Intel Arc Graphics 128EU Mobile vs NVIDIA Rubin GPU Comparison

Where Each One Wins

The Intel Arc Graphics 128EU Mobile and the NVIDIA Rubin GPU occupy entirely different segments of the hardware spectrum, and the recorded data reflects that divide clearly. The Intel part is an integrated graphics processor for mobile platforms, built around the Meteor Lake chip and using system shared memory. The NVIDIA Rubin GPU is a server-class accelerator built on the GR100 chip with dedicated HBM4 memory. There are no overlapping benchmark wins in the database, because the two products are not tested against one another in any recorded workload. Instead, the wins are structural: Intel wins on portability and integration, while NVIDIA wins on raw compute scale and memory subsystem capacity.

The Intel Arc Graphics 128EU Mobile delivers 4.608 TFLOPS of FP32 performance and 9.216 TFLOPS of FP16 performance, with a 28 W TDP. That power envelope places it in the integrated graphics class, where the entire GPU sits on the same package as the CPU. Its memory is system shared, meaning bandwidth is system dependent. The NVIDIA Rubin GPU, by contrast, delivers 130.0 TFLOPS of FP32 and 260.0 TFLOPS of FP16, with a 2300 W TDP and a suggested PSU of 2700 W. The gap in raw throughput is roughly 28 times in FP32 and 28 times in FP16, based on the recorded figures. The NVIDIA part also uses 288 GB of HBM4 memory across a 16384 bit bus, producing 22.1 TB/s of bandwidth, while the Intel part has no dedicated memory of its own.

For use-case segmentation, the Intel Arc Graphics 128EU Mobile wins in any scenario where the host system must remain lightweight and power-constrained. Its 28 W TDP and IGP slot width make it suitable for portable devices, and its display outputs are portable device dependent, which means it is designed to drive the integrated display of a laptop or similar form factor. The NVIDIA Rubin GPU wins in server deployments where compute density and memory bandwidth are the priority. Its SXM Module slot width, PCIe 6.0 x16 bus interface, and lack of display outputs confirm that it is not intended for local rendering on a screen but for accelerated compute in a data center.

The data also shows a difference in production status and release timing. Both are listed as Active, but the Intel part was released on 2023-12-13 while the NVIDIA part is dated 2025-12-31. The Intel part succeeds HD Graphics-M, while the NVIDIA part succeeds Server Blackwell. Neither has a successor listed in the database. The Intel part sits at the 50th percentile among all GPUs, and the NVIDIA part also sits at the 50th percentile, though that percentile figure does not reflect any direct comparison between the two, as neither has a benchmark score recorded.

Architecture Differences

The architectural split between the two parts is fundamental. The Intel Arc Graphics 128EU Mobile uses the Xe-LPG architecture on a 10 nm process node, fabricated by Intel. The NVIDIA Rubin GPU uses the Rubin architecture on a 3 nm process node, fabricated by TSMC. That process difference is substantial: the NVIDIA part packs 336,000 million transistors into a 1456 mm² die, yielding a transistor density of 230.8M per mm². The Intel part has no transistor count, die size, or density recorded in the database, so a direct density comparison is not possible from the available data.

The Intel GPU has 1024 shading units, 64 texture mapping units, and 32 raster output pipelines. It has no recorded ray tracing cores or tensor cores. The NVIDIA GPU has 28672 shading units, 896 texture mapping units, and 24 raster output pipelines. It has 896 tensor cores and no recorded ray tracing cores. The shading unit count difference is about 28 times in NVIDIA's favor, which aligns with the FP32 throughput gap. The texture mapping unit count is 14 times higher on the NVIDIA part, and its texture rate is 2,031.2 GTexel/s versus 144.0 GTexel/s on the Intel part.

The raster output pipeline counts are inverted: the Intel part has 32 ROPs and a pixel rate of 72.00 GPixel/s, while the NVIDIA part has 24 ROPs and a pixel rate of 54.41 GPixel/s. This is an unusual finding. Despite the NVIDIA part's vastly larger compute resources, its pixel rate is lower because it has fewer ROPs. The Intel part wins this specific rasterization metric. That makes sense for an integrated graphics part that must drive display output, whereas the NVIDIA server accelerator prioritizes compute and memory bandwidth over pixel throughput.

Clock behavior also differs. The Intel part has a 300 MHz base clock and a 2250 MHz boost clock. The NVIDIA part has a 700 MHz base clock and a 2267 MHz boost clock. The boost clocks are nearly identical, but the NVIDIA base clock is more than twice the Intel base clock. The NVIDIA memory clock is 2695 MHz with 10.8 Gbps effective data rate, while the Intel memory clock is system shared. The NVIDIA memory bus is 16384 bit wide, and the Intel bus width is system shared as well.

The API support profiles are also different. The Intel part supports DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.4. The NVIDIA part lists N/A for DirectX, OpenGL, and Vulkan. This reinforces the server positioning of the NVIDIA Rubin GPU: it is not designed for consumer graphics API workloads, while the Intel integrated part explicitly supports the standard graphics APIs. The NVIDIA part also has no display outputs, while the Intel part has portable device dependent outputs.

Head-to-Head Benchmarks

The database contains no head-to-head benchmark entries between the Intel Arc Graphics 128EU Mobile and the NVIDIA Rubin GPU. The winsA and winsB fields are both zero, and the headToHeadBenchmarks array is empty. This means no direct comparison scores exist in the recorded data. Any comparison must be drawn from the static specifications and the derived throughput figures.

