Intel Arc Graphics 32EU vs NVIDIA Jetson Orin Nano Super Comparison

Head-to-Head Benchmarks

The recorded data presents an unusual comparison: the Intel Arc Graphics 32EU has a single benchmark score, while the NVIDIA Jetson Orin Nano Super has no benchmark entries in the database. This asymmetry shapes the entire analysis. The Intel part scores 733 points in 3DMark Steel Nomad DX12, placing it in the 3rd percentile of all GPUs. The NVIDIA part, lacking any recorded benchmark, holds a 50th percentile placement based on its specifications and category, but its average benchmark score is recorded as 0.

Looking at the nearest rivals for the Intel Arc Graphics 32EU, the data shows a tight cluster. The Intel Arc Graphics 24EU and the Intel Arc Graphics 64EU both record an average score of 733, a delta of 0 percent. This indicates that within the Arc Graphics-M family, the 32EU configuration neither gains nor loses performance relative to its lower and higher EU siblings in this specific test. The benchmark results suggest that the 3DMark Steel Nomad workload is not sensitive to the shading unit count across these three Intel variants, at least within the measurement tolerance of the database.

The AMD Radeon HD 6470M sits just below the Intel part, scoring 723, which is a 1.4 percent deficit. The NVIDIA GeForce GT 415M sits just above, scoring 751, which is a 2.4 percent advantage for the NVIDIA part. These deltas are small, placing the Intel Arc Graphics 32EU in a narrow performance band where rival GPUs from over a decade ago are within a few percentage points. The data does not support any claim of a dominant head-to-head victory for either side in this comparison, primarily because the NVIDIA Jetson Orin Nano Super has no recorded benchmark scores to compare against.

For the NVIDIA Jetson Orin Nano Super, the database lists no benchmark entries and no nearest rivals. This means the head-to-head benchmark section is defined by absence. The only quantitative performance indicator for the NVIDIA part is its derived specification-based metrics: a pixel rate of 16.32 GPixel/s, a texture rate of 32.64 GTexel/s, and a FP32 compute of 2.089 TFLOPS. The Intel part delivers 15.60 GPixel/s, 31.20 GTexel/s, and 998.4 GFLOPS. These are not benchmark scores, but they are the only comparable numbers available. The NVIDIA part leads in each of these raw throughput metrics: 4.6 percent higher pixel rate, 4.6 percent higher texture rate, and 109.2 percent higher FP32 compute.

Where Each One Wins

The Intel Arc Graphics 32EU wins in the only actual benchmark recorded. It has a measurable 3DMark Steel Nomad DX12 score of 733, while the NVIDIA Jetson Orin Nano Super has no score at all. In a strict benchmark-driven comparison, the Intel part is the only one with demonstrated 3D performance data. The Intel part also wins on process technology, using a 3 nm node from TSMC compared to the 8 nm node from Samsung for the NVIDIA part. The Intel chip integrates 17,800 million transistors on a 243 mm² die, giving a transistor density of 73.3 million per mm². The NVIDIA die is 200 mm² with an unknown transistor count, so no density comparison is possible.

The NVIDIA Jetson Orin Nano Super wins in raw compute throughput and memory bandwidth. The FP32 figure of 2.089 TFLOPS is more than double the Intel part's 998.4 GFLOPS. The FP16 figure of 4.178 TFLOPS (2:1) also far exceeds the Intel part's 1.997 TFLOPS (2:1). The NVIDIA part has 1024 shading units, 32 texture mapping units, 16 render output units, and 32 tensor cores. The Intel part has 256 shading units, 16 TMUs, and 8 ROPs, with no tensor cores listed. The NVIDIA memory subsystem is concrete: 8 GB of LPDDR5 on a 128-bit bus delivering 102.4 GB/s. The Intel part relies on system shared memory with system dependent bandwidth, which is a fundamental architectural disadvantage for memory-bound workloads.

