Intel Arc B390 vs NVIDIA Jetson T5000 Comparison

Head-to-Head Benchmarks

The benchmark database shows a stark asymmetry between these two parts: the Intel Arc B390 has recorded benchmark results, while the NVIDIA Jetson T5000 has none. The only available score for the Arc B390 comes from 3dmark_3dmark_steel_nomad_dx12, where it posts 1482 points. This places the Arc B390 in the 9th percentile of all GPUs tracked by the database, a low standing that reflects its positioning among entry-level graphics solutions.

The nearest rivals for the Arc B390 in the database are all NVIDIA discrete mobile and OEM parts. The GeForce GT 520MX sits closest with an average score of 1463, putting the Arc B390 1.3% ahead. The GeForce 800M follows at 1460, a 1.5% gap. The GeForce GT 625 OEM averages 1446, 2.5% behind, and the GeForce GT 710 averages 1443, 2.7% behind. These deltas are narrow, indicating that the Arc B390's performance clusters tightly with these older, low-end parts rather than establishing any meaningful lead. The data shows a GPU that edges out its closest competition by single-digit percentage points, not a decisive victor.

For the Jetson T5000, the database contains no benchmark entries and no nearest rivals. Its average benchmark score is recorded as 0, and its percentile versus all GPUs is listed as 50, a placeholder value that does not reflect measured performance. The absence of data means no head-to-head comparison can be constructed from direct tests. Any comparative analysis must rely on the specification differences and the architectural characteristics recorded in the database, not on measured outcomes.

The compute throughput figures from the database offer a partial picture. The Arc B390 delivers 7.680 TFLOPS FP32 and 15.36 TFLOPS FP16 with a 2:1 ratio. The Jetson T5000 delivers 8.064 TFLOPS FP32 and 8.064 TFLOPS FP16 with a 1:1 ratio. In raw FP32, the Jetson T5000 is 5% ahead of the Arc B390. In FP16, the Arc B390 is 90% ahead, but this advantage comes with a caveat: the 2:1 ratio on the Arc B390 indicates that FP16 performance is achieved through packing two FP32 operations per clock, a common technique that does not necessarily translate to real-world gains across all workloads. The Jetson T5000's 1:1 FP16 ratio means its FP16 throughput is identical to its FP32 throughput, suggesting a different design priority.

FAQ

Q: Which GPU has a recorded benchmark score in the database?

A: Only the Intel Arc B390 has a benchmark entry, scoring 1482 points in 3dmark_3dmark_steel_nomad_dx12. The NVIDIA Jetson T5000 has no benchmark entries recorded.

Q: How does the Intel Arc B390 compare to its nearest rivals?

A: The Arc B390 leads the GeForce GT 520MX by 1.3%, the GeForce 800M by 1.5%, the GeForce GT 625 OEM by 2.5%, and the GeForce GT 710 by 2.7%.

Q: What is the memory configuration difference?

A: The Arc B390 uses system shared memory with a system dependent bus width and bandwidth. The Jetson T5000 has 128 GB of LPDDR5X memory on a 256-bit bus with 273.2 GB/s bandwidth.

Q: What are the clock speed differences?

A: The Arc B390 runs at a base clock of 300 MHz and boosts to 2500 MHz. The Jetson T5000 runs at a base clock of 1386 MHz and boosts to 1575 MHz.

Q: Which GPU has tensor cores?

A: The Jetson T5000 has 96 tensor cores. The Arc B390 has no tensor cores listed in the database.

Q: What is the process node difference?

A: The Arc B390 is manufactured on a 3 nm process at Intel. The Jetson T5000 is manufactured on a 5 nm process at TSMC.

Where Each One Wins

The Intel Arc B390 wins in situations where measured graphics performance matters and where the host system provides shared memory. Its recorded 3DMark Steel Nomad DX12 score of 1482, while modest, places it ahead of four comparable NVIDIA parts by margins from 1.3% to 2.7%. This indicates that for legacy or lightweight DX12 workloads, the Arc B390 delivers a measurable, if small, advantage over those specific rivals. Its boost clock of 2500 MHz is substantially higher than the Jetson T5000's 1575 MHz, and its FP16 output of 15.36 TFLOPS is double its FP32 rate, which can benefit workloads that exploit packed math. The Arc B390 also supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4, making it a functional graphics solution for portable devices with display outputs dependent on the host platform.

