Intel Graphics 24EU Mobile vs NVIDIA N1 20SM Comparison

Where Each One Wins

The recorded data for these two mobile graphics solutions presents a clear split in design intent and resulting capability. The Intel Graphics 24EU Mobile is configured for efficiency and basic display output, while the NVIDIA N1 20SM is built as a high-throughput compute and rendering engine. Their respective benchmark profiles, though limited in direct head-to-head measurements, show distinct areas of advantage based on their architectural specifications.

The Intel part, based on the Twin Lake chip with Xe-LP architecture, operates with a base clock of 300 MHz and a boost clock of 1000 MHz. It uses system shared memory with bandwidth described as system dependent. This configuration delivers a pixel rate of 4.000 GPixel/s and a texture rate of 12.00 GTexel/s. Its FP32 compute is 384.0 GFLOPS, with FP16 at 768.0 GFLOPS using a 2:1 ratio. The power envelope is 6 W, which is exceptionally low for a graphics processor. The data indicates this is an integrated graphics processor designed for minimal power draw, likely for basic 2D acceleration, video playback, and light 3D workloads in portable devices.

The NVIDIA N1 20SM, by contrast, is a far more substantial part. It uses the GB20B chip with Blackwell 2.0 architecture, built on a 5 nm process by TSMC. Its die size is 382 mm², indicating a large, complex silicon. The base clock is 741 MHz, boosting to 2346 MHz. It has 2560 shading units, 160 TMUs, and 24 ROPs. Critically, it includes 20 ray tracing cores and 80 tensor cores, features entirely absent from the Intel part. Memory is 128 GB of LPDDR5X on a 256-bit bus, delivering 273.2 GB/s of bandwidth. The pixel rate is 56.30 GPixel/s, and the texture rate is 375.4 GTexel/s. FP32 performance reaches 12.01 TFLOPS, with FP16 at the same 12.01 TFLOPS using a 1:1 ratio. The power draw is listed as unknown, but the specifications suggest a much higher consumption than the Intel part.

In terms of where each wins, the Intel part wins on power efficiency and integration simplicity. The 6 W TDP allows for fanless or passively cooled designs in thin and light laptops. The NVIDIA part wins on raw performance metrics across every compute category. Its FP32 output is over 31 times higher than the Intel part. The texture rate is over 31 times higher. The pixel rate is over 14 times higher. The NVIDIA part also has dedicated ray tracing and tensor cores, which the Intel part lacks entirely. The NVIDIA part's memory bandwidth of 273.2 GB/s versus the Intel part's system dependent shared memory represents a massive advantage for memory-intensive workloads.

For use-case segmentation, the Intel Graphics 24EU Mobile is suited for office productivity, web browsing, and media consumption where power draw is the primary constraint. The NVIDIA N1 20SM is suited for gaming, 3D rendering, machine learning inference, and any workload that demands high parallel compute throughput. The presence of 20 ray tracing cores and 80 tensor cores in the NVIDIA part specifically targets applications that use hardware-accelerated ray tracing and AI-based features like DLSS or similar technologies. The Intel part has no such dedicated hardware.

The interface differences also indicate intended usage. The Intel part uses a Ring Bus interface, typical for tightly integrated system-on-chip designs. The NVIDIA part uses PCIe 5.0 x16, which allows for high-bandwidth communication with the host processor and potentially external devices. The NVIDIA part also lists a display output of 1x HDMI, while the Intel part's display outputs are described as portable device dependent, meaning they vary by the specific laptop implementation.

The Verdict

The benchmark data indicates these are not direct competitors but rather two points on a wide spectrum of mobile graphics capability. The Intel Graphics 24EU Mobile is a low-power integrated solution for basic tasks. The NVIDIA N1 20SM is a high-performance integrated processor for demanding applications. Selection between them depends entirely on the user's workload and the device's power budget.

For users whose primary activities are document editing, web browsing, video streaming, and lightweight productivity applications, the Intel Graphics 24EU Mobile with its 6 W power draw is sufficient. Its system shared memory architecture means it does not require dedicated VRAM, reducing cost and complexity. The 192 shading units and 12 TMUs are adequate for 2D acceleration and basic video decode. Its FP32 output of 384.0 GFLOPS is enough for casual 3D graphics at low resolutions and detail settings.

