Intel Graphics 24EU Mobile vs NVIDIA Switch 2 GPU Comparison

Head-to-Head Benchmarks

The recorded data for this comparison is unusual: neither GPU has any benchmark entries in the database, and the head-to-head benchmark list is empty. Both parts show an average benchmark score of 0 and a percentile rank of 50 against all GPUs. This means any direct numerical performance comparison must be derived from their hardware specifications rather than measured workloads.

The FP32 compute figures reveal the most dramatic gap. The Intel Graphics 24EU Mobile delivers 384.0 GFLOPS, while the NVIDIA Switch 2 GPU reaches 4.301 TFLOPS. Converting that to the same unit, the Switch 2 GPU provides 4,301 GFLOPS, which is approximately 11.2 times higher. This is the single largest mathematical advantage in the comparison. The FP16 figures follow the same pattern: Intel lists 768.0 GFLOPS (2:1), while NVIDIA lists 8.602 TFLOPS (2:1), again an 11.2-fold difference.

Pixel throughput shows a similar hierarchy. The Intel part sustains 4.000 GPixel/s, while the NVIDIA part reaches 22.40 GPixel/s, which is 5.6 times faster. Texture rate tells a comparable story: Intel delivers 12.00 GTexel/s, NVIDIA delivers 67.20 GTexel/s, a 5.6 times advantage. These ratios are not arbitrary; they track the respective ROP and TMU counts. Intel has 4 ROPs and 12 TMUs, while NVIDIA has 16 ROPs and 48 TMUs, exactly four times more of each. Yet the pixel and texture rates are only 5.6 times higher, not 4 times, which reflects the clock speed differences between the two parts.

The clock speeds themselves are worth examining. Intel runs at a 300 MHz base and 1000 MHz boost. NVIDIA runs at a 561 MHz base and 1400 MHz boost. The NVIDIA boost clock is 1.4 times higher than Intel's boost, and its base clock is 1.87 times higher. These clock advantages compound with the shading unit count. Intel has 192 shading units, NVIDIA has 1536, which is exactly 8 times more. The combination of 8 times the shading units and 1.4 times the boost clock yields roughly the 11.2 times FP32 advantage observed.

Memory bandwidth presents a stark contrast. Intel uses system shared memory with bandwidth listed as "System Dependent," meaning no fixed figure exists. NVIDIA uses 12 GB of LPDDR5X on a 128-bit bus, providing 102.4 GB/s. The absence of a fixed bandwidth number for Intel means no direct ratio can be computed, but the architectural difference is clear: one depends entirely on the host platform's memory configuration, while the other has a dedicated, fixed allocation.

Architecture Differences

The two GPUs come from different manufacturers, different foundries, and different architectural generations. Intel's chip is called Twin Lake, built on the Xe-LP architecture, and belongs to the HD Graphics-T (Twin Lake) generation. NVIDIA's chip is called GA10B, built on the Ampere architecture, and belongs to the Console GPU (Nintendo) generation. The process nodes differ: Intel uses 10 nm silicon from Intel's own foundry, while NVIDIA uses 8 nm silicon from Samsung. The die size for Intel is unknown, but NVIDIA's die measures 200 mm².

The transistor counts for both parts are listed as unknown, so no transistor density comparison is possible. However, the die size difference alone suggests a substantial resource gap. A 200 mm² die allows for significantly more hardware blocks than the integrated Intel solution, which is reflected in the raw unit counts.

The shading unit count is the most telling architectural difference. Intel packs 192 shading units, 12 TMUs, and 4 ROPs. NVIDIA packs 1536 shading units, 48 TMUs, and 16 ROPs. NVIDIA has exactly 8 times the shading units, 4 times the TMUs, and 4 times the ROPs. Beyond these base units, NVIDIA adds dedicated hardware that Intel lacks entirely: 12 RT cores and 48 tensor cores. Intel lists no RT cores and no tensor cores. This means the NVIDIA part has hardware acceleration for ray tracing and tensor operations, while the Intel part must rely on compute shaders or other software approaches for those workloads.

