Intel Arc B390 vs NVIDIA GB10 Comparison

Head-to-Head Benchmarks

The recorded data shows no direct head-to-head benchmark runs between the Intel Arc B390 and the NVIDIA GB10. Instead, the database places both parts into separate performance tiers based on their respective test suites, and the gap between those tiers is substantial.

The Intel Arc B390 was measured in a single DirectX 12 test, 3DMark Steel Nomad, where it scored 1482. That result places it at the 9th percentile of all GPUs in the database, meaning roughly nine percent of recorded graphics processors score lower. Its nearest rivals in the database are all older, low-end NVIDIA parts. The GeForce GT 520MX scores 1463, which is 1.3 percent behind the Arc B390. The GeForce 800M trails by 1.5 percent at 1460. The GeForce GT 625 OEM sits at 1446, a 2.5 percent deficit, and the GeForce GT 710 scores 1443, 2.7 percent lower. These deltas show the Arc B390 is not dramatically faster than its closest neighbors; it leads them by single-digit percentage points, and the margin between first and fourth place in that cluster is under three percent.

The NVIDIA GB10, by contrast, was measured with Geekbench OpenCL and Vulkan workloads, scoring 120137 and 114648 respectively, for an average of 117393. That lands it at the 95th percentile of all GPUs, a completely different performance stratum. Its nearest rivals are professional and workstation-class parts. The RTX 4000 SFF Ada Generation scores 117088, only 0.3 percent behind the GB10. The AMD Radeon PRO W7700 sits 1.3 percent ahead at 118976. The Tesla V100 SXM2 16 GB trails by 2.6 percent at 114395, and the RTX A5500 Mobile is 3.0 percent behind at 113944. The GB10 is effectively neck-and-neck with these high-end workstation accelerators, with all four rivals within a three percent band.

Comparing the two headline numbers directly, the GB10's average benchmark score of 117393 is roughly 79 times higher than the Arc B390's 1482. No single test appears in both benchmark lists, so the database does not provide a like-for-like frame rate comparison. However, the percentile placement tells the story: the Arc B390 sits among legacy entry-level mobile parts, while the GB10 competes with current-generation workstation GPUs. The wins column confirms this asymmetry, with the database recording zero wins for the Arc B390 and zero wins for the GB10, since no head-to-head tests exist.

The Verdict

The data indicates these two products are not competitors in any meaningful sense. A user selecting between them would be choosing between fundamentally different hardware classes.

The Intel Arc B390 is an integrated graphics processor embedded in a Panther Lake chip, built on Intel's 3 nm process with the Xe3-LPG architecture. Its 1482 Steel Nomad score and 9th percentile ranking place it squarely in the entry-level segment, alongside GPUs from a decade ago. It delivers 7.680 TFLOPS of FP32 compute, 60.00 GPixel/s of pixel throughput, and 120.0 GTexel/s of texture fill. It has no dedicated memory, relying on system shared RAM with bandwidth described as system dependent.

The NVIDIA GB10 is a server-class Blackwell 2.0 part with 29.71 TFLOPS of FP32 compute, 116.1 GPixel/s pixel rate, and 928.5 GTexel/s texture rate. Its 273.2 GB/s of memory bandwidth from 128 GB of LPDDR5X on a 256-bit bus places it in an entirely different performance envelope. The 95th percentile ranking and its proximity to the RTX 4000 SFF Ada Generation (0.3 percent ahead) and Radeon PRO W7700 (1.3 percent behind) confirm it belongs in professional compute territory.

From the benchmark data alone, the GB10 is the clear choice for any workload that requires substantial compute throughput, large memory capacity, or high-bandwidth access. The Arc B390 is suitable only for basic graphics output on a portable device, where its IGP form factor and 80 W TDP make it appropriate for lightweight tasks. The GB10 carries a 140 W TDP and a 300 W suggested PSU, uses a PCIe 5.0 x16 interface, and measures 150 mm by 51 mm by 150 mm. These physical and power characteristics reflect a part designed for sustained, demanding workloads.

Where Each One Wins

The Intel Arc B390 wins in scenarios where power draw and physical footprint are the primary constraints. Its 80 W TDP is well below the GB10's 140 W, and its IGP slot width means it requires no power connectors and no dedicated expansion slot. It supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4, making it a functional graphics solution for a portable device with system shared memory. Its display outputs are listed as portable device dependent, reinforcing its role as an integrated solution rather than a discrete card.

The NVIDIA GB10 wins on every raw performance metric recorded. Its FP32 throughput of 29.71 TFLOPS is roughly 3.9 times the Arc B390's 7.680 TFLOPS. Its texture rate of 928.5 GTexel/s is about 7.7 times higher than 120.0 GTexel/s. Its pixel rate of 116.1 GPixel/s nearly doubles the Arc B390's 60.00 GPixel/s. The GB10 also has 6144 shading units versus 1536, 384 texture mapping units versus 48, 48 ROPs versus 24, 48 RT cores versus 12, and 384 tensor cores where the Arc B390 lists none. Memory is the starkest divide: the GB10 has 128 GB of dedicated LPDDR5X with 273.2 GB/s bandwidth, while the Arc B390 has no dedicated memory at all.