For users who run games, 3D modeling software, video editing with effects, or machine learning workloads, the NVIDIA N1 20SM is the clear choice. Its 2560 shading units, 12.01 TFLOPS of FP32 compute, and 273.2 GB/s memory bandwidth provide the throughput needed for these applications. The 20 ray tracing cores enable hardware-accelerated ray tracing, and the 80 tensor cores accelerate AI-based features. The 128 GB of LPDDR5X memory is substantial, though this is shared with the system, and the 256-bit bus width ensures high data transfer rates.

The percentile data places both parts at the 50th percentile against all GPUs, but this is likely misleading given the vast difference in their specifications. The Intel part's low power draw is its primary virtue. The NVIDIA part's performance is its primary virtue. There is no scenario where the Intel part outperforms the NVIDIA part in any compute benchmark, as the data shows NVIDIA leading in every measured rate and throughput metric.

The release dates differ, with the Intel part appearing in the database as released in 2024 and the NVIDIA part in 2026. This suggests a generational gap, with the NVIDIA part being a more recent design benefiting from newer architecture and process technology. The 5 nm TSMC process versus the 10 nm Intel process also contributes to the NVIDIA part's higher density and performance.

Users with strict power constraints, such as those requiring maximum battery life in a thin laptop, should select the Intel part. Users with performance requirements should select the NVIDIA part, accepting the likely higher power draw and larger die size. The data does not support any other conclusion.

Head-to-Head Benchmarks

The direct head-to-head benchmark array is empty in the database, so the comparison relies on the recorded specification-based rates and throughput values. The largest wins for the NVIDIA N1 20SM appear across all compute metrics.

In FP32 compute, the NVIDIA part delivers 12.01 TFLOPS versus the Intel part's 384.0 GFLOPS. This is a factor of approximately 31.3 times higher performance. For FP16, the NVIDIA part delivers 12.01 TFLOPS with a 1:1 ratio, while the Intel part delivers 768.0 GFLOPS with a 2:1 ratio. The NVIDIA part's FP16 output is over 15 times higher. The NVIDIA part's 1:1 FP16 to FP32 ratio means it does not sacrifice precision for throughput, while the Intel part's 2:1 ratio indicates a reduced precision mode.

The texture rate shows the NVIDIA part at 375.4 GTexel/s versus the Intel part's 12.00 GTexel/s. This is a 31.3 times advantage. The pixel rate shows the NVIDIA part at 56.30 GPixel/s versus the Intel part's 4.000 GPixel/s, a 14.1 times advantage. These differences reflect the NVIDIA part's 160 TMUs versus the Intel part's 12 TMUs, and the NVIDIA part's 24 ROPs versus the Intel part's 4 ROPs.

Memory bandwidth is a major differentiator. The NVIDIA part has 273.2 GB/s from its 256-bit bus and LPDDR5X memory. The Intel part's bandwidth is listed as system dependent, meaning it shares the system memory bus and has no dedicated high-speed memory interface. In practice, the NVIDIA part's dedicated memory path will provide significantly higher sustained bandwidth for texture-heavy and compute-heavy workloads.

Clock speeds also differ substantially. The NVIDIA part boosts to 2346 MHz, while the Intel part boosts to 1000 MHz. The NVIDIA part's base clock of 741 MHz is lower than the Intel part's boost, but the NVIDIA part's much higher shader count and memory bandwidth compensate. The NVIDIA part's higher boost clock, combined with 2560 shading units, yields the massive FP32 advantage.

The NVIDIA part includes 20 ray tracing cores and 80 tensor cores, which the Intel part does not have. These are not directly comparable to any Intel part specification, but their presence enables features that the Intel part cannot perform in hardware. Ray tracing workloads would either need to be computed on the shading units, which would be far slower, or omitted entirely. Tensor core workloads like AI upscaling or inference would similarly fall back to shader-based computation.

The NVIDIA part's die size of 382 mm² versus the Intel part's unknown die size suggests a much larger silicon area, allowing for the higher component counts. The NVIDIA part's process node of 5 nm from TSMC versus the Intel part's 10 nm from Intel also indicates a manufacturing advantage in density and power efficiency per transistor.

The Intel part's only wins are in power consumption and integration simplicity. At 6 W, it draws a fraction of what the NVIDIA part likely requires, though the NVIDIA part's TDP is listed as unknown. The Intel part's Ring Bus interface is simpler than the NVIDIA part's PCIe 5.0 x16, which may be easier to implement in very small form factors. The Intel part also has no power connectors listed, while the NVIDIA part lists none as well, but the NVIDIA part's higher performance suggests it will need more power delivery.

