Intel Arc Pro B390

# Intel Arc Pro B390: Integrated Graphics for the Panther Lake Era

The Intel Arc Pro B390 is an integrated graphics processor (IGP) built on the Xe3-LPG architecture, part of the Arc Graphics-WM (Panther Lake) generation. Fabricated on Intel's 3 nm process, this chip combines 1536 shading units, 48 texture mapping units, and 24 raster operation pipelines into a power envelope of just 80 W. With a base clock of 300 MHz and a boost clock of 2500 MHz, the B390 delivers a peak FP32 throughput of 7.680 TFLOPS and a pixel rate of 60.00 GPixel/s. It occupies the 50th percentile among all GPUs, placing it squarely in the mid-range of the performance spectrum, a notable achievement for an integrated solution.

Power and Cooling

The Intel Arc Pro B390 carries a thermal design power (TDP) of 80 W. This figure represents the maximum heat the cooling solution must dissipate under sustained load. For an integrated graphics processor, this is a modest requirement that fits comfortably within the thermal headroom of a typical laptop or compact desktop chassis. The low power draw means that a dedicated cooling solution is unnecessary; the chip relies on the system's existing thermal management, as indicated by its IGP slot width.

Because the B390 is an integrated part, it requires no power connectors. The absence of PCIe power cables (6-pin or 8-pin) simplifies installation and reduces cable clutter. This contrasts sharply with discrete graphics cards, which often demand supplementary power. The system's power supply unit (PSU) must only account for the 80 W TDP of the GPU, plus the rest of the platform's components. Benchmark data does not specify a suggested PSU wattage, so builders should size their power supply based on the entire system's requirements, but the GPU itself adds minimal strain.

The bus interface is IGP, meaning the B390 communicates with the rest of the system through the processor's integrated memory controller rather than a dedicated PCIe slot. This design inherently limits the bandwidth available to the GPU, but it also eliminates the need for slot-based power delivery. The 3 nm process node contributes to the efficiency here; smaller transistors typically reduce leakage current and improve power scaling. The result is an integrated GPU that can sustain its boost clock without exotic cooling, making it suitable for thin-and-light systems where thermal capacity is at a premium.

Ray Tracing and Feature Set

The Intel Arc Pro B390 includes 12 dedicated ray tracing cores. This hardware accelerates the BVH traversal and ray-triangle intersection tests that form the backbone of real-time ray tracing. While 12 RT cores is a modest count compared to high-end discrete GPUs, it represents a significant capability for an integrated part. Games that support ray-traced shadows or reflections can offload these calculations from the shader units, preserving frame rate in hybrid rendering scenarios.

In terms of API support, the B390 is fully compliant with DirectX 12 Ultimate, specifically the 12_2 feature level. This is the most recent DirectX 12 feature level, encompassing hardware ray tracing, variable rate shading, and mesh shaders. The Vulkan 1.4 API is also supported, providing low-overhead access to the GPU for cross-platform titles and compute workloads. OpenGL 4.6 compatibility ensures legacy applications and professional software continue to function without issue.

Notably, the FACT PACK does not list tensor core information for the B390. This means the data does not confirm dedicated AI acceleration hardware. The absence of tensor cores in the specifications does not preclude machine learning workloads, shader-based compute can handle these tasks, but it suggests the B390 may not excel at AI inference compared to parts with dedicated tensor units. The FP16 throughput of 15.36 TFLOPS (at a 2:1 ratio) indicates the shader units can perform half-precision math at double the FP32 rate, which can accelerate some AI and compute operations, but this is not a substitute for dedicated tensor hardware.

Memory Subsystem

The B390 uses "System Shared" memory for both VRAM capacity and type. This means the GPU does not have its own dedicated video memory; instead, it dynamically allocates a portion of the system's main RAM. The bus width is also listed as "System Shared," which reflects that the memory bandwidth is determined by the platform's memory controller rather than a dedicated GPU bus.

Consequently, the memory bandwidth is "System Dependent." In practical terms, the B390's performance scales with the speed and configuration of the host system's RAM. A dual-channel DDR5 setup at high transfer rates will provide substantially more bandwidth than a single-channel configuration. This is a critical consideration for gaming at high resolutions, the B390's ability to feed its 1536 shading units depends entirely on the system memory's throughput.

For 1080p gaming, the shared memory architecture is typically sufficient, as the bandwidth demands are moderate. At 1440p or 4K, however, the increased texture and geometry data can saturate the memory bus, leading to stuttering or reduced frame rates. The 60.00 GPixel/s pixel rate suggests the ROPs can fill a 4K frame (approximately 8.3 million pixels) in about 0.14 milliseconds, but the memory subsystem must keep pace with the data transfer. Users planning high-resolution gaming should pair the B390 with fast, dual-channel memory to mitigate this bottleneck.

How It Compares

The FACT PACK provides no nearest rivals for the Intel Arc Pro B390. This absence of comparative data means the benchmark database does not currently list specific competing GPUs with calculable deltas. Without this information, a direct numerical comparison to other parts is not possible from the available facts.

What can be stated is the B390's percentile ranking. At the 50th percentile among all GPUs, it sits at the median of the performance distribution. This implies it outperforms roughly half of all GPUs in the database while trailing the other half. For an integrated part, this is a strong position, as IGPs typically occupy the lower quartiles. The B390 appears to close the gap between integrated and entry-level discrete graphics, though the lack of rival data prevents quantifying this gap.

