Intel Arc G3 Extreme vs NVIDIA GeForce RTX 5090 Mobile Comparison

Head-to-Head Benchmarks

The recorded data for the Intel Arc G3 Extreme contains no benchmark entries, while the NVIDIA GeForce RTX 5090 Mobile has a full suite of ten recorded tests. This makes direct numerical comparison impossible for most workloads, but the NVIDIA part's results can be examined on their own terms and against its nearest rivals in the database.

The GeForce RTX 5090 Mobile delivers an average benchmark score of 45,152 across all recorded tests. That places it at the 84th percentile of all GPUs in the database. Its nearest rival, the AMD Radeon Pro 5500 XT, scores 45,384, which is 0.5% higher. The NVIDIA GeForce RTX 4070 Ti trails by 0.8% with a score of 44,795. The Intel Arc A730M sits 1% ahead at 45,592, and the NVIDIA RTX 5880 Ada Generation leads by 1.8% at 45,972.

In individual tests, the RTX 5090 Mobile shows a clear split between compute-oriented and graphics-oriented workloads. Its Geekbench OpenCL score reaches 201,834, while the Vulkan result is close behind at 198,405. The PassMark G3D score of 30,034 is the strongest traditional graphics result. The PassMark GPU Compute result of 13,401 indicates that compute throughput is substantial but not dominant relative to its own graphics scores.

The DirectX legacy tests reveal an interesting pattern. PassMark DirectX 9 scores 324, which is the highest of the four DirectX tests. DirectX 11 scores 269, DirectX 10 scores 183, and DirectX 12 scores 138. The DirectX 12 result being the lowest among these is notable, as modern workloads typically favor newer API versions. The 2D graphics score of 1,057 is modest compared to the 3D result.

The 3DMark Steel Nomad DX12 test produces a score of 5,871. This is the only modern synthetic gaming benchmark in the dataset. When combined with the PassMark G3D result, the data indicates that the RTX 5090 Mobile is a high-end mobile part that competes with desktop-class GPUs from the previous generation. Its average score of 45,152 places it within 2% of the four nearest rivals listed, all of which are desktop or high-end mobile parts.

The Intel Arc G3 Extreme has no recorded benchmark scores and an average score of zero. Its percentile ranking of 50 is a default value, not a measured result. The database contains no wins for the Intel part and no wins for the NVIDIA part in head-to-head testing, because no head-to-head benchmark entries exist. The data confirms that the NVIDIA part is fully characterized, while the Intel part currently lacks empirical performance data.

Architecture Differences

The two GPUs come from different manufacturers and use fundamentally different process technologies. Intel builds the Arc G3 Extreme on a 3 nm node at its own foundry. NVIDIA builds the RTX 5090 Mobile on a 5 nm node at TSMC. The Intel chip, codenamed Panther Lake, uses the Xe3-LPG architecture and belongs to the Arc Graphics-M (Panther Lake) generation. NVIDIA's chip, GB203, uses the Blackwell 2.0 architecture and belongs to the GeForce 50 Mobile generation.

The transistor counts differ dramatically. NVIDIA's GB203 contains 45,600 million transistors on a 378 mm² die, yielding a transistor density of 120.6M per mm². The Intel part's transistor count and die size are listed as unknown, so no density comparison is possible. The manufacturing process difference partially explains the NVIDIA die size, as a larger node typically requires more area for the same transistor count.

Clock behavior also differs. The Intel part has a base clock of 300 MHz and a boost clock of 2500 MHz. The NVIDIA part has a base clock of 990 MHz and a boost clock of 1515 MHz. The Intel part's boost clock is substantially higher, but its base clock is much lower. The NVIDIA part runs its memory at 1750 MHz with 28 Gbps effective speed. The Intel part uses system shared memory with no dedicated memory clock.

The memory subsystems are completely different. NVIDIA uses 24 GB of GDDR7 on a 256 bit bus, delivering 896.0 GB/s of bandwidth. Intel uses system shared memory, meaning the GPU draws from the host system's RAM. The memory type, bus width, and bandwidth are all listed as system dependent. This is a fundamental architectural difference: the NVIDIA part has dedicated, high-bandwidth VRAM, while the Intel part relies on shared system resources.

The compute unit counts show NVIDIA's advantage in raw hardware. NVIDIA has 10,496 shading units, 328 TMUs, 112 ROPs, 82 RT cores, and 328 tensor cores. Intel has 1,536 shading units, 48 TMUs, 24 ROPs, and 12 RT cores. The Intel part has no tensor cores listed. NVIDIA's pixel rate is 169.7 GPixel/s versus Intel's 60.00 GPixel/s. NVIDIA's texture rate is 496.9 GTexel/s versus Intel's 120.0 GTexel/s.

Floating point performance follows the same pattern. NVIDIA delivers 31.80 TFLOPS for FP32 and 31.80 TFLOPS for FP16 with a 1:1 ratio. Intel delivers 7.680 TFLOPS for FP32 and 15.36 TFLOPS for FP16 with a 2:1 ratio. NVIDIA's FP32 throughput is roughly four times higher, while its FP16 throughput is roughly two times higher.

Both parts support DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. Both are integrated into the host device with no slot width, no power connectors, and portable-device-dependent display outputs. The NVIDIA part uses a PCIe 5.0 x16 bus interface, while the Intel part uses an IGP interface.

Where Each One Wins

The NVIDIA GeForce RTX 5090 Mobile wins every category where data exists. Its 24 GB of GDDR7 memory with 896.0 GB/s bandwidth is clearly superior to shared system memory. Its 82 RT cores and 328 tensor cores provide dedicated hardware for ray tracing and AI workloads, while the Intel part has 12 RT cores and no tensor cores. The NVIDIA part's 31.80 TFLOPS FP32 throughput more than quadruples the Intel part's 7.680 TFLOPS.

