Intel Arc Pro B65 vs NVIDIA GeForce RTX 5090 SE Comparison

Where Each One Wins

The Intel Arc Pro B65 and the NVIDIA GeForce RTX 5090 SE occupy entirely different segments of the GPU market, and the benchmark data reflects that split clearly. The Arc Pro B65 is a professional workstation card built around the Xe2-HPG architecture, specifically the BMG-G21 chip. It prioritizes compute capacity, memory size, and stability over raw frame throughput. The RTX 5090 SE, on the other hand, is a high-end consumer GeForce 50-series part using the Blackwell 2.0 architecture with the GB202 chip. Its design targets maximum rendering performance and high-refresh gaming workloads.

The recorded data shows that the Arc Pro B65 wins in scenarios that demand large memory allocations. It ships with 32 GB of GDDR6 memory on a 256-bit bus, delivering 608.0 GB/s of bandwidth. That capacity is double the 24 GB found on the RTX 5090 SE, which uses GDDR7 memory on a 384-bit bus for 1.34 TB/s. For tasks like complex 3D scene rendering, large dataset visualization, or multi-application professional workflows, the Arc Pro B65 has a clear advantage in memory footprint. The RTX 5090 SE cannot match that capacity, so any workload that exceeds 24 GB will favor the Intel card.

The RTX 5090 SE wins in raw compute throughput. Its FP32 performance is 66.94 TFLOPS, compared to 12.29 TFLOPS for the Arc Pro B65. That is a massive gap. The NVIDIA card also leads in texture rate with 1,045.9 GTexel/s versus 384.0 GTexel/s, and pixel rate with 380.3 GPixel/s versus 192.0 GPixel/s. These numbers indicate that the RTX 5090 SE is the stronger choice for any workload where raw shading power matters, including high-resolution gaming, real-time ray tracing, and GPU-accelerated rendering with heavy shader complexity.

The architecture differences reinforce this split. The RTX 5090 SE has 14,080 shading units, 440 texture mapping units, 160 raster operations pipelines, 110 ray tracing cores, and 440 tensor cores. The Arc Pro B65 has 2,560 shading units, 160 TMUs, 80 ROPs, and 20 ray tracing cores, with no tensor core count listed. That disparity in core counts explains the compute gap. The NVIDIA card is also denser, with 122.9M transistors per mm² on a 750 mm² die, versus 72.1M per mm² on a 272 mm² die for the Intel part.

The Verdict

The data indicates that the RTX 5090 SE is the superior performer in absolute terms across nearly every compute metric. Its FP32 throughput is roughly 5.4 times higher than the Arc Pro B65. Its texture rate is about 2.7 times higher, and its pixel rate is about 2 times higher. The RTX 5090 SE also has more than 5 times the shading units and 5.5 times the ray tracing cores. For any user prioritizing raw performance, the choice is unambiguous.

The Arc Pro B65, however, offers a different set of strengths. Its 32 GB memory capacity is the standout feature. The RTX 5090 SE has 24 GB, which is still substantial but 8 GB less. For applications that need to hold very large textures, models, or simulation data in VRAM, the Intel card is the only option that fits. The Arc Pro B65 also has a much lower power draw at 200 W TDP versus 500 W TDP for the NVIDIA card, and its 8-pin power connector is more flexible than the 16-pin connector required by the RTX 5090 SE. The suggested PSU for the Intel card is 550 W, while the NVIDIA card recommends a 900 W unit.

The production status for both cards is listed as Active. The RTX 5090 SE has a launch MSRP of 1,499 USD. The Arc Pro B65 has no launch MSRP listed. Both cards use PCIe 5.0 x16 interfaces and support DirectX 12 Ultimate, OpenGL 4.6, and Vulkan 1.4.

For a user who needs maximum memory capacity in a workstation environment with modest power requirements, the Arc Pro B65 is the appropriate pick. For a user who needs maximum compute performance, the RTX 5090 SE is the clear choice.

Head-to-Head Benchmarks

The head-to-head benchmark table is empty in the recorded data, meaning there are no direct comparative test scores available. However, the specification-level comparisons provide a clear picture of where each card wins.

The largest win for the RTX 5090 SE is in FP32 compute. The NVIDIA card delivers 66.94 TFLOPS, which is 54.65 TFLOPS more than the Arc Pro B65's 12.29 TFLOPS. That is a 5.45 times advantage. The FP16 comparison is similarly lopsided, with the RTX 5090 SE achieving 66.94 TFLOPS (1:1 ratio) versus 24.58 TFLOPS (2:1 ratio) for the Intel card.

Memory bandwidth is another decisive win for the NVIDIA card. With 1.34 TB/s versus 608.0 GB/s, the RTX 5090 SE has more than twice the bandwidth. This matters for high-resolution textures and memory-intensive rendering passes. The NVIDIA card also uses GDDR7 memory, while the Intel card uses GDDR6.

The RTX 5090 SE wins in texture rate by a factor of 2.72, with 1,045.9 GTexel/s versus 384.0 GTexel/s. Its pixel rate advantage is smaller but still significant: 380.3 GPixel/s versus 192.0 GPixel/s, a 1.98 times difference.