Memory architecture differs fundamentally. Intel uses system shared memory for both size and type, with a system shared bus width and system dependent bandwidth. NVIDIA has 12 GB of dedicated LPDDR5X memory on a 128-bit bus with a fixed 102.4 GB/s bandwidth. The NVIDIA memory clock is listed as 800 MHz, which is specified as 6.4 Gbps effective. Intel's memory clock is also listed as system shared, meaning it adapts to whatever the host platform provides.

The power envelopes differ by a factor of roughly 6.7. Intel is rated at 6 W TDP, while NVIDIA is rated at 40 W TDP. This power difference aligns with the performance difference, though the ratio is not identical. The API support shows some overlap: both support DirectX 12, OpenGL 4.6, and Vulkan 1.4. However, NVIDIA supports DirectX 12 Ultimate (12_2), while Intel supports DirectX 12 (12_1). This means NVIDIA meets the full DirectX 12 Ultimate feature set, including ray tracing and mesh shaders, while Intel only meets the 12_1 feature level.

The bus interface also differs. Intel uses a Ring Bus, which is typical for integrated graphics sharing the CPU's memory subsystem. NVIDIA lists no bus interface, which is consistent with a dedicated console GPU that connects directly to the console's memory controller. The display outputs differ as well: Intel lists "Portable Device Dependent," meaning the outputs depend on the host device, while NVIDIA lists "No outputs," indicating this GPU does not drive displays directly.

FAQ

Q: Which GPU has more shading units?

A: The NVIDIA Switch 2 GPU has 1536 shading units, which is exactly 8 times the 192 shading units found in the Intel Graphics 24EU Mobile.

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

A: No. The Intel part lists no RT cores, while the NVIDIA Switch 2 GPU includes 12 RT cores. Additionally, NVIDIA supports DirectX 12 Ultimate (12_2), which includes ray tracing features, whereas Intel only supports DirectX 12 (12_1).

Q: What is the memory bandwidth difference?

A: The NVIDIA Switch 2 GPU has a fixed 102.4 GB/s bandwidth from 12 GB of LPDDR5X on a 128-bit bus. The Intel Graphics 24EU Mobile uses system shared memory with bandwidth listed as "System Dependent," meaning no fixed number exists.

Q: How do the power requirements compare?

A: The Intel Graphics 24EU Mobile has a 6 W TDP, while the NVIDIA Switch 2 GPU has a 40 W TDP. This makes the NVIDIA part approximately 6.7 times more power-hungry.

Q: What is the FP32 compute difference?

A: The Intel part delivers 384.0 GFLOPS, while the NVIDIA part delivers 4.301 TFLOPS. In the same units, NVIDIA provides 4,301 GFLOPS, which is about 11.2 times higher.

Q: Which GPU has dedicated tensor cores?

A: Only the NVIDIA Switch 2 GPU includes tensor cores, with 48 of them. The Intel Graphics 24EU Mobile lists no tensor cores.

Specification Differences

The following fields differ between the two parts:

• Manufacturer: Intel versus NVIDIA
• Chip: Twin Lake versus GA10B
• Architecture: Xe-LP versus Ampere
• Generation: HD Graphics-T (Twin Lake) versus Console GPU (Nintendo)
• Process Node: 10 nm versus 8 nm
• Foundry: Intel versus Samsung
• Die Size: unknown versus 200 mm²
• Base Clock: 300 MHz versus 561 MHz
• Boost Clock: 1000 MHz versus 1400 MHz
• Memory Clock: System Shared versus 800 MHz 6.4 Gbps effective
• Memory Size: System Shared versus 12 GB
• Memory Type: System Shared versus LPDDR5X
• Memory Bus Width: System Shared versus 128 bit
• Memory Bandwidth: System Dependent versus 102.4 GB/s
• Shading Units: 192 versus 1536
• TMUs: 12 versus 48
• ROPs: 4 versus 16
• RT Cores: null versus 12
• Tensor Cores: null versus 48
• Pixel Rate: 4.000 GPixel/s versus 22.40 GPixel/s
• Texture Rate: 12.00 GTexel/s versus 67.20 GTexel/s
• FP32: 384.0 GFLOPS versus 4.301 TFLOPS
• FP16: 768.0 GFLOPS (2:1) versus 8.602 TFLOPS (2:1)
• TDP: 6 W versus 40 W
• Slot Width: IGP versus null
• Bus Interface: Ring Bus versus null
• Display Outputs: Portable Device Dependent versus No outputs
• DirectX Support: 12 (12_1) versus 12 Ultimate (12_2)
• Dimensions: null versus 272 mm 10.7 inches length, 116 mm 4.6 inches height, 14 mm 0.6 inches width
• Release Date: 2024-12-31 versus 2025-06-04
• Launch MSRP: null versus 449 USD

