Intel Arc Pro B390 vs NVIDIA GeForce RTX 5090 D V2 Comparison

The Verdict

The Intel Arc Pro B390 and NVIDIA GeForce RTX 5090 D V2 occupy completely different positions in the hardware landscape. The database shows the RTX 5090 D V2 with a benchmark score of 16,504 in 3DMark Steel Nomad DX12, placing it in the 59th percentile of all GPUs. The Arc Pro B390 has no recorded benchmark scores and sits at the 50th percentile, indicating it is not designed for the same workload class.

The RTX 5090 D V2 is a discrete, dual-slot add-in card with 24 GB of GDDR7 memory, a 384-bit bus, and 1.34 TB/s of bandwidth. It targets high-end desktop workloads, with a 575 W TDP and a 950 W suggested PSU. The Arc Pro B390 is an integrated graphics processor (IGP) with no dedicated memory, relying entirely on system shared memory. Its 80 W TDP and lack of power connectors confirm it is meant for compact or portable systems where a discrete GPU cannot fit.

The data indicates the RTX 5090 D V2 is the only viable choice for any user needing discrete GPU performance. The Arc Pro B390, with its integrated nature and system-dependent memory bandwidth, cannot compete in raw throughput. However, for systems where power draw, physical space, and thermal constraints dominate, the Arc Pro B390 offers a functional graphics solution with modern API support. The RTX 5090 D V2 is for users with a full-size chassis, a robust PSU, and workloads that demand high-end rendering or compute. The Arc Pro B390 is for users building ultra-compact or portable devices where a discrete GPU is physically impossible.

Architecture Differences

The two GPUs come from different manufacturers, using different process nodes and architectures. The Intel Arc Pro B390 uses the Panther Lake chip with the Xe3-LPG architecture, built on a 3 nm process at Intel's foundry. It belongs to the Arc Graphics-WM (Panther Lake) generation, with its predecessor listed as HD Graphics-WM. The NVIDIA GeForce RTX 5090 D V2 uses the GB202 chip with the Blackwell 2.0 architecture, built on a 5 nm process at TSMC. It belongs to the GeForce 50-series, with its predecessor listed as GeForce 40 and successor as GeForce 60.

The RTX 5090 D V2 has a massive transistor count of 92,200 million on a 750 mm² die, giving a transistor density of 122.9M per mm². The Arc Pro B390's transistor count and die size are recorded as unknown in the database. This size difference is reflected in the shading units: the RTX 5090 D V2 has 21,760 shading units, while the Arc Pro B390 has 1,536. The RTX 5090 D V2 also carries 680 texture mapping units and 176 raster output pipelines, compared to 48 TMUs and 24 ROPs on the Arc Pro B390.

Ray tracing hardware differs significantly. The RTX 5090 D V2 has 170 RT cores and 680 tensor cores, while the Arc Pro B390 has 12 RT cores and no tensor cores recorded. The memory subsystem is entirely different: the RTX 5090 D V2 uses 24 GB of GDDR7 on a 384-bit bus with 1.34 TB/s bandwidth, while the Arc Pro B390 uses system shared memory with system-dependent bandwidth. The RTX 5090 D V2 has a dedicated 16-pin power connector and requires a 950 W suggested PSU, whereas the Arc Pro B390 has no power connectors and runs entirely off the system's integrated power delivery.

Clock behavior also diverges. The Arc Pro B390 has a low base clock of 300 MHz but boosts to 2,500 MHz, while the RTX 5090 D V2 has a base clock of 2,017 MHz and a boost of 2,407 MHz. The RTX 5090 D V2 also lists a memory clock of 1,750 MHz (28 Gbps effective), while the Arc Pro B390's memory clock is tied to system shared memory. Both support DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4.

Head-to-Head Benchmarks

The database contains only one recorded benchmark for the RTX 5090 D V2: a 3DMark Steel Nomad DX12 score of 16,504. The Arc Pro B390 has no benchmark entries, so no direct head-to-head comparison is possible from the recorded data. The RTX 5090 D V2's average benchmark score is 16,504, which matches its single result.

The RTX 5090 D V2's nearest rivals in the database provide context for its performance tier. The NVIDIA T400 scores 16,508, which is essentially identical (0% delta). The AMD Radeon PRO W7500 scores 16,415, putting the RTX 5090 D V2 0.5% ahead. The NVIDIA RTX PRO 6000 Blackwell scores 16,408, a 0.6% delta. The AMD Radeon RX 5700 XT scores 16,361, making the RTX 5090 D V2 0.9% faster. These deltas are all under 1%, indicating that the RTX 5090 D V2 sits in a tightly clustered performance band around the 16,400 to 16,500 score range in this specific test.

