Intel Arc Pro B65

Power and Cooling, TDP, PSU recommendation, connector requirements

The Intel Arc Pro B65 is specified with a 200 W TDP, which places it in the mid-range power tier for professional graphics cards. This figure represents the maximum board power draw under sustained workloads, and the data indicates that a 550 W power supply is the suggested minimum for systems housing this card. The recommendation accounts for typical platform power draws from the CPU, motherboard, storage, and other peripherals, ensuring stable operation without overstressing the PSU.

Power delivery is handled through a single 8-pin PCIe power connector. This is a straightforward requirement, as many modern power supplies include at least one 8-pin cable, and adapters from dual 6-pin configurations are commonly available. The single-connector design simplifies cable management and reduces the number of potential failure points compared to multi-connector boards. The card occupies a dual-slot form factor, meaning it will require two expansion slots worth of space in the chassis, which is standard for cards in this performance class with active cooling solutions.

The 200 W TDP combined with the 550 W PSU recommendation suggests a power efficiency profile that is neither exceptionally frugal nor overly demanding. In a workstation environment, this allows for deployment in systems with moderate power budgets, leaving ample headroom for high-core-count CPUs and multiple storage drives. The boost clock is locked at 2400 MHz, identical to the base clock, which indicates a fixed-frequency operation profile rather than a dynamic boost behavior. This could imply consistent power draw characteristics across varied workloads, making thermal and power planning more predictable for system integrators.

Ray Tracing and Feature Set, RT/tensor cores, API support from facts

The Arc Pro B65 is equipped with 20 dedicated ray tracing cores, enabling hardware-accelerated ray tracing workloads. In professional contexts, this accelerates tasks such as photorealistic rendering, architectural visualization, and product design previews where light transport simulation is critical. The presence of dedicated RT hardware means the card can offload these computations from the shader array, improving performance in ray-traced workflows compared to software-based implementations.

API support is comprehensive for modern graphics workloads. The card supports DirectX 12 Ultimate with feature level 12_2, which encompasses the full DirectX Raytracing (DXR) tier, mesh shaders, variable rate shading, and sampler feedback. This places it in the current generation of graphics API compatibility. Vulkan 1.4 support provides cross-platform access to advanced rendering features, including ray tracing extensions where applicable. OpenGL 4.6 is also supported, ensuring compatibility with legacy professional applications that have not yet transitioned to newer APIs.

The Xe2-HPG architecture underpins this feature set, representing Intel's second-generation high-performance graphics architecture. While tensor core counts are not specified in the data, the architecture's ray tracing implementation is integrated into the GPU's compute units. The combination of 20 RT cores and 2560 shading units suggests a balanced approach where ray tracing acceleration is available without compromising traditional rasterization throughput. For professionals working in DCC applications like Blender, Maya, or Cinema 4D, the RT cores provide acceleration for viewport rendering and final frame rendering when ray tracing is enabled.

Memory Subsystem, VRAM size/type, bus width, bandwidth and what it means for high resolutions

The Arc Pro B65 ships with 32 GB of GDDR6 memory across a 256-bit bus interface. This configuration yields a memory bandwidth of 608.0 GB/s, a figure that positions the card for high-resolution and large-dataset workloads. The memory operates at 2375 MHz, which translates to 19 Gbps effective data rate.

The 32 GB capacity is a defining characteristic for professional use cases. Large 3D scenes with high-resolution textures, complex CAD assemblies, or multi-layer compositing projects in video editing can easily exceed 16 GB of VRAM usage. With 32 GB, the card can hold substantially larger working sets in memory, reducing the need for data streaming from system RAM or storage. This is particularly impactful for GPU-accelerated rendering, where the entire scene geometry and texture cache must reside in VRAM for optimal performance.

The 256-bit bus width and resulting 608.0 GB/s bandwidth provide the data throughput necessary to feed the 2560 shading units and 20 RT cores. At 4K resolution and above, texture fetches and framebuffer operations demand high memory bandwidth to avoid stalls. The bandwidth figure here is competitive for the card's performance class, ensuring that memory bandwidth does not become a bottleneck in most professional workloads. For multi-display configurations driving four DisplayPort 2.1 outputs, the memory subsystem can handle the framebuffer requirements for each display simultaneously.