The predecessor to the B390 is the "HD Graphics-WM" series. The B390 represents a generational leap, moving from that older design to the Xe3-LPG architecture with dedicated RT cores and DirectX 12 Ultimate support. The 3 nm process node also marks a significant manufacturing advancement. However, without benchmark scores for the predecessor, a quantitative performance delta cannot be derived.

Benchmark Performance

The benchmark section for the Intel Arc Pro B390 is empty, and the average benchmark score is recorded as 0. This is an unusual data point, as it suggests no synthetic or real-world tests have been logged in the database for this part. The percentile ranking of 50 is likewise based on the available data, which may be sparse or estimated.

What can be analyzed is the theoretical throughput derived from the clock speeds and core counts. The FP32 performance of 7.680 TFLOPS is a strong indicator of raw shader throughput. To contextualize this, consider that a typical integrated GPU in the same power class might deliver 3-4 TFLOPS; the B390's figure is roughly double that, based on the specification sheet. The texture fill rate of 120.0 GTexel/s suggests the 48 TMUs can process 120 billion texels per second, which is adequate for 1080p texture-heavy workloads.

The pixel rate of 60.00 GPixel/s, derived from the 24 ROPs at the 2500 MHz boost clock, is a more modest figure. This limits the fill-rate-bound performance, such as when applying heavy post-processing effects or running at very high resolutions. In practice, the B390 will likely excel in compute-bound scenarios (thanks to the high FP32 throughput) but may fall behind in pixel-bound situations. The FP16 rate of 15.36 TFLOPS (2:1 ratio) doubles the throughput for half-precision operations, which can accelerate certain shader effects and machine learning inference.

Given the lack of benchmark scores, these theoretical numbers serve as the primary performance guide. They indicate a GPU that is competitive for 1080p gaming at medium settings and capable of light 1440p workloads, though the memory bandwidth will be the limiting factor in real-world tests.

Who Should Consider It

The Intel Arc Pro B390 is best suited for users who need a capable integrated GPU without the power and space requirements of a discrete card. The 80 W TDP and IGP form factor make it ideal for compact desktops, mini PCs, and laptops where thermal and space constraints are paramount. The 50th percentile ranking suggests it offers balanced performance for general-purpose computing, including office work, media playback, and light content creation.

For gaming, the B390 targets 1080p resolution at medium settings. The 7.680 TFLOPS FP32 performance is sufficient for most modern titles at this resolution, provided the system memory is fast enough to feed the GPU. Users with dual-channel DDR5 RAM will see the best results; single-channel configurations will bottleneck the B390's performance. At 1440p, the B390 can handle esports titles and older games, but demanding AAA releases may require reduced settings. The 4K resolution is not recommended for gaming, as the pixel rate and memory bandwidth are insufficient for smooth frame rates in graphically intensive titles.

The 12 RT cores enable entry-level ray tracing. Gamers who prioritize ray-traced effects should consider this a bonus rather than a primary feature; the B390 can enable RT in less demanding games or at lower resolutions, but it will struggle with heavy RT workloads. Professional users running OpenGL 4.6 applications will find full compatibility, and the DirectX 12 Ultimate support future-proofs the B390 for upcoming titles that leverage mesh shaders and variable rate shading.

This GPU is not intended for enthusiasts seeking high-refresh-rate or high-resolution gaming. It is a pragmatic choice for users who want a single, energy-efficient solution for everyday computing and casual gaming. The "System Dependent" memory bandwidth is a critical caveat, the B390's real-world performance hinges on the rest of the system's configuration. Builders should allocate a generous portion of their budget to fast, dual-channel RAM to unlock the B390's full potential.

FAQ

Q: Does the Intel Arc Pro B390 have its own dedicated video memory?

A: No. The B390 uses "System Shared" memory for both capacity and type, meaning it dynamically allocates from the system's main RAM. The bus width is also system-shared, and the bandwidth is listed as "System Dependent."

Q: What power connectors does the B390 require?

A: None. The power connectors field is listed as "None," and the slot width is "IGP." The GPU draws its 80 W TDP directly from the system's power delivery, requiring no additional cables.

Q: Can the B390 handle ray tracing?

A: Yes, it includes 12 dedicated ray tracing cores. The B390 supports DirectX 12 Ultimate (12_2), which mandates hardware ray tracing, as well as Vulkan 1.4. Ray tracing performance will be modest given the integrated nature of the GPU.

Q: What is the maximum supported DirectX version?

A: The B390 supports DirectX 12 Ultimate, specifically the 12_2 feature level. It also supports Vulkan 1.4 and OpenGL 4.6.

Q: Is the B390 suitable for 4K gaming?

A: The data suggests it is not ideal. The pixel rate of 60.00 GPixel/s and the system-dependent memory bandwidth will likely bottleneck 4K workloads. The 7.680 TFLOPS FP32 performance is more aligned with 1080p gaming at medium settings.

Q: What process node is the B390 built on?

A: The B390 is fabricated on a 3 nm process at Intel. This advanced node contributes to the 80 W TDP while allowing a boost clock of 2500 MHz.