Intel Arc Graphics 24EU vs NVIDIA Rubin GPU Comparison

Intel Arc Graphics 24EU and NVIDIA Rubin GPU occupy opposite ends of the hardware spectrum, and the recorded data confirms that their roles in a system could not be more distinct. The database contains only a single benchmark result for the Intel part, the 3DMark Steel Nomad DX12 test, where it scores 733. The NVIDIA Rubin GPU has no benchmark entries in the database, which limits direct numerical comparison. However, the specifications and the few available metrics for the Intel part provide a clear picture of what each product is designed to do.

Head-to-Head Benchmarks

The only recorded benchmark score in this comparison belongs to Intel Arc Graphics 24EU. In 3DMark Steel Nomad DX12, it delivers a score of 733 points. This places the Intel integrated graphics solution at the 3rd percentile among all GPUs in the database. That percentile ranking means 97 percent of recorded graphics processors outperform it in this specific workload. The score itself is modest, and the data indicates that this is an entry-level graphics solution intended for basic display output and light graphical tasks.

The nearest rivals for Intel Arc Graphics 24EU provide context for its performance level. Intel Arc Graphics 32EU and Intel Arc Graphics 64EU both achieve an average score of 733, matching the 24EU model exactly with a delta of 0 percent. This suggests that within the same architecture family, the number of execution units does not meaningfully change the result in this particular benchmark, at least according to the recorded data. AMD Radeon HD 6470M scores 723, which is 1.4 percent lower than the Intel part. NVIDIA GeForce GT 415M scores 751, which is 2.4 percent higher. These are very small differences, all within a narrow band around 733 points. The data shows that Intel Arc Graphics 24EU performs nearly identically to its immediate siblings and sits within a few percentage points of decade-old discrete mobile GPUs.

NVIDIA Rubin GPU has no benchmark scores in the database. Its average benchmark score is listed as 0, and its nearest rivals list is empty. This absence of data means no direct head-to-head benchmark comparison is possible. The Intel part has a concrete score; the NVIDIA part does not. What the database does provide for the Rubin GPU is a percentile of 50, which suggests a mid-pack ranking in the overall distribution, but without a score, this cannot be verified or compared numerically. The wins total in the head-to-head section is 0 for both products, reflecting the lack of paired benchmark results.

The architectural disparity between the two parts is immense. Intel Arc Graphics 24EU operates with a boost clock of 2000 MHz and a base clock of 300 MHz. NVIDIA Rubin GPU runs at a base clock of 700 MHz and boosts to 2267 MHz. The NVIDIA part has a higher boost clock by 267 MHz, but clock speed alone does not explain the performance gap. The raw compute figures tell a more complete story. Intel Arc Graphics 24EU delivers 768.0 GFLOPS of FP32 performance and 1.536 TFLOPS of FP16 performance. NVIDIA Rubin GPU delivers 130.0 TFLOPS of FP32 and 260.0 TFLOPS of FP16. The NVIDIA part is roughly 169 times faster in FP32 and 169 times faster in FP16, based strictly on the recorded specifications. These numbers represent different classes of hardware entirely.

The Verdict

The data indicates that these two products serve completely different purposes. Intel Arc Graphics 24EU is an integrated graphics processor, marked as IGP in slot width, with a TDP of 65 W. It is part of the Arrow Lake-S chip and uses the Xe-LPG architecture. Its 3rd percentile ranking and modest 733 benchmark score show that it is not designed for demanding workloads. It exists to provide display output and basic acceleration for a general-purpose desktop processor. The recorded data shows it supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4, which means it can handle modern APIs, but its 192 shading units, 12 texture mapping units, and 6 raster output pipelines limit its actual throughput.

NVIDIA Rubin GPU is a server-class accelerator. Its slot width is SXM Module, its TDP is 2300 W, and the database suggests a 2700 W power supply. This is not a product for a desktop computer. It has no display outputs, meaning it is not intended to drive monitors. It supports no graphics APIs in the database, with DirectX, OpenGL, and Vulkan all listed as N/A. This is a compute device, designed for data center workloads such as AI training and inference. Its 28672 shading units, 896 texture mapping units, 896 tensor cores, and 288 GB of HBM4 memory with 22.1 TB/s of bandwidth place it in an entirely different performance class.