The Intel Arc G3 Extreme has no benchmark scores in the database, so it cannot claim a measured win in any workload category. Its architectural advantages are limited to its higher boost clock of 2500 MHz and its smaller 3 nm process node. The 2:1 FP16 ratio means the Intel part can reach 15.36 TFLOPS for FP16 work, but this is still below NVIDIA's 31.80 TFLOPS for the same precision.

The Intel part's system shared memory could be advantageous in specific system configurations where unified memory access reduces data copying overhead. However, the database records no benchmark data to confirm this. The NVIDIA part's 95 W TDP is higher than the Intel part's 80 W TDP, which may affect system thermal budgets, but the performance gap in raw specifications is large enough that the NVIDIA part is the clear choice for demanding workloads.

For legacy DirectX workloads, the RTX 5090 Mobile shows higher scores in DirectX 9 (324) than DirectX 12 (138). This suggests the part handles older APIs well, though the reasons are not documented in the data. The Intel part has no comparable scores.

Specification Differences

The two parts differ in nearly every specification field:

| Specification | Intel Arc G3 Extreme | NVIDIA GeForce RTX 5090 Mobile |

|---|---|---|

| Process node | 3 nm | 5 nm |

| Foundry | Intel | TSMC |

| Transistors | unknown | 45,600 million |

| Die size | unknown | 378 mm² |

| Base clock | 300 MHz | 990 MHz |

| Boost clock | 2500 MHz | 1515 MHz |

| Memory size | System Shared | 24 GB |

| Memory type | System Shared | GDDR7 |

| Memory bus width | System Shared | 256 bit |

| Memory bandwidth | System Dependent | 896.0 GB/s |

| Shading units | 1,536 | 10,496 |

| TMUs | 48 | 328 |

| ROPs | 24 | 112 |

| RT cores | 12 | 82 |

| Tensor cores | null | 328 |

| Pixel rate | 60.00 GPixel/s | 169.7 GPixel/s |

| Texture rate | 120.0 GTexel/s | 496.9 GTexel/s |

| FP32 | 7.680 TFLOPS | 31.80 TFLOPS |

| FP16 | 15.36 TFLOPS (2:1) | 31.80 TFLOPS (1:1) |

| TDP | 80 W | 95 W |

| Bus interface | IGP | PCIe 5.0 x16 |

| Release date | 2026-05-31 | 2025-03-26 |

| Predecessor | null | GeForce 40 Mobile |

The NVIDIA part has a later predecessor listed, while the Intel part has none. Both are marked as Active production status. Neither has a launch MSRP in the database. The NVIDIA part uses a dedicated memory bus and PCIe interface, while the Intel part is fully integrated.

FAQ

Q: How does the Intel Arc G3 Extreme compare to the NVIDIA GeForce RTX 5090 Mobile in raw compute performance?

A: The NVIDIA part delivers 31.80 TFLOPS FP32 and 31.80 TFLOPS FP16. The Intel part delivers 7.680 TFLOPS FP32 and 15.36 TFLOPS FP16. NVIDIA is roughly four times faster in FP32 and two times faster in FP16.

Q: What memory configurations do the two GPUs use?

A: The RTX 5090 Mobile uses 24 GB of GDDR7 on a 256 bit bus with 896.0 GB/s bandwidth. The Arc G3 Extreme uses system shared memory, with size, type, bus width, and bandwidth all dependent on the host system.

Q: Which GPU has more ray tracing cores?

A: The RTX 5090 Mobile has 82 RT cores. The Arc G3 Extreme has 12 RT cores.

Q: What is the average benchmark score for each GPU?

A: The RTX 5090 Mobile has an average benchmark score of 45,152 across ten recorded tests. The Arc G3 Extreme has no recorded benchmarks and an average score of zero.

Q: Are there any benchmarks where the Arc G3 Extreme wins?

A: The database contains no benchmark entries for the Intel part, so no wins are recorded for it in any test.

Q: What process nodes do the two GPUs use?

A: Intel builds the Arc G3 Extreme on a 3 nm node at its own foundry. NVIDIA builds the RTX 5090 Mobile on a 5 nm node at TSMC.

The Verdict

The data points to a single conclusion: the NVIDIA GeForce RTX 5090 Mobile is the stronger part in every measurable aspect. Its 31.80 TFLOPS FP32 throughput, 896.0 GB/s memory bandwidth, 82 RT cores, and 328 tensor cores place it at the 84th percentile of all GPUs in the database. Its average benchmark score of 45,152 puts it within 2% of desktop-class rivals like the AMD Radeon Pro 5500 XT and NVIDIA RTX 5880 Ada Generation.

The Intel Arc G3 Extreme has no recorded performance data. Its specifications show a capable integrated part with a 3 nm process, a 2500 MHz boost clock, and 12 RT cores, but the lack of benchmarks means the database cannot validate its real-world behavior. Its 80 W TDP is lower than the NVIDIA part's 95 W, and its system shared memory approach may simplify system design, but these are architectural traits rather than demonstrated performance advantages.

For users who need a mobile GPU with known performance, the RTX 5090 Mobile is the only part with empirical data in the database. Its 24 GB GDDR7 memory and 896.0 GB/s bandwidth make it suitable for memory-intensive workloads. Its DirectX 9 score of 324 versus DirectX 12 score of 138 suggests strong legacy API performance. The Intel part requires further benchmarking before it can be assessed against the NVIDIA part on equal terms. The recorded data favors NVIDIA decisively.