The NVIDIA part also wins on power efficiency in terms of compute per watt. The recorded TDP for the NVIDIA part is 25 W, while the Intel part is rated at 65 W. The NVIDIA part delivers 2.089 TFLOPS at 25 W, which is 83.56 GFLOPS per watt. The Intel part delivers 998.4 GFLOPS at 65 W, which is 15.36 GFLOPS per watt. The NVIDIA part is over five times more efficient in this specific metric. The NVIDIA part also has a physical form factor advantage, with dimensions of 70 mm by 45 mm, while the Intel part is an IGP with no dimensions listed.

The Verdict

The data presents two fundamentally different products. The Intel Arc Graphics 32EU is an integrated GPU within the Arrow Lake-S chip, built for a desktop or laptop environment where it shares system memory and depends on the motherboard for display outputs. The NVIDIA Jetson Orin Nano Super is a portable device, 70 mm by 45 mm, with its own 8 GB LPDDR5 memory, a PCIe 4.0 x4 interface, and a 25 W TDP. The launch MSRP for the NVIDIA part is 249 USD.

For a user requiring a measured 3D benchmark score, the Intel part is the only choice with data. The 733 score in Steel Nomad places it at the 3rd percentile, and the closest rival data shows a 2.4 percent advantage for the GeForce GT 415M and a 1.4 percent advantage over the Radeon HD 6470M. This is not a high-performance part, but it is a functioning one in the database.

For a user requiring compute throughput, the NVIDIA part is clearly superior based on the recorded specifications. The 2.089 TFLOPS FP32 figure is more than double the Intel part, and the 102.4 GB/s memory bandwidth is a dedicated resource, not a system dependent variable. The 32 tensor cores, which the Intel part lacks entirely, indicate a purpose-built AI or inference workload capability. The 50th percentile placement, despite zero benchmark scores, suggests the database categorizes it as a mid-range part based on its specifications.

The process node difference also matters. The Intel part uses a 3 nm TSMC process, while the NVIDIA part uses an 8 nm Samsung process. The Intel part is denser, with 73.3 million transistors per mm² against an unknown figure for NVIDIA. The NVIDIA part compensates with a lower 25 W TDP, which is less than half of the Intel part's 65 W TDP. The data does not support a single winner. It supports a split verdict: Intel for benchmark-verified graphics, NVIDIA for specification-driven compute and efficiency.

FAQ

Q: What is the only recorded benchmark score for the Intel Arc Graphics 32EU?

A: The Intel Arc Graphics 32EU records a score of 733 in the 3DMark Steel Nomad DX12 test, placing it in the 3rd percentile of all GPUs.

Q: Does the NVIDIA Jetson Orin Nano Super have any benchmark scores?

A: No, the database lists no benchmark entries for the NVIDIA Jetson Orin Nano Super, and its average benchmark score is recorded as 0.

Q: How does the Intel Arc Graphics 32EU compare to its nearest rivals?

A: The Intel Arc Graphics 24EU and Intel Arc Graphics 64EU both score 733, a 0 percent delta. The AMD Radeon HD 6470M scores 723, a 1.4 percent deficit, and the NVIDIA GeForce GT 415M scores 751, a 2.4 percent advantage.

Q: What is the FP32 compute difference between the two parts?

A: The NVIDIA Jetson Orin Nano Super delivers 2.089 TFLOPS, while the Intel Arc Graphics 32EU delivers 998.4 GFLOPS. The NVIDIA part is 109.2 percent higher.

Q: What memory configuration does each part use?

A: The NVIDIA Jetson Orin Nano Super uses 8 GB of LPDDR5 on a 128-bit bus with 102.4 GB/s bandwidth. The Intel Arc Graphics 32EU uses system shared memory with system dependent bandwidth.

Q: What is the TDP for each part?

A: The NVIDIA Jetson Orin Nano Super has a 25 W TDP, while the Intel Arc Graphics 32EU has a 65 W TDP.