The NVIDIA Jetson T5000 wins in compute-oriented and memory-intensive scenarios. Its FP32 throughput of 8.064 TFLOPS exceeds the Arc B390's 7.680 TFLOPS. It carries 128 GB of LPDDR5X memory with 273.2 GB/s bandwidth, a fixed and substantial pool compared to the Arc B390's system dependent memory. The 256-bit bus width provides a wide path for data movement, and the 96 tensor cores enable accelerated AI inference workloads that the Arc B390 cannot handle. The Jetson T5000 also has 20 ray tracing cores versus 12 on the Arc B390, and a higher texture rate of 126.0 GTexel/s versus 120.0 GTexel/s. Its PCIe 5.0 x8 interface allows for high-bandwidth host communication, and its 300 W suggested PSU indicates a system designed for sustained compute rather than mobile efficiency.

The specification data points to a clear split: the Arc B390 is a graphics-first part for portable devices, while the Jetson T5000 is a compute-first module for server or edge deployments. The Arc B390 has the higher pixel rate at 60.00 GPixel/s versus 50.40 GPixel/s, and the higher boost clock, but the Jetson T5000 counters with more shading units (2560 versus 1536), more TMUs (80 versus 48), more ROPs (32 versus 24), and a larger die size at 391 mm².

Specification Differences

The two parts diverge across nearly every measurable specification. The Arc B390 uses the Intel Panther Lake chip with Xe3-LPG architecture, built on a 3 nm process at Intel. The Jetson T5000 uses the NVIDIA GB10B chip with Blackwell architecture, built on a 5 nm process at TSMC. The Arc B390 has a die size listed as unknown, while the Jetson T5000 has a die size of 391 mm².

Clock speeds differ significantly. The Arc B390 has a base clock of 300 MHz and a boost clock of 2500 MHz. The Jetson T5000 has a base clock of 1386 MHz and a boost clock of 1575 MHz, with memory clocked at 1067 MHz (8.5 Gbps effective). The Arc B390's memory is system shared with system dependent bandwidth, while the Jetson T5000 has 128 GB of LPDDR5X on a 256-bit bus with 273.2 GB/s bandwidth.

Compute resources differ in count and type. The Arc B390 has 1536 shading units, 48 TMUs, 24 ROPs, and 12 ray tracing cores, with no tensor cores. The Jetson T5000 has 2560 shading units, 80 TMUs, 32 ROPs, 20 ray tracing cores, and 96 tensor cores. The Arc B390 achieves a pixel rate of 60.00 GPixel/s and a texture rate of 120.0 GTexel/s. The Jetson T5000 achieves a pixel rate of 50.40 GPixel/s and a texture rate of 126.0 GTexel/s.

FP32 and FP16 throughput differ in both magnitude and ratio. The Arc B390 delivers 7.680 TFLOPS FP32 and 15.36 TFLOPS FP16 (2:1). The Jetson T5000 delivers 8.064 TFLOPS FP32 and 8.064 TFLOPS FP16 (1:1).

Power and interface specifications also differ. The Arc B390 has a TDP of 80 W, uses an IGP slot width, has no power connectors, and connects via an IGP bus interface. The Jetson T5000 has a TDP of 120 W, uses an IGP slot width, has no power connectors, lists a suggested PSU of 300 W, and connects via PCIe 5.0 x8. The Arc B390 has display outputs described as portable device dependent, while the Jetson T5000 has no display outputs.

API support is another differentiator. The Arc B390 supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The Jetson T5000 lists N/A for DirectX, OpenGL, and Vulkan. Physical dimensions differ as well: the Jetson T5000 measures 87 mm in length, 100 mm in height, and 15 mm in width, while the Arc B390 has no listed dimensions.

Release dates and successors also differ. The Arc B390 was released on 2026-01-26, with no predecessor or successor listed. The Jetson T5000 was released on 2025-08-26, with its predecessor listed as Server Hopper and its successor as Server Rubin. The Jetson T5000 has a launch MSRP of 2,999 USD.