FAQ

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

A: The NVIDIA N1 20SM delivers 12.01 TFLOPS of FP32 compute, while the Intel Graphics 24EU Mobile delivers 384.0 GFLOPS. The NVIDIA part is approximately 31.3 times faster in this metric.

Q: Does the Intel Graphics 24EU Mobile support ray tracing?

A: No. The Intel part has no ray tracing cores listed in the database. The NVIDIA N1 20SM includes 20 ray tracing cores.

Q: What memory configurations do these parts use?

A: The Intel part uses system shared memory with system dependent bandwidth. The NVIDIA part uses 128 GB of LPDDR5X on a 256-bit bus with 273.2 GB/s bandwidth.

Q: Which part has a higher boost clock?

A: The NVIDIA N1 20SM has a boost clock of 2346 MHz. The Intel Graphics 24EU Mobile has a boost clock of 1000 MHz.

Q: What is the power draw of the Intel Graphics 24EU Mobile?

A: The Intel part has a TDP of 6 W. The NVIDIA part's TDP is listed as unknown in the database.

Q: What process nodes are these parts built on?

A: The Intel part is built on a 10 nm process by Intel. The NVIDIA part is built on a 5 nm process by TSMC.

Architecture Differences

The architectural divergence between these two integrated graphics processors is fundamental. The Intel Graphics 24EU Mobile uses the Xe-LP architecture, which is Intel's low-power graphics design. It is part of the HD Graphics-T generation for Twin Lake chips. The NVIDIA N1 20SM uses the Blackwell 2.0 architecture, which is NVIDIA's latest IGP design for the N1x generation.

The compute resources differ dramatically. The Intel part has 192 shading units, 12 TMUs, and 4 ROPs. The NVIDIA part has 2560 shading units, 160 TMUs, and 24 ROPs. These counts directly determine the pixel and texture rates. The NVIDIA part's 160 TMUs allow for 375.4 GTexel/s texture filtering, while the Intel part's 12 TMUs manage 12.00 GTexel/s. The NVIDIA part's 24 ROPs output 56.30 GPixel/s, while the Intel part's 4 ROPs output 4.000 GPixel/s.

The NVIDIA part includes dedicated hardware not present in the Intel part. It has 20 ray tracing cores and 80 tensor cores. The ray tracing cores accelerate bounding volume hierarchy traversal and ray-triangle intersection tests for real-time ray tracing. The tensor cores accelerate matrix multiplication for deep learning inference and training. The Intel part has no equivalent hardware, meaning any such workloads must be handled by the general-purpose shading units.

Memory architecture is another key difference. The Intel part uses system shared memory with no dedicated VRAM. Its memory type, bus width, and bandwidth are all listed as system dependent. This means performance scales with the host system's memory configuration. The NVIDIA part has 128 GB of LPDDR5X on a 256-bit bus, providing 273.2 GB/s of bandwidth. This dedicated memory path reduces latency and increases sustained throughput for large datasets.

Clock specifications also differ. The Intel part has a base clock of 300 MHz and a boost of 1000 MHz. The NVIDIA part has a base of 741 MHz and a boost of 2346 MHz. The NVIDIA part's higher clocks, combined with its larger shader array, produce its substantial compute advantage. The Intel part's low clocks reflect its 6 W power budget.

The process technology differs. The Intel part is on a 10 nm process from Intel's foundry. The NVIDIA part is on a 5 nm process from TSMC. The smaller process node allows for higher transistor density and improved power efficiency. The NVIDIA part's die size is 382 mm², which is large for an IGP but accommodates the high component counts. The Intel part's die size is unknown.

The bus interface differs as well. The Intel part uses a Ring Bus, which is a common on-die interconnect for integrated processors. The NVIDIA part uses PCIe 5.0 x16, which provides high-bandwidth communication with the host CPU and potentially allows for external expansion. This suggests the NVIDIA part may be implemented as a discrete-style package or a multi-chip module with a PCIe link.

Feature support also varies. The Intel part supports DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.4. The NVIDIA part lists its API support as N/A for DirectX, OpenGL, and Vulkan. This is unusual and may indicate the database has not yet recorded the NVIDIA part's API support, or that the part uses a proprietary or different API stack. The display outputs also differ: the Intel part is portable device dependent, while the NVIDIA part lists 1x HDMI.

The release dates place the Intel part in 2024 and the NVIDIA part in 2026. The NVIDIA part is a more recent design, likely benefiting from architectural improvements and manufacturing advances. The power connectors are listed as none for the NVIDIA part and null for the Intel part, though the NVIDIA part's higher performance suggests it will require more power delivery from the system.