For compute-intensive workloads such as OpenCL or Vulkan rendering, the database shows the GB10 is the only viable option. Its Geekbench OpenCL score of 120137 and Vulkan score of 114648 are orders of magnitude beyond anything the Arc B390 could produce in an equivalent test, given the 9th percentile placement. The GB10's predecessor is listed as Server Hopper and its successor as Server Rubin, confirming it sits in a product line aimed at server deployment, not consumer portable devices.

FAQ

Q: How much faster is the NVIDIA GB10 than the Intel Arc B390?

A: The GB10 has an average benchmark score of 117393, while the Arc B390 scores 1482 in its single recorded test. The GB10 sits at the 95th percentile of all GPUs, while the Arc B390 sits at the 9th percentile.

Q: What are the closest competitors to the Intel Arc B390?

A: The nearest rivals are the NVIDIA GeForce GT 520MX (1463, 1.3 percent behind), GeForce 800M (1460, 1.5 percent behind), GeForce GT 625 OEM (1446, 2.5 percent behind), and GeForce GT 710 (1443, 2.7 percent behind).

Q: What are the closest competitors to the NVIDIA GB10?

A: The nearest rivals are the NVIDIA RTX 4000 SFF Ada Generation (117088, 0.3 percent behind), AMD Radeon PRO W7700 (118976, 1.3 percent ahead), Tesla V100 SXM2 16 GB (114395, 2.6 percent behind), and RTX A5500 Mobile (113944, 3.0 percent behind).

Q: Does the Intel Arc B390 have dedicated memory?

A: No. Its memory size, type, and bus width are all listed as system shared, and its bandwidth is system dependent. The NVIDIA GB10 has 128 GB of LPDDR5X on a 256-bit bus with 273.2 GB/s bandwidth.

Q: What API support does each GPU offer?

A: The Intel Arc B390 supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The NVIDIA GB10 lists N/A for DirectX, OpenGL, and Vulkan, reflecting its server-oriented design without standard graphics API support.

Q: What is the production status and release timeline for these parts?

A: Both are listed as active. The Intel Arc B390 was released on 2026-01-26, and the NVIDIA GB10 was released on 2025-10-14.

Architecture Differences

The Intel Arc B390 uses the Xe3-LPG architecture built on Intel's 3 nm process, with the chip codenamed Panther Lake. It belongs to the Arc Graphics-M (Panther Lake) generation. The NVIDIA GB10 uses the Blackwell 2.0 architecture on a 5 nm process from TSMC, with the chip designated GB20B and the generation listed as Server Blackwell (Bxx).

The compute layouts differ substantially. The Arc B390 has 1536 shading units, 48 texture mapping units, 24 ROPs, and 12 RT cores. The GB10 has 6144 shading units, 384 TMUs, 48 ROPs, 48 RT cores, and 384 tensor cores. The GB10's tensor core count indicates a machine-learning-oriented design, while the Arc B390 lists no tensor cores at all.

Clock behavior also differs. The Arc B390 has a base clock of 300 MHz and a boost clock of 2500 MHz. The GB10 has a base clock of 1665 MHz and a boost clock of 2418 MHz. Despite the lower base clock, the Arc B390's higher boost clock does not compensate for its far smaller shader array. The GB10's memory clock is listed as 1067 MHz with 8.5 Gbps effective, while the Arc B390 has no dedicated memory clock, using system shared memory.

The GB10 has a die size of 382 mm², while the Arc B390's die size is unknown. Both use different foundries: Intel for the Arc B390, TSMC for the GB10. The GB10's bus interface is PCIe 5.0 x16, while the Arc B390 uses an IGP bus interface with no power connectors. The GB10 also has no power connectors despite its 140 W TDP and 300 W suggested PSU, which is consistent with a server board that supplies power through the slot. The Arc B390's 80 W TDP and IGP form factor mean it draws power from the host processor package.

Display outputs differ as well. The Arc B390's outputs are portable device dependent, while the GB10 has a single HDMI output. The GB10's physical dimensions are 150 mm by 51 mm by 150 mm, while the Arc B390 has no listed dimensions, consistent with an integrated part. The GB10 lists a launch MSRP of 3,999 USD. The Arc B390 has no launch MSRP, reflecting its integrated nature.

The FP16 compute rates also reveal architectural intent. The Arc B390 delivers 15.36 TFLOPS FP16 at a 2:1 ratio relative to its 7.680 TFLOPS FP32. The GB10 delivers 29.71 TFLOPS FP16 at a 1:1 ratio, meaning its FP16 throughput matches FP32 exactly. This difference indicates the GB10's Blackwell architecture is designed for workloads that require equal precision in both formats, a common trait in server compute accelerators. The Arc B390's 2:1 FP16 ratio is typical of consumer graphics parts where FP16 is used for visual effects rather than general compute.