The choice between these two is not a choice at all for a single user. A system with Intel Arc Graphics 24EU is a mainstream desktop or laptop that needs integrated graphics for everyday use. A system with NVIDIA Rubin GPU is a server node that requires massive parallel compute power. The data does not support any scenario where these two products compete for the same purchase decision. The Intel part is for general computing with light graphics; the NVIDIA part is for high-performance computing with no graphics output at all.

Architecture Differences

The two products use the same 3 nm process node and the same foundry, TSMC, but that is where the similarities end. Intel Arc Graphics 24EU is built on the Xe-LPG architecture and belongs to the Arc Graphics (Arrow Lake) generation. It is part of the Arrow Lake-S chip, which contains 17,800 million transistors on a 243 mm² die, producing a transistor density of 73.3 million transistors per square millimeter. The GPU itself has 192 shading units, 12 TMUs, and 6 ROPs. It has no dedicated ray tracing cores and no tensor cores. Its memory is system shared, meaning it uses the host system's RAM rather than dedicated VRAM. The memory bus width is also system shared, and bandwidth is system dependent. This integrated design keeps power consumption at 65 W and uses a ring bus interface.

NVIDIA Rubin GPU is built on the Rubin architecture and belongs to the Server Rubin (Rxx) generation. Its chip is designated GR100, and it contains 336,000 million transistors on a 1456 mm² die, yielding a transistor density of 230.8 million transistors per square millimeter. The GPU has 28672 shading units, 896 TMUs, and 24 ROPs. It also has 896 tensor cores, which are essential for AI and machine learning workloads. Memory is far more substantial: 288 GB of HBM4 on a 16384-bit bus, providing 22.1 TB/s of bandwidth. Memory clock is listed as 2695 MHz, with 10.8 Gbps effective. The bus interface is PCIe 6.0 x16. The TDP is 2300 W, and the slot width is SXM Module. This is a massive, power-hungry accelerator designed for installation in server racks.

The transistor counts illustrate the scale difference. Intel's integrated GPU is part of a processor with 17,800 million transistors, while NVIDIA's Rubin GPU chip contains 336,000 million transistors. That is approximately 19 times more transistors in the NVIDIA chip. The die size difference is similarly stark: 243 mm² for the Intel chip versus 1456 mm² for the NVIDIA chip. The transistor density also differs, with NVIDIA packing over three times more transistors per square millimeter. These are fundamentally different engineering approaches. The Intel part is optimized for low power and integration into a general-purpose CPU. The NVIDIA part is optimized for maximum compute throughput in a server environment.

Production status for both is listed as Active. The Intel part has a release date of October 23, 2024, while the NVIDIA Rubin GPU has a release date of December 31, 2025. The Intel part lists HD Graphics as its predecessor, while the NVIDIA part lists Server Blackwell as its predecessor. Neither has a successor listed. The Intel part has no launch MSRP in the database, and neither does the NVIDIA part.

FAQ

Q: What is the performance difference between Intel Arc Graphics 24EU and NVIDIA Rubin GPU?

A: The database contains no direct benchmark comparison. Intel Arc Graphics 24EU scores 733 in 3DMark Steel Nomad DX12, while NVIDIA Rubin GPU has no recorded benchmark scores. However, FP32 compute figures show NVIDIA at 130.0 TFLOPS versus Intel's 768.0 GFLOPS, indicating a massive performance gap in raw compute capability.

Q: Which product has more shading units?

A: NVIDIA Rubin GPU has 28672 shading units, while Intel Arc Graphics 24EU has 192 shading units. The NVIDIA part also has 896 tensor cores, which the Intel part lacks entirely.

Q: What memory configurations do these products use?