Architecture Differences

The architectural divide is fundamental. The Arc B390 is built on Intel's Xe3-LPG architecture, part of the Arc Graphics-M generation for Panther Lake. It uses a 3 nm process at Intel's foundry. The Jetson T5000 uses NVIDIA's Blackwell architecture, part of the Server Blackwell generation, built on a 5 nm process at TSMC. These are different design philosophies: one targets integrated graphics in mobile processors, the other targets embedded and edge server compute.

The core counts reflect these priorities. The Jetson T5000 has 2560 shading units, 80 TMUs, 32 ROPs, and 20 ray tracing cores, alongside 96 tensor cores. The Arc B390 has 1536 shading units, 48 TMUs, 24 ROPs, and 12 ray tracing cores, with no tensor cores. The Jetson T5000's tensor core count indicates a hardware path for AI workloads, a feature entirely absent from the Arc B390's specifications.

Memory architecture reinforces the split. The Jetson T5000 has 128 GB of dedicated LPDDR5X memory with 273.2 GB/s bandwidth, a configuration suited to large datasets and inference models. The Arc B390 uses system shared memory with bandwidth dependent on the host system, a design typical of integrated GPUs where memory is shared with the CPU.

FP16 handling differs. The Arc B390's 2:1 FP16 ratio means it can process two FP16 operations per FP32 operation, doubling throughput to 15.36 TFLOPS. The Jetson T5000's 1:1 ratio means its FP16 throughput equals its FP32 throughput at 8.064 TFLOPS. This suggests the Arc B390 is optimized for workloads that can use packed math, while the Jetson T5000 maintains uniform precision across formats.

API support shows a similar contrast. The Arc B390 supports DirectX 12 Ultimate, OpenGL 4.6, and Vulkan 1.4, making it a viable graphics device. The Jetson T5000 lists N/A for all three APIs, indicating it is not intended for conventional graphics rendering. Its display outputs are listed as none, reinforcing its role as a compute module rather than a display adapter.

Process node and foundry also differ. The Arc B390 uses a 3 nm process at Intel, while the Jetson T5000 uses a 5 nm process at TSMC. The Jetson T5000's die size is 391 mm², a large die for a 5 nm process, reflecting the high core counts and 128 GB memory controller. The Arc B390's die size is unknown, but its lower core counts and system shared memory suggest a smaller, more integrated design.

The Verdict

The data directs each part to a distinct audience. The Intel Arc B390 is a graphics solution for portable devices. Its only recorded benchmark score of 1482 in 3DMark Steel Nomad DX12, while low in the overall percentile ranking at 9, places it ahead of four NVIDIA rivals by margins of 1.3% to 2.7%. It supports DirectX 12 Ultimate, OpenGL 4.6, and Vulkan 1.4, and its boost clock of 2500 MHz is high for an integrated part. Its FP16 throughput of 15.36 TFLOPS nearly doubles its FP32 output, and its pixel rate of 60.00 GPixel/s exceeds the Jetson T5000's 50.40 GPixel/s. For systems requiring display output and standard graphics APIs, the Arc B390 is the only viable choice between the two.

The NVIDIA Jetson T5000 is a compute module for server and edge deployments. It has no recorded benchmarks, no display outputs, and no graphics API support. Its strengths lie in raw compute: 8.064 TFLOPS FP32, 8.064 TFLOPS FP16, 96 tensor cores, and 128 GB of LPDDR5X memory with 273.2 GB/s bandwidth. Its 256-bit memory bus and PCIe 5.0 x8 interface support high-throughput data movement. The 300 W suggested PSU indicates a system designed for sustained operation. Its 120 W TDP is higher than the Arc B390's 80 W, a trade-off for the larger core count and dedicated memory.

The recorded data does not support a direct performance comparison, as the Jetson T5000 has no benchmark scores in the database. The Arc B390's 9th percentile ranking places it among low-end GPUs, and its narrow wins over the GeForce GT 520MX, GeForce 800M, GeForce GT 625 OEM, and GeForce GT 710 show it does not separate itself from those parts in any meaningful way. The Jetson T5000's 50th percentile is a placeholder, not a measured result.

The choice comes down to workload. A system that needs graphics rendering, display outputs, and standard API support should use the Arc B390. A system that needs AI inference, large memory capacity, and sustained compute should use the Jetson T5000. The specification sheets make this division explicit: the Arc B390 is an integrated GPU for portable devices, and the Jetson T5000 is a server compute module with no display functionality. Neither part serves the other's intended role.