Fields that are identical include OpenGL 4.6, Vulkan 1.4, production status of Active, and unknown transistor counts.

Where Each One Wins

The Intel Graphics 24EU Mobile wins in efficiency and integration. Its 6 W TDP is dramatically lower than NVIDIA's 40 W, and it uses system shared memory, meaning it does not require dedicated VRAM allocation. Its Ring Bus interface and IGP slot width confirm it is designed to be embedded into a portable device, with display outputs that depend on the host. This makes it suitable for light computing tasks where power draw is the primary constraint and where the host platform already provides memory.

The NVIDIA Switch 2 GPU wins in every raw performance metric recorded. It has higher clocks (base and boost), more shading units, TMUs, ROPs, RT cores, and tensor cores. Its FP32 compute is 11.2 times higher, its pixel rate is 5.6 times higher, and its texture rate is 5.6 times higher. It has dedicated memory with a fixed bandwidth of 102.4 GB/s, which removes dependence on host memory configuration. The DirectX 12 Ultimate support adds ray tracing and mesh shader capabilities that Intel cannot match with its 12_1 feature level.

For workloads that involve heavy parallel compute, such as 3D rendering or physics simulation, the NVIDIA part's 1536 shading units and 4.301 TFLOPS FP32 throughput provide a massive advantage. For ray-traced scenes, the 12 RT cores are essential, as Intel has no equivalent hardware. For machine learning or tensor operations, the 48 tensor cores are the only option between the two, since Intel lists none.

The Intel part retains a niche for ultra-low-power portable contexts. Its 6 W TDP allows deployment in devices where a 40 W GPU would be thermally or electrically infeasible. The system shared memory also means no separate memory chips are needed, simplifying the host design. However, the performance ceiling is low, and the absence of fixed memory bandwidth means performance varies with the host's memory subsystem.

The Verdict

The data indicates two completely different product categories. The Intel Graphics 24EU Mobile is an integrated GPU with 192 shading units, a 1000 MHz boost clock, 384.0 GFLOPS FP32, and a 6 W TDP. The NVIDIA Switch 2 GPU is a dedicated console GPU with 1536 shading units, a 1400 MHz boost clock, 4.301 TFLOPS FP32, a 40 W TDP, and a 200 mm² die from Samsung's 8 nm process.

The benchmark database shows no measured scores for either part, so the verdict rests on specification analysis. The NVIDIA part is architecturally superior in every compute dimension: 8 times the shading units, 4 times the TMUs, 4 times the ROPs, 11.2 times the FP32 throughput, 5.6 times the pixel rate, and 5.6 times the texture rate. It also adds RT cores and tensor cores that Intel lacks entirely. The fixed 102.4 GB/s memory bandwidth is a clear advantage over Intel's system dependent bandwidth.

The Intel part wins only in power consumption and integration simplicity. Its 6 W TDP is suitable for battery-constrained portable devices, and its system shared memory eliminates dedicated VRAM requirements. The release date of 2024-12-31 for Intel versus 2025-06-04 for NVIDIA shows the Intel part came first, but the NVIDIA part launched at a 449 USD MSRP, which the database records as its launch price.

A user requiring ray tracing, tensor acceleration, high FP32 throughput, or fixed memory bandwidth should select the NVIDIA Switch 2 GPU. A user constrained to a 6 W power envelope with no need for discrete memory or advanced acceleration blocks would be limited to the Intel Graphics 24EU Mobile. The data does not support any scenario where the Intel part outperforms the NVIDIA part in raw throughput. The NVIDIA part is the clear performance leader, while the Intel part serves a narrow low-power niche.