Intel Arc A380M vs Intel Arc Pro B390 Comparison

Where Each One Wins

The Intel Arc A380M and Intel Arc Pro B390 occupy different positions in the mobile graphics landscape, and the recorded data shows a clear split in their intended workloads. The A380M, built on the Alchemist architecture, is a discrete MXM module with dedicated GDDR6 memory. The B390, an integrated graphics processor on the Panther Lake chip, uses system-shared memory. This fundamental difference shapes where each part excels.

The A380M wins in memory bandwidth. It provides 186.0 GB/s of dedicated bandwidth over a 96-bit bus, while the B390's bandwidth is listed as "System Dependent," meaning it shares the system memory controller and is subject to contention from the CPU and other components. For workloads that are bandwidth-sensitive, such as texture-heavy scenes or large framebuffer operations, the discrete memory arrangement gives the A380M a structural advantage.

The B390 wins in raw compute throughput. Its FP32 performance is 7.680 TFLOPS versus 4.096 TFLOPS for the A380M, a 87.5% advantage. The B390 also has 1536 shading units compared to 1024, and 12 ray tracing cores versus 8. For compute-heavy tasks such as rendering, simulation, or AI inference (when using FP32 paths), the B390's larger shader array and higher boost clock of 2500 MHz versus 2000 MHz deliver markedly higher peak throughput.

The A380M counters with higher texture and pixel rates relative to its smaller footprint. It achieves 128.0 GTexel/s and 64.00 GPixel/s, while the B390 records 120.0 GTexel/s and 60.00 GPixel/s. Despite having fewer TMUs (64 vs 48) and ROPs (32 vs 24), the A380M's higher clock speeds at base (1550 MHz vs 300 MHz) and its dedicated memory allow it to exceed the B390 in these fill-rate metrics. This suggests the A380M is better suited to rasterization-bound scenes where fill rate matters more than raw shader throughput.

Power consumption is another differentiator. The A380M draws 35 W, while the B390 is rated at 80 W. The B390's higher power envelope enables its larger compute array, but the A380M delivers more performance per watt in fill-rate terms. The A380M is a capable air-cooled MXM module, while the B390 is an IGP integrated into the Panther Lake processor, sharing the system's thermal solution.

The production status for both is "Active," and both sit at the 50th percentile in the database's GPU rankings. Neither has benchmark scores or nearest rivals recorded, so the analysis relies on architectural and specification data rather than measured performance tests.

Architecture Differences

The two GPUs come from different generations and foundries. The A380M uses the DG2-128 chip built on TSMC's 6 nm process, part of the Xe-HPG architecture, and belongs to the Alchemist generation under the Arc 3 Mobile designation. The B390 uses the Panther Lake chip built on Intel's 3 nm process, part of the Xe3-LPG architecture, and belongs to the Arc Graphics-WM (Panther Lake) generation.

Transistor counts and die sizes differ sharply. The A380M packs 7,200 million transistors into a 157 mm² die, yielding a density of 45.9M / mm². The B390's transistor count and die size are listed as "unknown," but its 3 nm Intel process suggests a newer manufacturing node. The A380M's 6 nm TSMC process is a mature node, while the B390's 3 nm Intel process represents a more advanced fabrication technology.

Memory architecture is a major divergence. The A380M uses 6 GB of GDDR6 on a 96-bit bus with 186.0 GB/s bandwidth. The B390 uses system-shared memory with no dedicated VRAM, a "System Shared" bus width, and bandwidth that varies with the host system. This makes the A380M a self-contained graphics solution, while the B390 relies entirely on the platform's memory subsystem.

Clocking strategies also differ. The A380M has a base clock of 1550 MHz and a boost of 2000 MHz, with memory running at 1937 MHz (15.5 Gbps effective). The B390 has a very low base clock of 300 MHz and a boost of 2500 MHz, indicating aggressive power management and a wide dynamic range. The B390's boost clock is 25% higher than the A380M's, but its base clock is dramatically lower, reflecting its integrated nature and thermal constraints.

Shader configurations diverge in count and ratio. The B390 has 1536 shading units, 48 TMUs, and 24 ROPs. The A380M has 1024 shading units, 64 TMUs, and 32 ROPs. The B390 allocates more transistors to compute, while the A380M maintains a more balanced mix for traditional graphics. Ray tracing cores follow the same pattern: 12 for the B390, 8 for the A380M.

The B390 has a predecessor, HD Graphics-WM, while the A380M has no listed predecessor or successor. The B390's release date is 2026-01-26, roughly three years after the A380M's 2023-01-23 release. Both support DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4, so API feature parity is complete.

Form factor and interface differ fundamentally. The A380M is an MXM Module with an MXM-A (3.1) bus interface, designed for upgradeable laptop graphics. The B390 is an IGP with a bus interface listed as "IGP," meaning it is fused into the processor and cannot be replaced or upgraded. The A380M has no power connectors listed, while the B390 explicitly has "None" for power connectors, as it draws power through the socket.

Head-to-Head Benchmarks

The database records no direct head-to-head benchmark scores for these two GPUs, with both winsA and winsB at zero. The comparison must therefore be constructed from the recorded specifications, which reveal clear winners in specific metrics.

FP32 compute is the B390's largest victory. At 7.680 TFLOPS, it is 87.5% ahead of the A380M's 4.096 TFLOPS. This means the B390 can process nearly twice as many floating-point operations per second, making it far stronger for compute-bound tasks like scientific visualization, rendering, or any workload that stresses shader ALUs. FP16 performance follows the same pattern: 15.36 TFLOPS for the B390 versus 8.192 TFLOPS for the A380M, an 87.5% advantage again.