Architecture Differences

The Intel Arc Graphics 32EU is built on the Xe-LPG architecture, which is a graphics-focused design within the Arc Graphics-M generation for Arrow Lake. It uses a 3 nm TSMC process with 17,800 million transistors on a 243 mm² die. The shading unit count is 256, with 16 TMUs and 8 ROPs. There are no tensor cores and no RT cores listed. The bus interface is a Ring Bus, which ties it to the host CPU's internal fabric. The memory is system shared, meaning there is no dedicated VRAM, and bandwidth is system dependent. The base clock is 300 MHz with a boost clock of 1950 MHz. The pixel rate is 15.60 GPixel/s and the texture rate is 31.20 GTexel/s.

The NVIDIA Jetson Orin Nano Super is built on the Ampere architecture, which is a compute-oriented design within the Tegra generation. It uses an 8 nm Samsung process with a 200 mm² die and an unknown transistor count. The shading unit count is 1024, which is four times the Intel part, with 32 TMUs and 16 ROPs. It includes 32 tensor cores, which are absent from the Intel part. The memory is 8 GB of LPDDR5 on a 128-bit bus with 102.4 GB/s bandwidth, a fixed and dedicated resource. The bus interface is PCIe 4.0 x4, which allows external connectivity. The memory clock is 800 MHz with 6.4 Gbps effective speed. The pixel rate is 16.32 GPixel/s and the texture rate is 32.64 GTexel/s. The base and boost clocks are not listed in the database.

The architectural philosophies differ sharply. The Intel part integrates graphics into a CPU package, relying on shared resources and a ring bus for data movement. The NVIDIA part is a standalone system-on-module with its own memory and a PCIe interface, designed for embedded or edge deployments. The process node difference, 3 nm versus 8 nm, gives Intel a density advantage, but the transistor count for NVIDIA is unknown, so the comparison cannot be completed. The tensor cores on the NVIDIA part point to a machine learning orientation, while the Intel part has no such hardware listed.

Specification Differences

The two parts differ across nearly every recorded specification field. The process node is 3 nm for Intel and 8 nm for NVIDIA. The foundry is TSMC for Intel and Samsung for NVIDIA. The transistor count is 17,800 million for Intel and unknown for NVIDIA. The die size is 243 mm² for Intel and 200 mm² for NVIDIA. The transistor density is 73.3M per mm² for Intel and null for NVIDIA.

The shading units are 256 for Intel and 1024 for NVIDIA. The TMUs are 16 for Intel and 32 for NVIDIA. The ROPs are 8 for Intel and 16 for NVIDIA. The tensor cores are null for Intel and 32 for NVIDIA. The RT cores are null for both. The pixel rate is 15.60 GPixel/s for Intel and 16.32 GPixel/s for NVIDIA. The texture rate is 31.20 GTexel/s for Intel and 32.64 GTexel/s for NVIDIA. The FP32 compute is 998.4 GFLOPS for Intel and 2.089 TFLOPS for NVIDIA. The FP16 compute is 1.997 TFLOPS (2:1) for Intel and 4.178 TFLOPS (2:1) for NVIDIA.

The TDP is 65 W for Intel and 25 W for NVIDIA. The slot width is IGP for both. The bus interface is Ring Bus for Intel and PCIe 4.0 x4 for NVIDIA. The display outputs are motherboard dependent for Intel and portable device dependent for NVIDIA. The memory size is system shared for Intel and 8 GB for NVIDIA. The memory type is system shared for Intel and LPDDR5 for NVIDIA. The bus width is system shared for Intel and 128 bit for NVIDIA. The bandwidth is system dependent for Intel and 102.4 GB/s for NVIDIA. The memory clock is system shared for Intel and 800 MHz 6.4 Gbps effective for NVIDIA.

The base clock is 300 MHz for Intel and null for NVIDIA. The boost clock is 1950 MHz for Intel and null for NVIDIA. The dimensions are null for Intel and 70 mm by 45 mm for NVIDIA. The release date is 2024-10-23 for Intel and 2024-12-16 for NVIDIA. The production status is active for both. The launch MSRP is null for Intel and 249 USD for NVIDIA. The APIs are identical: DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4 for both.