A: Intel Arc Graphics 24EU uses system shared memory, with size, type, and bus width all marked as system shared. NVIDIA Rubin GPU has 288 GB of HBM4 memory on a 16384-bit bus with 22.1 TB/s of bandwidth.

Q: Can NVIDIA Rubin GPU output video to a display?

A: No. The database lists its display outputs as "No outputs." Intel Arc Graphics 24EU has display outputs listed as "Motherboard Dependent," meaning the motherboard determines what video connections are available.

Q: What is the power consumption difference?

A: Intel Arc Graphics 24EU has a TDP of 65 W. NVIDIA Rubin GPU has a TDP of 2300 W, and the database suggests a 2700 W power supply for systems using it.

Q: Are these products from the same generation or architecture?

A: No. Intel uses the Xe-LPG architecture in the Arc Graphics (Arrow Lake) generation. NVIDIA uses the Rubin architecture in the Server Rubin (Rxx) generation. Both use a 3 nm process from TSMC, but the designs are unrelated.

Where Each One Wins

Intel Arc Graphics 24EU wins in scenarios requiring low power consumption. Its 65 W TDP is a fraction of the NVIDIA part's 2300 W. It also wins on integration, as it is an IGP that shares system memory and requires no separate graphics card. For a desktop or laptop processor, this integrated solution provides display output and basic 3D acceleration without additional hardware. Its support for DirectX 12 Ultimate, OpenGL 4.6, and Vulkan 1.4 means it can run modern operating systems and applications with graphical interfaces. The 733 benchmark score, while low in absolute terms, is only 2.4 percent below the NVIDIA GeForce GT 415M and 1.4 percent above the AMD Radeon HD 6470M, showing it is competitive with older discrete mobile GPUs.

NVIDIA Rubin GPU wins in every compute-heavy scenario. Its 130.0 TFLOPS of FP32 performance and 260.0 TFLOPS of FP16 performance place it in a class that the Intel part cannot approach. The 896 tensor cores are designed for AI acceleration, and the 288 GB of HBM4 memory with 22.1 TB/s bandwidth allows it to handle massive datasets. The 16384-bit memory bus is orders of magnitude wider than the system shared memory of the Intel part. The PCIe 6.0 x16 interface provides high-bandwidth connectivity to the host system. The lack of display outputs confirms its role as a compute accelerator, not a graphics card. The 50th percentile ranking in the database, while not backed by a benchmark score, suggests it sits in the middle of the distribution for all recorded GPUs, which is notable given the lack of any benchmark entry.

The data shows that Intel Arc Graphics 24EU is for systems where graphics are a secondary concern. It is part of a processor, not a standalone component. Its 192 shading units and 12 TMUs handle basic rendering, and its 6 ROPs limit pixel output to 12.00 GPixel/s. Texture rate is 24.00 GTexel/s, and FP32 output is 768.0 GFLOPS. These are small numbers, but they are sufficient for everyday computing, video playback, and light productivity. The 3rd percentile ranking reflects its position as an entry-level solution.

NVIDIA Rubin GPU is for systems where graphics rendering is irrelevant and compute throughput is everything. Its 28672 shading units, 896 TMUs, and 24 ROPs are configured for parallel processing, not display output. The pixel rate of 54.41 GPixel/s and texture rate of 2,031.2 GTexel/s are over four times and over eighty times higher than the Intel part, respectively. These figures indicate a device built for massive parallel workloads. The FP16 performance of 260.0 TFLOPS, double the FP32 figure, shows a design optimized for mixed-precision compute, common in AI training.

The release dates differ by over a year, with Intel launching on October 23, 2024, and NVIDIA scheduled for December 31, 2025. Both are marked as Active in production status. The Intel part has no launch MSRP in the database, and neither does the NVIDIA part. The predecessor names differ as well: HD Graphics for Intel, Server Blackwell for NVIDIA. These are separate product lines with separate goals. The benchmark data, while incomplete for the NVIDIA part, is consistent with this division. Intel Arc Graphics 24EU has a recorded score; NVIDIA Rubin GPU does not. The specifications alone make the intended use cases clear.