 GB GDDR6 memory with 224.0 GB/s bandwidth provides predictable performance for texture-heavy rendering and large dataset operations. The B370's system shared memory has system-dependent bandwidth, which introduces variability based on the host platform's memory configuration. The A570M also wins on raw geometry throughput: its 128 TMUs and 64 ROPs produce 166.4 GTexel/s texture rate and 83.20 GPixel/s pixel rate, versus 96.00 GTexel/s and 48.00 GPixel/s for the B370. That is 73.3% higher texture fill and 73.3% higher pixel fill.

The Arc B370 wins on compute density and efficiency. Its 6.144 TFLOPS FP32 output comes from a 25 W TDP, which is a ratio of 245.8 GFLOPS per watt. The A570M's 5.325 TFLOPS from 75 W yields 71.0 GFLOPS per watt. The B370 delivers 3.5 times the FP32 throughput per watt. Additionally, the B370's boost clock of 2400 MHz is 84.6% higher than the A570M's 1300 MHz boost, which helps it close the shading-unit deficit. The B370 also uses a 3 nm process from Intel foundry, versus 6 nm from TSMC for the A570M, giving it a manufacturing advantage in transistor density potential.

The A570M wins in the percentile ranking. Its 88th percentile places it among high-end mobile GPUs, while the B370's 5th percentile indicates entry-level positioning. The nearest rivals confirm this split: the A570M trades within 0.7% of the AMD Radeon RX 6950 XT, AMD Radeon RX 5600 OEM, NVIDIA P102-100, and AMD Radeon PRO V710. The B370 trades within 1.4% of the ATI Mobility Radeon HD 5570, ATI Radeon HD 5770, AMD Radeon HD 7650M, and AMD FirePro M2000.

Architecture Differences

The Arc A570M uses the DG2-256 chip built on the Xe-HPG architecture, which is the foundation of the Alchemist generation. The Arc B370 uses the Panther Lake chip built on the Xe3-LPG architecture, which is the graphics component of the Arc Graphics-M (Panther Lake) generation. This is a two-generation architectural jump for the B370, moving from Xe-HPG directly to Xe3-LPG.

The Xe3-LPG architecture in the B370 achieves higher performance per shading unit. The B370 has 1280 shading units producing 6.144 TFLOPS FP32, which is 4.8 TFLOPS per 1000 shading units. The A570M has 2048 shading units producing 5.325 TFLOPS FP32, which is 2.6 TFLOPS per 1000 shading units. The B370's architecture is 84.6% more efficient per shading unit, likely due to the higher boost clock and architectural improvements.

The processing node differs significantly. The A570M uses a 6 nm process at TSMC with 11,500 million transistors on a 269 mm² die, yielding a transistor density of 42.8 million per mm². The B370 uses a 3 nm process at Intel foundry with unknown transistor count and die size. The 3 nm node allows for denser transistor packing, which contributes to the B370's higher clock speeds despite the lower TDP.

The memory architecture is fundamentally different. The A570M has dedicated GDDR6 memory with a 128-bit bus and fixed bandwidth. The B370 uses system shared memory with system-dependent bandwidth. This means the B370's memory performance depends entirely on the host system's memory configuration, whereas the A570M has deterministic memory performance.

Both GPUs support DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. Both have ray tracing cores: 16 RT cores for the A570M and 10 RT cores for the B370. Neither has dedicated tensor cores listed.

Specification Differences

| Specification | Intel Arc A570M | Intel Arc B370 |

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

| Chip | DG2-256 | Panther Lake |

| Architecture | Xe-HPG | Xe3-LPG |

| Generation | Alchemist (Arc 5 Mobile) | Arc Graphics-M (Panther Lake) |

| Process Node | 6 nm | 3 nm |

| Foundry | TSMC | Intel |

| Transistors | 11,500 million | unknown |

| Die Size | 269 mm² | unknown |

| Base Clock | 900 MHz | 300 MHz |

| Boost Clock | 1300 MHz | 2400 MHz |

| Memory Type | GDDR6 | System Shared |

| Memory Size | 8 GB | System Shared |

| Memory Bus | 128 bit | System Shared |

| Memory Bandwidth | 224.0 GB/s | System Dependent |

| Memory Clock | 1750 MHz (14 Gbps effective) | System Shared |

| Shading Units | 2048 | 1280 |

| TMUs | 128 | 40 |

| ROPs | 64 | 20 |

| RT Cores | 16 | 10 |

| Pixel Rate | 83.20 GPixel/s | 48.00 GPixel/s |

| Texture Rate | 166.4 GTexel/s | 96.00 GTexel/s |

| FP32 | 5.325 TFLOPS | 6.144 TFLOPS |

| FP16 | 10.65 TFLOPS (2:1) | 12.29 TFLOPS (2:1) |

| TDP | 75 W | 25 W |

| Bus Interface | PCIe 4.0 x8 | IGP |

| Power Connectors | Not specified | None |

| Release Date | 2023-07-31 | 2026-01-26 |

The release date gap is substantial: the A570M launched on 2023-07-31, while the B370 launches on 2026-01-26. This 30-month gap explains the architectural and process node improvements in the B370.