The largest single-metric win for the NVIDIA Rubin GPU in the recorded data is FP32 throughput. The NVIDIA part delivers 130.0 TFLOPS, while the Intel part delivers 4.608 TFLOPS. That is a 28.2 times advantage for NVIDIA. The FP16 comparison shows the same ratio: 260.0 TFLOPS versus 9.216 TFLOPS, again roughly 28 times. The texture rate comparison shows a 14.1 times advantage for NVIDIA, at 2,031.2 GTexel/s versus 144.0 GTexel/s. The shading unit count of 28672 versus 1024 is a 28 times advantage. The tensor core count of 896 versus zero recorded tensor cores on the Intel part is not directly comparable, but the NVIDIA part clearly carries dedicated tensor hardware while the Intel part does not list any.

The memory subsystem is another decisive NVIDIA win. The NVIDIA part has 288 GB of HBM4 memory, a 16384 bit bus, and 22.1 TB/s of bandwidth. The Intel part has system shared memory with system dependent bandwidth. There is no numeric bandwidth figure for the Intel part, so the comparison is qualitative: dedicated HBM4 with a massive bus versus shared system memory with unspecified bandwidth.

The Intel Arc Graphics 128EU Mobile wins on pixel rate and ROP count. The Intel part has 32 ROPs and 72.00 GPixel/s, while the NVIDIA part has 24 ROPs and 54.41 GPixel/s. That gives Intel a 1.33 times pixel rate advantage. The Intel part also wins on power efficiency in absolute terms: 28 W TDP versus 2300 W TDP. The NVIDIA part requires a suggested PSU of 2700 W, while the Intel part has no suggested PSU listed because it is an IGP drawing power from the host platform.

The clock comparison is mixed. The NVIDIA part has a higher base clock at 700 MHz versus 300 MHz, but the boost clocks are close: 2267 MHz for NVIDIA and 2250 MHz for Intel. The NVIDIA part also has a higher transistor count, 336,000 million, and a larger die at 1456 mm², though the Intel part has no comparable figures recorded.

The Verdict

The data shows two products built for different jobs. The Intel Arc Graphics 128EU Mobile is an integrated mobile GPU with 1024 shading units, 4.608 TFLOPS of FP32, 28 W TDP, and full support for DirectX 12, OpenGL 4.6, and Vulkan 1.4. It has no dedicated memory, no tensor cores, and no server-oriented features. Its pixel rate of 72.00 GPixel/s is actually higher than the NVIDIA part's 54.41 GPixel/s, and its ROP count of 32 exceeds the NVIDIA part's 24. For a mobile integrated part, that rasterization capability is relevant, especially for driving portable device displays.

The NVIDIA Rubin GPU is a server accelerator with 28672 shading units, 896 tensor cores, 130.0 TFLOPS of FP32, 260.0 TFLOPS of FP16, 288 GB of HBM4, and 22.1 TB/s of bandwidth. It has a 2300 W TDP and uses an SXM Module slot. It has no display outputs and no consumer graphics API support. The recorded data indicates it is meant for compute workloads where memory bandwidth and FP32 or FP16 throughput dominate. Its 28 times FP32 advantage over the Intel part and its 22.1 TB/s memory bandwidth are the defining characteristics.

Who should pick which, strictly from the data: a system builder targeting a portable, integrated platform with standard graphics API support and low power draw should use the Intel part. A data center operator needing massive compute throughput, tensor core capability, and dedicated high-bandwidth memory should use the NVIDIA part. The two do not compete in the same workload space, and the database does not record any direct benchmark between them. The Intel part is the only one of the two with any graphics API support, and the NVIDIA part is the only one with tensor cores or dedicated memory. The choice is dictated by form factor and workload, not by relative performance in a shared test.

FAQ

Q: Which GPU has higher FP32 performance?

A: The NVIDIA Rubin GPU has 130.0 TFLOPS of FP32, while the Intel Arc Graphics 128EU Mobile has 4.608 TFLOPS. The NVIDIA part is roughly 28 times higher.

Q: Does the Intel Arc Graphics 128EU Mobile have dedicated memory?

A: No. Its memory size, type, and bus width are all listed as system shared, and its bandwidth is system dependent.

Q: What is the memory configuration of the NVIDIA Rubin GPU?

A: It has 288 GB of HBM4 memory, a 16384 bit bus, and 22.1 TB/s of bandwidth. The memory clock is 2695 MHz with 10.8 Gbps effective data rate.

Q: Which GPU supports DirectX?

A: The Intel Arc Graphics 128EU Mobile supports DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.4. The NVIDIA Rubin GPU lists N/A for DirectX, OpenGL, and Vulkan.

Q: What is the pixel rate of each GPU?

A: The Intel Arc Graphics 128EU Mobile has a pixel rate of 72.00 GPixel/s with 32 ROPs. The NVIDIA Rubin GPU has a pixel rate of 54.41 GPixel/s with 24 ROPs.

Q: What are the power requirements of each GPU?

A: The Intel Arc Graphics 128EU Mobile has a 28 W TDP and is an IGP. The NVIDIA Rubin GPU has a 2300 W TDP and a suggested PSU of 2700 W.