Intel Arc Graphics 112EU Mobile vs NVIDIA N1 20SM Comparison

Intel Arc Graphics 112EU Mobile and NVIDIA N1 20SM are both integrated graphics solutions, but they target fundamentally different segments of the mobile market. The Arc part comes from Intel’s Meteor Lake platform, built on the Xe-LPG architecture, while the N1 20SM is an NVIDIA Blackwell 2.0 IGP using the GB20B chip. Neither part has recorded benchmark scores in the database, and both sit at the 50th percentile among all GPUs, so the analysis below is driven entirely by their architectural and specification data.

Where Each One Wins

The Intel Arc Graphics 112EU Mobile is positioned for graphics-oriented workloads within a Meteor Lake mobile system. It delivers 896 shading units, 56 texture mapping units, and 24 raster operation units. Its floating-point throughput reaches 3.942 TFLOPS for FP32 and 7.885 TFLOPS for FP16 with a 2:1 ratio. That FP16 capability is a notable asset for applications that can exploit half-precision math, such as certain compute and media workloads. The part also exposes a full DirectX 12 (12_1) feature set, OpenGL 4.6, and Vulkan 1.4, making it broadly compatible with Windows gaming and graphics applications. The 2200 MHz boost clock is high for an integrated part, and the 123.2 GTexel/s texture rate gives it substantial fill-rate headroom for its class.

The NVIDIA N1 20SM wins decisively in raw compute throughput. With 2560 shading units and a 2346 MHz boost clock, it delivers 12.01 TFLOPS of FP32 performance, which is roughly three times the Intel part’s FP32 output. It also matches that figure for FP16 at a 1:1 ratio, meaning it does not rely on rate-limited half-precision execution. The N1 20SM includes 20 ray tracing cores and 80 tensor cores, features entirely absent from the Intel Arc 112EU. Any workload involving ray-traced effects or tensor-accelerated AI inference will find the NVIDIA part far more capable. The N1 also has a massive memory advantage: 128 GB of LPDDR5X on a 256-bit bus with 273.2 GB/s of bandwidth, compared to the Intel part’s system-shared memory with system-dependent bandwidth.

Architecture Differences

The two GPUs come from different foundries and process nodes. Intel uses its own 10 nm process for the Meteor Lake chip, while NVIDIA uses TSMC’s 5 nm node for the GB20B die. The N1 20SM die measures 382 mm², a large integrated die for a mobile IGP. Intel’s die size is not recorded in the database. The process difference alone does not dictate performance, but it does help explain the N1’s higher transistor density potential and efficiency characteristics.

The Intel Arc Graphics 112EU Mobile uses the Xe-LPG architecture, which is Intel’s gaming-oriented graphics architecture for Meteor Lake. It has no dedicated ray tracing units and no tensor cores in this configuration. The shading array consists of 896 shaders, 56 TMUs, and 24 ROPs. Memory is entirely system-shared, with no dedicated VRAM, and the bus interface is listed as Ring Bus, tying it directly to the CPU’s internal fabric. The boost clock is 2200 MHz, and the base clock is 300 MHz. Power draw is listed at 65 W, which is high for an integrated part but plausible for a Meteor Lake mobile solution under load.

The NVIDIA N1 20SM uses the Blackwell 2.0 architecture on the GB20B chip. It is a substantially larger configuration with 2560 shaders, 160 TMUs, and 24 ROPs. The 20 ray tracing cores and 80 tensor cores are the headline architectural additions. The memory subsystem is completely different: 128 GB of LPDDR5X on a 256-bit interface, running at 1067 MHz with 8.5 Gbps effective speed, yielding 273.2 GB/s of bandwidth. The bus interface is PCIe 5.0 x16, and the display output is listed as a single HDMI port. The TDP is unknown in the database, but the power connectors are listed as None, consistent with an IGP.

The API support also diverges sharply. Intel lists DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.4. NVIDIA lists N/A for DirectX, OpenGL, and Vulkan, which suggests the N1 20SM may not be intended for conventional PC graphics API workloads. The two parts are architecturally aimed at different roles: the Intel part is a general-purpose mobile GPU for graphics and compute, while the NVIDIA part appears optimized for bandwidth-heavy, tensor-heavy, or ray-tracing workloads, possibly in a data-center or edge context.

Head-to-Head Benchmarks

The database contains no head-to-head benchmark entries for these two parts, and neither has individual benchmark scores. The comparison must therefore rely on specification-derived metrics. The largest numerical gaps are in texture rate, memory bandwidth, and FP32 throughput.

Texture rate: the NVIDIA N1 20SM delivers 375.4 GTexel/s against the Intel Arc 112EU’s 123.2 GTexel/s. That is a 3.05x advantage for NVIDIA, driven by the 160 TMUs versus 56. For texture-bound workloads, the N1 will complete roughly three times the texture fetch work per second, assuming similar efficiency per TMU.