The Arc Pro B65 wins in memory capacity. Its 32 GB is 8 GB more than the RTX 5090 SE's 24 GB. That is a 33% capacity advantage. For workloads that spill beyond 24 GB, the Intel card will complete tasks that the NVIDIA card cannot start without swapping to system memory.

The Arc Pro B65 also wins on power efficiency. At 200 W TDP, it draws 300 W less than the RTX 5090 SE's 500 W TDP. The suggested PSU rating is 550 W versus 900 W, a 350 W difference. This makes the Intel card substantially easier to integrate into existing systems without upgrading the power supply.

The RTX 5090 SE has a smaller physical footprint at 267 mm in length, 111 mm in height, and 40 mm in width. The Arc Pro B65 has no listed dimensions, so a direct size comparison is not possible from the data.

FAQ

Q: Which card has more memory?

A: The Intel Arc Pro B65 has 32 GB of GDDR6 memory, while the NVIDIA GeForce RTX 5090 SE has 24 GB of GDDR7 memory. The Intel card has an 8 GB capacity advantage.

Q: Which card has higher FP32 performance?

A: The NVIDIA GeForce RTX 5090 SE achieves 66.94 TFLOPS, which is more than 5 times the 12.29 TFLOPS of the Intel Arc Pro B65.

Q: What are the power requirements for each card?

A: The Intel Arc Pro B65 has a 200 W TDP and requires a 550 W suggested PSU with a single 8-pin power connector. The NVIDIA GeForce RTX 5090 SE has a 500 W TDP and requires a 900 W suggested PSU with a single 16-pin power connector.

Q: Do both cards support the same APIs?

A: Yes, both cards support DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4.

Q: Which card has more ray tracing cores?

A: The NVIDIA GeForce RTX 5090 SE has 110 ray tracing cores, while the Intel Arc Pro B65 has 20. The NVIDIA card has 5.5 times more ray tracing hardware.

Q: What memory bandwidth does each card provide?

A: The NVIDIA GeForce RTX 5090 SE provides 1.34 TB/s of bandwidth, while the Intel Arc Pro B65 provides 608.0 GB/s. The NVIDIA card has more than twice the bandwidth.

Architecture Differences

The two cards are built on fundamentally different architectures. The Intel Arc Pro B65 uses Xe2-HPG architecture on a chip called BMG-G21. This chip is manufactured on a 5 nm process at TSMC and contains 19,600 million transistors on a 272 mm² die. The transistor density is 72.1M per mm². The card belongs to the Battlemage (Pro Series) generation.

The NVIDIA GeForce RTX 5090 SE uses Blackwell 2.0 architecture on a chip called GB202. It is also manufactured on a 5 nm process at TSMC but contains 92,200 million transistors on a 750 mm² die. The transistor density is 122.9M per mm². The card belongs to the GeForce 50 generation and has a predecessor listed as GeForce 40 and a successor listed as GeForce 60.

The core configuration differs dramatically. The RTX 5090 SE has 14,080 shading units, 440 TMUs, 160 ROPs, 110 RT cores, and 440 tensor cores. The Arc Pro B65 has 2,560 shading units, 160 TMUs, 80 ROPs, and 20 RT cores, with no tensor core count listed. This means the NVIDIA card has 5.5 times more shading units, 2.75 times more TMUs, 2 times more ROPs, and 5.5 times more RT cores.

Clock speeds also differ. The Intel card has a base clock and boost clock both at 2400 MHz. The NVIDIA card has a base clock of 1740 MHz and a boost clock of 2377 MHz. The Intel card has a higher sustained clock, but the NVIDIA card compensates with far more execution units. Memory clocks are 2375 MHz (19 Gbps effective) for the Intel card and 1750 MHz (28 Gbps effective) for the NVIDIA card.

The memory subsystems reflect different priorities. The Arc Pro B65 uses 32 GB of GDDR6 on a 256-bit bus. The RTX 5090 SE uses 24 GB of GDDR7 on a 384-bit bus. The wider bus and newer memory type give the NVIDIA card its 1.34 TB/s bandwidth, while the Intel card's narrower bus limits it to 608.0 GB/s despite the larger capacity.

Power delivery and physical design differ as well. The Intel card uses a single 8-pin power connector, while the NVIDIA card uses a single 16-pin connector. The RTX 5090 SE is listed as a dual-slot card with dimensions of 267 mm in length, 111 mm in height, and 40 mm in width. The Arc Pro B65 is also dual-slot but has no listed dimensions. Display outputs differ, with the Arc Pro B65 offering 4x DisplayPort 2.1 and the RTX 5090 SE offering 1x HDMI 2.1b and 3x DisplayPort 2.1b.

The FP16 capability also diverges. The Intel card achieves 24.58 TFLOPS at a 2:1 ratio, meaning it halves throughput for FP16. The NVIDIA card achieves 66.94 TFLOPS at a 1:1 ratio, meaning it maintains full throughput for FP16. This makes the RTX 5090 SE particularly strong for workloads that leverage FP16 compute, including certain AI inference and rendering tasks.