Head-to-Head Benchmarks

The head-to-head benchmark comparison contains no direct shared tests between the two GPUs. The A570M recorded a Geekbench OpenCL score of 58239, and the B370 recorded a 3DMark Steel Nomad DX12 score of 1184. These benchmarks measure different workloads: OpenCL compute versus DirectX 12 gaming rasterization. The scores cannot be directly compared, but they can be contextualized through their percentile rankings and nearest rivals.

The A570M's 58239 Geekbench OpenCL score places it at the 88th percentile of all GPUs. Its nearest rival, the AMD Radeon RX 6950 XT, scores 58392, which is 0.3% higher. The AMD Radeon RX 5600 OEM scores 58085, which is 0.3% lower. The NVIDIA P102-100 scores 58528, which is 0.5% higher. The AMD Radeon PRO V710 scores 58657, which is 0.7% higher. This places the A570M in a tight cluster of high-performance GPUs, all within 0.7% of each other.

The B370's 1184 3DMark Steel Nomad DX12 score places it at the 5th percentile of all GPUs. Its nearest rival, the ATI Mobility Radeon HD 5570, scores 1186, which is 0.2% higher. The ATI Radeon HD 5770 scores 1190, which is 0.5% higher. The AMD Radeon HD 7650M scores 1192, which is 0.7% higher. The AMD FirePro M2000 scores 1168, which is 1.4% lower. This places the B370 in a cluster of legacy and entry-level GPUs, all within 1.4% of each other.

The architecture efficiency differences show up in the compute-to-fill-rate ratio. The A570M has a FP32 to texture rate ratio of 32.0 GFLOPS per GTexel/s, while the B370 has a ratio of 64.0 GFLOPS per GTexel/s. The B370 allocates more of its silicon to compute throughput relative to texture work, which aligns with its higher FP32 output despite fewer TMUs.

The clock speed difference is the most dramatic specification gap. The B370's 2400 MHz boost clock is 84.6% higher than the A570M's 1300 MHz boost. The base clocks tell a different story: the A570M's 900 MHz base is 200% higher than the B370's 300 MHz base. This suggests the B370 relies heavily on boost behavior to reach its performance targets, while the A570M maintains a higher sustained clock.

The Verdict

The data shows two GPUs designed for entirely different roles. The Arc A570M is a dedicated mobile GPU with its own memory subsystem, high fill rates, and a large shading unit count. It sits at the 88th percentile, competing with the AMD Radeon RX 6950 XT and the AMD Radeon PRO V710 within 0.7%. Its 8 GB GDDR6 memory and 224.0 GB/s bandwidth make it suitable for applications that require consistent memory throughput and substantial texture processing. The 75 W TDP is appropriate for a GPU with 2048 shading units and 16 RT cores.

The Arc B370 is an integrated-class GPU built on a 3 nm process with a 25 W TDP. It delivers higher FP32 compute than the A570M despite having 37.5% fewer shading units. The 2400 MHz boost clock and Xe3-LPG architecture enable this efficiency. However, its system shared memory and 5th percentile ranking place it firmly in entry-level territory. The nearest rivals, including the ATI Mobility Radeon HD 5570 and AMD Radeon HD 7650M, are over a decade old, which indicates the B370's performance class is modest.

The production status for both GPUs is active. The A570M launched on 2023-07-31, and the B370 launches on 2026-01-26. The 30-month gap between releases explains the process node difference: TSMC 6 nm for the A570M versus Intel 3 nm for the B370. The B370's higher FP32 output per shading unit and per watt confirms architectural progress, but the A570M's dedicated memory and higher fill rates give it advantages in rasterization-heavy workloads.

For workloads that depend on texture rate, pixel rate, or memory bandwidth, the A570M is the clear choice. Its 166.4 GTexel/s and 83.20 GPixel/s are 73.3% higher than the B370's corresponding rates. For compute-heavy workloads that fit within system shared memory and for power-constrained environments, the B370's 6.144 TFLOPS FP32 at 25 W offers superior efficiency. The B370's 3 nm process and 84.6% higher boost clock indicate a newer design point optimized for low power and high compute density.

The benchmark data does not include a shared test between the two GPUs, so a direct performance comparison is impossible. The percentile rankings provide the clearest signal: the A570M at 88th percentile is a high-end part, and the B370 at 5th percentile is an entry-level part. The A570M belongs in systems requiring dedicated graphics memory and high fill rates. The B370 belongs in ultra-portable devices where 25 W power draw and integrated memory are mandatory. Each GPU wins in its intended category: the A570M for memory-bound and fill-rate-bound tasks, the B370 for compute-per-watt and architectural efficiency.