The Arc Pro B390's lack of benchmark data means the database cannot assign it a comparative position. Its 50th percentile ranking versus the RTX 5090 D V2's 59th percentile suggests the RTX 5090 D V2 outperforms a majority of all GPUs in the database, while the Arc Pro B390 sits at the median. Without a recorded score, the Arc Pro B390's actual performance relative to the RTX 5090 D V2 cannot be quantified. What the data does show is that the RTX 5090 D V2 delivers 104.8 TFLOPS of FP32 compute and 423.6 GPixel/s pixel throughput, while the Arc Pro B390 delivers 7.680 TFLOPS and 60.00 GPixel/s. The texture rates are 1,636.8 GTexel/s versus 120.0 GTexel/s, a factor of 13.6x difference in favor of the RTX 5090 D V2.

Specification Differences

The two GPUs differ in nearly every specification field. The Arc Pro B390 uses a 3 nm process at Intel, while the RTX 5090 D V2 uses a 5 nm process at TSMC. The RTX 5090 D V2 has 92,200 million transistors on a 750 mm² die; the Arc Pro B390's transistor count and die size are unknown. The Arc Pro B390 has a base clock of 300 MHz and boost of 2,500 MHz; the RTX 5090 D V2 has a base of 2,017 MHz and boost of 2,407 MHz.

Memory is fundamentally different: the Arc Pro B390 uses system shared memory with system-dependent bandwidth, while the RTX 5090 D V2 has 24 GB of GDDR7 on a 384-bit bus with 1.34 TB/s bandwidth. The Arc Pro B390's memory clock is listed as "System Shared" while the RTX 5090 D V2 runs at 1,750 MHz (28 Gbps effective).

Compute resources differ by an order of magnitude. The Arc Pro B390 has 1,536 shading units, 48 TMUs, 24 ROPs, and 12 RT cores. The RTX 5090 D V2 has 21,760 shading units, 680 TMUs, 176 ROPs, 170 RT cores, and 680 tensor cores. The Arc Pro B390 has no tensor cores recorded.

The RTX 5090 D V2 delivers 104.8 TFLOPS FP32 and 104.8 TFLOPS FP16 (1:1 ratio). The Arc Pro B390 delivers 7.680 TFLOPS FP32 and 15.36 TFLOPS FP16 (2:1 ratio). Pixel rates are 423.6 GPixel/s versus 60.00 GPixel/s, and texture rates are 1,636.8 GTexel/s versus 120.0 GTexel/s.

Power and physical specifications diverge sharply. The Arc Pro B390 has an 80 W TDP, is an IGP with no power connectors, and has no suggested PSU. The RTX 5090 D V2 has a 575 W TDP, is dual-slot, uses one 16-pin connector, and requires a 950 W suggested PSU. The Arc Pro B390 uses an IGP bus interface, while the RTX 5090 D V2 uses PCIe 5.0 x16. The RTX 5090 D V2 measures 304 mm by 137 mm by 48 mm; the Arc Pro B390 has no recorded dimensions.

Display outputs differ: the Arc Pro B390 is "Portable Device Dependent" while the RTX 5090 D V2 has 1x HDMI 2.1b and 3x DisplayPort 2.1b. Release dates also differ: the Arc Pro B390 launched on 2026-01-26, while the RTX 5090 D V2 launched on 2025-08-14. The RTX 5090 D V2 has a launch MSRP of 2,299 USD; the Arc Pro B390 has no launch MSRP recorded. Both support DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4.

FAQ

Q: Which GPU has more raw compute performance?

A: The RTX 5090 D V2 delivers 104.8 TFLOPS of FP32 compute, compared to 7.680 TFLOPS for the Arc Pro B390. The RTX 5090 D V2 also has 21,760 shading units versus 1,536, and 680 TMUs versus 48.

Q: Can the Arc Pro B390 be installed in a desktop PC as a discrete card?

A: No. The Arc Pro B390 is an IGP with a slot width of "IGP" and no power connectors. It uses the IGP bus interface and has no recorded dimensions, indicating it is integrated into a processor rather than a separate card.

Q: What memory does each GPU use?

A: The Arc Pro B390 uses system shared memory with system-dependent bandwidth. The RTX 5090 D V2 uses 24 GB of GDDR7 on a 384-bit bus with 1.34 TB/s bandwidth and a memory clock of 1,750 MHz (28 Gbps effective).