The 32 GB capacity also enables larger compute workloads, such as machine learning inference or scientific simulations, where datasets must be resident in GPU memory. The GDDR6 memory type offers a balance between cost and performance, providing sufficient bandwidth for most professional tasks without the premium associated with HBM memory stacks.

How It Compares, position vs each nearest rival

The Arc Pro B65 currently has no listed nearest rivals in the benchmark database. This absence of comparative data makes it challenging to position the card against specific competing products from AMD or NVIDIA. The percentile rank of 50 against all GPUs indicates that the card sits at the median performance level in the database, meaning half of all tracked GPUs score higher and half score lower.

Without rival benchmarks, the comparison must rely on architectural positioning. The card's specifications, 32 GB memory, 2560 shading units, and 200 W TDP, suggest it targets the professional workstation segment rather than consumer gaming. In this segment, memory capacity and compute precision often matter more than raw frame rates. The 50th percentile ranking implies that while not a top-tier performer, it is firmly in the middle of the performance distribution, which could represent a balanced option for general professional use.

The lack of benchmark scores (average score of 0) and empty nearest rivals field means the database has not yet accumulated sufficient performance data for this card. This is common for newly released or niche professional products. The production status is listed as "Active," and the release date is March 31, 2026, suggesting the card is in early availability. As more benchmark results are submitted, the comparative positioning will become clearer.

Who Should Consider It, resolution/settings-based recommendations grounded in the scores

Given the 50th percentile ranking and the 32 GB memory capacity, the Arc Pro B65 is suited for professionals who prioritize memory capacity over peak compute throughput. At 1080p and 1440p resolutions, the card's 12.29 TFLOPS of FP32 compute and 608.0 GB/s bandwidth are more than adequate for demanding professional applications. At 4K resolution, the memory bandwidth becomes a more critical factor, and the 608.0 GB/s figure supports smooth interaction with high-resolution textures and complex scenes.

For GPU-accelerated rendering, the 32 GB memory capacity allows for scenes that would exceed the VRAM limits of 16 GB cards. This makes the card a candidate for artists working on large product visualizations, architectural flythroughs, or cinematic pre-visualization where texture detail and geometry complexity are high. The 20 RT cores provide acceleration for ray-traced rendering, though the 12.29 TFLOPS FP32 performance suggests that final-frame renders may be slower than higher-end workstation cards.

Video editors working with 6K or 8K timelines will benefit from the 32 GB capacity, as preview caches and effect stacks can consume significant VRAM. The four DisplayPort 2.1 outputs support multi-monitor setups common in editing bays or financial trading floors where multiple high-resolution displays are used simultaneously. The 200 W TDP keeps power requirements modest, making the card suitable for workstations where multiple GPUs might be installed.

The card is not positioned for high-refresh-rate gaming, as the 50th percentile ranking indicates mid-range performance at best. However, for professionals whose primary workload is not frame-rate-sensitive, the Arc Pro B65 offers a compelling combination of memory capacity and feature support.

FAQ, 4-6 Q&A pairs, each answerable from FACT PACK data

Q: What is the memory bandwidth of the Intel Arc Pro B65?

A: The card features 608.0 GB/s of memory bandwidth, achieved through a 256-bit bus interface with 32 GB of GDDR6 memory running at an effective 19 Gbps.

Q: Does the Arc Pro B65 support hardware ray tracing?

A: Yes, the card includes 20 dedicated ray tracing cores, which accelerate ray-traced workloads in compatible applications. These RT cores operate within the Xe2-HPG architecture.

Q: What is the recommended power supply wattage for this card?

A: The suggested PSU rating is 550 W. The card itself has a 200 W TDP and requires a single 8-pin power connector.

Q: Which display outputs are available on the Arc Pro B65?

A: The card provides four DisplayPort 2.1 outputs, enabling multi-monitor configurations with up to four displays connected directly.

Q: What is the transistor count and die size of the BMG-G21 chip?

A: The BMG-G21 chip contains 19,600 million transistors on a 272 mm² die, fabricated on TSMC's 5 nm process, yielding a transistor density of 72.1 million per square millimeter.