Memory bandwidth is the A380M's biggest win. Its 186.0 GB/s of dedicated GDDR6 bandwidth is fixed and guaranteed, while the B390's bandwidth is "System Dependent." In a typical mobile platform, shared memory bandwidth is often lower than dedicated graphics memory bandwidth due to contention, and it is never higher than the system's total memory bandwidth. The A380M's dedicated path avoids this contention entirely.

Fill rates are close but favor the A380M. The A380M's 128.0 GTexel/s texture rate beats the B390's 120.0 GTexel/s by 6.7%. Its pixel rate of 64.00 GPixel/s beats the B390's 60.00 GPixel/s by 6.7% as well. These margins are small but consistent, suggesting the A380M is slightly better at traditional rasterization output.

Clock speeds tell a mixed story. The B390's boost clock of 2500 MHz is 25% higher than the A380M's 2000 MHz, but the A380M's base clock of 1550 MHz is 416.7% higher than the B390's 300 MHz. The B390's wide clock range indicates it can idle very low and boost very high, but sustained performance depends on cooling and power delivery. The A380M's narrower range suggests more consistent performance under load.

Shader unit count favors the B390 by 50% (1536 vs 1024), and ray tracing cores favor the B390 by 50% as well (12 vs 8). However, the A380M's higher TMU and ROP counts (64 vs 48 and 32 vs 24, respectively) mean it can maintain fill-rate throughput despite fewer shaders, a tradeoff that historically favors resolution and texture quality over compute-heavy effects.

Power efficiency is a nuanced comparison. The A380M draws 35 W and delivers 4.096 TFLOPS, yielding roughly 117 GFLOPS per watt. The B390 draws 80 W and delivers 7.680 TFLOPS, yielding 96 GFLOPS per watt. The A380M is approximately 22% more power-efficient in FP32 compute per watt, despite being on an older 6 nm process. In fill-rate efficiency, the A380M delivers 3.66 GTexel/s per watt versus 1.50 GTexel/s per watt for the B390, a 144% advantage.

FAQ

Q: Which GPU has higher FP32 compute performance?

A: The Intel Arc Pro B390 delivers 7.680 TFLOPS, which is 87.5% higher than the Intel Arc A380M's 4.096 TFLOPS.

Q: How does memory configuration differ between the two?

A: The A380M uses 6 GB of dedicated GDDR6 memory on a 96-bit bus with 186.0 GB/s bandwidth. The B390 uses system-shared memory with a "System Shared" bus width and "System Dependent" bandwidth, meaning it shares the host system's memory.

Q: Which GPU has better texture and pixel fill rates?

A: The A380M has higher fill rates: 128.0 GTexel/s and 64.00 GPixel/s, compared to the B390's 120.0 GTexel/s and 60.00 GPixel/s, a 6.7% advantage in both metrics.

Q: What are the power requirements for each GPU?

A: The A380M is rated at 35 W TDP and uses an MXM module form factor with no power connectors listed. The B390 is rated at 80 W TDP, uses an IGP form factor, and explicitly lists "None" for power connectors.

Q: Are the two GPUs from the same architecture generation?

A: No. The A380M uses the Xe-HPG architecture on TSMC's 6 nm process (Alchemist generation, DG2-128 chip). The B390 uses the Xe3-LPG architecture on Intel's 3 nm process (Arc Graphics-WM, Panther Lake chip).

Q: Do both GPUs support the same graphics APIs?

A: Yes. Both support DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4, providing full API feature parity.

Specification Differences

| Specification | Intel Arc A380M | Intel Arc Pro B390 |

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

| Chip | DG2-128 | Panther Lake |

| Architecture | Xe-HPG | Xe3-LPG |

| Generation | Alchemist (Arc 3 Mobile) | Arc Graphics-WM (Panther Lake) |

| Process Node | 6 nm (TSMC) | 3 nm (Intel) |

| Transistors | 7,200 million | unknown |

| Die Size | 157 mm² | unknown |

| Transistor Density | 45.9M / mm² | null |

| Base Clock | 1550 MHz | 300 MHz |

| Boost Clock | 2000 MHz | 2500 MHz |

| Memory Clock | 1937 MHz, 15.5 Gbps effective | System Shared |

| Memory Size | 6 GB GDDR6 | System Shared |

| Memory Bus Width | 96 bit | System Shared |

| Memory Bandwidth | 186.0 GB/s | System Dependent |

| Shading Units | 1024 | 1536 |

| TMUs | 64 | 48 |

| ROPs | 32 | 24 |

| Ray Tracing Cores | 8 | 12 |

| Pixel Rate | 64.00 GPixel/s | 60.00 GPixel/s |

| Texture Rate | 128.0 GTexel/s | 120.0 GTexel/s |

| FP32 | 4.096 TFLOPS | 7.680 TFLOPS |

| FP16 | 8.192 TFLOPS (2:1) | 15.36 TFLOPS (2:1) |

| TDP | 35 W | 80 W |

| Slot Width | MXM Module | IGP |

| Power Connectors | null | None |

| Bus Interface | MXM-A (3.1) | IGP |

| Release Date | 2023-01-23 | 2026-01-26 |

| Predecessor | null | HD Graphics-WM |

| Production Status | Active | Active |

The two GPUs also share identical display outputs ("Portable Device Dependent") and identical API support (DirectX 12 Ultimate, OpenGL 4.6, Vulkan 1.4). Neither has a listed launch MSRP. Both sit at the 50th percentile in the database, with no benchmark scores or nearest rivals recorded. The B390's transistor count, die size, and transistor density are unknown, while the A380M's are fully specified. The A380M has no predecessor or successor listed, while the B390 lists HD Graphics-WM as its predecessor.