Q: How does the RTX 5090 D V2 compare to its nearest rivals in the database?

A: The RTX 5090 D V2 scores 16,504 in 3DMark Steel Nomad DX12. The NVIDIA T400 scores 16,508 (0% delta), the AMD Radeon PRO W7500 scores 16,415 (0.5% slower), the NVIDIA RTX PRO 6000 Blackwell scores 16,408 (0.6% slower), and the AMD Radeon RX 5700 XT scores 16,361 (0.9% slower).

Q: What is the power requirement difference?

A: The Arc Pro B390 has an 80 W TDP and no power connectors. The RTX 5090 D V2 has a 575 W TDP, uses one 16-pin power connector, and requires a 950 W suggested PSU.

Q: Which GPU supports newer display outputs?

A: The RTX 5090 D V2 has 1x HDMI 2.1b and 3x DisplayPort 2.1b. The Arc Pro B390's display outputs are listed as "Portable Device Dependent," meaning they depend on the portable device it is integrated into.

Where Each One Wins

The RTX 5090 D V2 wins in every measurable performance category. Its FP32 throughput of 104.8 TFLOPS is over 13 times higher than the Arc Pro B390's 7.680 TFLOPS. Its texture rate of 1,636.8 GTexel/s dwarfs the 120.0 GTexel/s of the Arc Pro B390. Its pixel rate of 423.6 GPixel/s is roughly 7 times the 60.00 GPixel/s of the Arc Pro B390. The RTX 5090 D V2 also has 170 RT cores and 680 tensor cores, enabling hardware ray tracing and tensor operations that the Arc Pro B390 cannot match with only 12 RT cores and no tensor cores.

The RTX 5090 D V2 wins in memory capacity and bandwidth. With 24 GB of GDDR7, a 384-bit bus, and 1.34 TB/s bandwidth, it provides dedicated high-speed memory. The Arc Pro B390's system shared memory is system-dependent, meaning its bandwidth varies with the host platform and is shared with CPU operations. This gives the RTX 5090 D V2 a clear advantage in memory-intensive workloads such as large texture datasets or high-resolution rendering.

The RTX 5090 D V2 wins in API support parity but also in physical connectivity. Both support DirectX 12 Ultimate, OpenGL 4.6, and Vulkan 1.4, but the RTX 5090 D V2 offers dedicated HDMI 2.1b and DisplayPort 2.1b outputs, while the Arc Pro B390's outputs depend on the portable device. The RTX 5090 D V2 also uses PCIe 5.0 x16, a high-bandwidth interface, while the Arc Pro B390 is limited to an IGP bus.

The Arc Pro B390 wins in power efficiency and physical footprint. Its 80 W TDP is vastly lower than the 575 W TDP of the RTX 5090 D V2. It requires no power connectors and no suggested PSU, whereas the RTX 5090 D V2 needs a 16-pin connector and a 950 W PSU. The Arc Pro B390 is an IGP, meaning it occupies no expansion slot and has no length, height, or width to accommodate. The RTX 5090 D V2 is a dual-slot card measuring 304 mm by 137 mm by 48 mm. For compact systems, embedded designs, or portable devices where a discrete GPU cannot physically fit, the Arc Pro B390 is the only option between the two.

The Arc Pro B390 also wins on release timing, launching on 2026-01-26, which is after the RTX 5090 D V2's 2025-08-14 release date. It uses a newer process node at 3 nm versus 5 nm, though this does not translate into performance advantages in the recorded data. The Arc Pro B390's integrated nature means it can be part of a system with no discrete GPU at all, reducing system complexity and avoiding the need for a dedicated power supply rating.

In practical terms, the RTX 5090 D V2 is the clear choice for any workload requiring discrete GPU performance: high-end gaming, 3D rendering, AI inference with tensor cores, or any task that benefits from 24 GB of dedicated GDDR7 memory. The Arc Pro B390 is suited for systems where the GPU must be integrated into the processor, such as thin-and-light laptops, mini PCs, or embedded devices, and where the 80 W TDP fits within the system's thermal and power budget. The database's benchmark data only covers the RTX 5090 D V2, showing its score of 16,504 in 3DMark Steel Nomad DX12, which places it near several rivals within a 0.9% performance band. The Arc Pro B390 has no benchmark record, so its actual performance remains unquantified, but its specifications clearly position it as a low-power integrated solution rather than a discrete performance part.