FP32 throughput: 12.01 TFLOPS for NVIDIA versus 3.942 TFLOPS for Intel, a 3.05x difference. The 2560 shaders at 2346 MHz produce this figure, while Intel’s 896 shaders at 2200 MHz produce the lower result. This affects all general compute, vertex processing, and non-accelerated graphics paths.

FP16 throughput: NVIDIA lists 12.01 TFLOPS at a 1:1 ratio with FP32, meaning full-rate half-precision. Intel lists 7.885 TFLOPS at a 2:1 ratio, which is exactly double its FP32 rate. Even with the ratio advantage, Intel’s absolute FP16 throughput is still lower than NVIDIA’s by roughly 1.52x.

Pixel rate: NVIDIA’s 56.30 GPixel/s narrowly edges out Intel’s 52.80 GPixel/s, a 6.6% advantage. Both parts have 24 ROPs, and the difference comes from the higher boost clock on the NVIDIA part. Raster output is the one metric where the two parts are close.

Memory bandwidth: NVIDIA provides 273.2 GB/s from dedicated LPDDR5X, while Intel’s bandwidth is listed as System Dependent. The 256-bit bus versus a shared system memory interface means the N1 has a guaranteed, fixed bandwidth ceiling, while the Intel part’s bandwidth varies with the host memory configuration. In any scenario with heavy memory traffic, the NVIDIA part has a structural advantage that cannot be matched by the Intel IGP.

Clock speeds: Intel boosts to 2200 MHz, NVIDIA boosts to 2346 MHz. NVIDIA’s base clock of 741 MHz is higher than Intel’s 300 MHz base, indicating the N1 sustains higher idle-to-load operating points. Neither part lists a game clock.

FAQ

Q: Which GPU has higher FP32 compute throughput?

A: The NVIDIA N1 20SM, with 12.01 TFLOPS versus 3.942 TFLOPS for the Intel Arc Graphics 112EU Mobile, a 3.05x advantage.

Q: Does the Intel Arc 112EU support DirectX 12?

A: Yes, the Intel part lists DirectX 12 (12_1), along with OpenGL 4.6 and Vulkan 1.4. The NVIDIA N1 20SM lists N/A for all three APIs.

Q: How much memory does each GPU provide?

A: The NVIDIA N1 20SM has 128 GB of LPDDR5X on a 256-bit bus with 273.2 GB/s bandwidth. The Intel Arc 112EU uses system shared memory with system-dependent bandwidth.

Q: Which GPU has ray tracing cores?

A: Only the NVIDIA N1 20SM, which includes 20 ray tracing cores and 80 tensor cores. The Intel Arc 112EU has neither.

Q: What is the process node for each GPU?

A: Intel uses a 10 nm process at Intel for the Meteor Lake chip. NVIDIA uses TSMC’s 5 nm process for the GB20B chip, with a 382 mm² die size.

Q: Are these GPUs similar in raster output?

A: They are close. NVIDIA’s pixel rate is 56.30 GPixel/s, while Intel’s is 52.80 GPixel/s, a 6.6% difference. Both have 24 ROPs.

Specification Differences

• Process node: Intel 10 nm versus TSMC 5 nm for NVIDIA.
• Die size: Intel not recorded; NVIDIA 382 mm².
• Architecture: Intel Xe-LPG versus NVIDIA Blackwell 2.0.
• Chip: Intel Meteor Lake versus NVIDIA GB20B.
• Shading units: 896 versus 2560.
• Texture mapping units: 56 versus 160.
• Ray tracing cores: none versus 20.
• Tensor cores: none versus 80.
• Boost clock: 2200 MHz versus 2346 MHz.
• Base clock: 300 MHz versus 741 MHz.
• FP32 throughput: 3.942 TFLOPS versus 12.01 TFLOPS.
• FP16 throughput: 7.885 TFLOPS (2:1) versus 12.01 TFLOPS (1:1).
• Texture rate: 123.2 GTexel/s versus 375.4 GTexel/s.
• Pixel rate: 52.80 GPixel/s versus 56.30 GPixel/s.
• Memory size: System Shared versus 128 GB.
• Memory type: System Shared versus LPDDR5X.
• Memory bus: System Shared versus 256 bit.
• Memory bandwidth: System Dependent versus 273.2 GB/s.
• Memory clock: System Shared versus 1067 MHz 8.5 Gbps effective.
• TDP: 65 W versus unknown.
• Bus interface: Ring Bus versus PCIe 5.0 x16.
• Display outputs: Portable Device Dependent versus 1x HDMI.
• Power connectors: none listed versus None.
• API support: DirectX 12 (12_1), OpenGL 4.6, Vulkan 1.4 versus N/A for all three.
• Release date: 2023-12-13 versus 2026-05-31.