Q: Does the Arc Pro B65 support DirectX 12 Ultimate?

A: Yes, the card supports DirectX 12 Ultimate with feature level 12_2, along with Vulkan 1.4 and OpenGL 4.6 API support.

Benchmark Performance, analyze scores vs rivals with exact % deltas

The Arc Pro B65 currently has no benchmark scores recorded in the database, with an average benchmark score of 0 and no nearest rivals listed. This absence of quantitative performance data presents a challenge for analysis. The percentile rank of 50 against all GPUs provides a relative positioning, indicating that the card sits exactly at the median of all tracked graphics cards. This percentile is derived from historical data on similar products or from partial submissions, but the specific benchmark results that contributed to this ranking are not itemized.

Without rival delta percentages, we cannot state precise performance advantages or deficits relative to competing products. The 12.29 TFLOPS FP32 compute rate and 384.0 GTexel/s texture fill rate provide theoretical peak figures, but real-world application performance depends on driver optimization, workload characteristics, and thermal behavior. The 192.0 GPixel/s pixel fill rate suggests capable rasterization throughput for the card's class.

The lack of benchmark data may be due to the card's recent release date of March 31, 2026. Professional graphics cards often take time to accumulate benchmark submissions as they ship to workstation OEMs and system integrators. The 50th percentile ranking, however, indicates that when compared to the entire GPU landscape, including consumer gaming cards, the Arc Pro B65 performs at the midpoint. This implies that for professional tasks requiring 32 GB memory, the card provides a reasonable balance of compute and memory capabilities.

The FP16 compute rate of 24.58 TFLOPS (at 2:1 ratio) indicates that the card can handle half-precision workloads at twice the FP32 rate. This is relevant for machine learning inference and certain scientific computing tasks where FP16 precision is sufficient. The absence of a tensor core count in the specifications suggests that AI acceleration may rely on the shader-based FP16 path rather than dedicated tensor hardware.

Architecture and Design, chip, node, transistor count, core configuration

The Arc Pro B65 is built on the BMG-G21 chip, which is fabricated on TSMC's 5 nm process node. This manufacturing process enables the integration of 19,600 million transistors on a die measuring 272 mm², resulting in a transistor density of 72.1 million per square millimeter. The 5 nm node represents a mature manufacturing process that balances power efficiency with density, allowing the 200 W TDP to support the card's feature set.

The chip's architecture is Xe2-HPG, Intel's second-generation high-performance graphics architecture, and this card belongs to the Battlemage generation, specifically the Pro Series. The Xe2-HPG architecture builds upon the foundation established by the first-generation Xe-HPG architecture, with improvements in ray tracing performance and compute efficiency. The BMG-G21 chip is a full configuration, featuring 2560 shading units arranged across the GPU's compute slices.

The core configuration includes 160 texture mapping units (TMUs) and 80 render output units (ROPs). This distribution provides a texture fill rate of 384.0 GTexel/s and a pixel fill rate of 192.0 GPixel/s. The 20 ray tracing cores are integrated alongside the shading units, with each RT core handling bounding volume hierarchy traversal and ray-primitive intersection tests. The 2560 shading units deliver 12.29 TFLOPS of FP32 compute, with FP16 performance reaching 24.58 TFLOPS at a 2:1 ratio.

The card operates at a fixed 2400 MHz clock speed for both base and boost states, which simplifies thermal management and ensures consistent performance across workloads. The memory subsystem uses 32 GB of GDDR6 across a 256-bit bus, with memory clocked at 2375 MHz for 19 Gbps effective data rate. The combination of the 5 nm process and 200 W TDP indicates a power efficiency that is competitive for the performance class, and the dual-slot cooling solution is adequate for the thermal output.

The bus interface is PCIe 5.0 x16, providing 128 GB/s of bidirectional bandwidth between the GPU and host system. This high-bandwidth interface is beneficial for workloads that involve significant data transfer between system memory and VRAM, such as when streaming geometry or textures from storage. The four DisplayPort 2.1 outputs support the latest display standard, enabling high refresh rates at 4K and above, as well as multi-display configurations without daisy-chaining.