Intel’s New Arc 140V “Xe2” GPU Outshines Radeon 780M and Rivals GTX 1650 in CompuBench Trials

Intel’s Arc 140V integrated GPU, featuring Xe2 GPU cores, has been examined in CompuBench, revealing impressive graphics performance. The Lunar Lake CPUs are undergoing increased testing as the launch nears. While several benchmarks for the Core Ultra 200V CPUs have emerged, fewer evaluations for their Xe2 iGPUs have appeared. Recently, the Intel Arc 140V integrated graphics, included in Intel Core Ultra 7 and Ultra 9 processors, underwent another assessment on CompuBench, this time showcasing a more accurate performance representation.

Before diving into the results, it’s important to note that CompuBench primarily provides OpenCl results, which may not be the best platform for GPU performance comparisons. However, new GFXBench benchmarks indicate that the Arc 140V is competitive with entry-level discrete GPUs, such as NVIDIA’s GTX 1650, and approaches the performance of the Radeon 890M iGPU, which features 16 compute units based on the RDNA 3.5 architecture.

The Compute tab shows significant improvements in the Level Set Segmentation -128 and 256 scores, with the Arc 140V demonstrating a 40% and 10% uplift, respectively, compared to previous tests. This performance increase widens the gap between the Arc 140V and the Radeon 780M, which scored 2390.41mVoxels/s and 2972.968mVoxels/s, respectively, likely due to the latest Intel GPU driver support.

The table below summarises the performance results:

Graphics Test:
– Intel Arc 140V “Xe2”
– AMD 890M “RDNA 3.5”
– NVIDIA GTX 1650
– Intel Arc A730M
– AMD 780M “RDNA 3”

– Level Set Segmentation – 128: 3780.274 mVoxels/s | 5269.016 mVoxels/s | 4054.334 mVoxel/s | 3394.234 mVoxel/s | 2390.241 mVoxels/s
– Level Set Segmentation – 256: 5325.429 mVoxels/s | 6006.334 mVoxels/s | 4084.295 mVoxel/s | 3480.865 mVoxel/s | 2972.968 mVoxels/s
– Ocean Surface Simulation: 1536.516 Iterations/s | 1093.617 Iterations/s | 1369.795 Iterations/s | 1347.085 Iterations/s | 923.408 Iterations/s
– Catmull-Clark Subdivision Level 3: 178.286 mTriangles/s | 156.979 mTriangles/s | 158.324 mTriangles/s | 120.751 mTriangles/s | 129.778 mTriangles/s
– Catmull-Clark Subdivision Level 5: 118.072 mTriangles/s | 80.397 mTriangles/s | 173.261 mTriangles/s | 116.816 mTriangles/s | 83.533 mTriangles/s
– Vertex Connection and Merging: 4.853 mPixels/s | 3.904 mPixels/s | 7.189 mPixels/s | 6.102 mPixels/s | 3.495 mPixels/s
– Subsurface Scattering: 4147.462 mSample/s | 3169.361 mSample/s | 3406.279 mSample/s | 2598.702 mSample/s | 3056.836 mSample/s
– Subsurface Scattering multiple view: 3831.172 mSample/s | 3022.308 mSample/s | 3258.196 mSample/s | 2479.355 mSample/s | 2802.52 mSample/s
– TV-L1 Optical Flow: 22.1 mPixels/s | 18.577 mPixels/s | 22.692 mPixels/s | 18.834 mPixels/s | 25.251 mPixels/s

The averages recorded were:
– Intel Arc 140V “Xe2”: 100%
– AMD 890M “RDNA 3.5”: 104%
– NVIDIA GTX 1650: 103%
– Intel Arc A730M: 83%
– AMD 780M “RDNA 3”: 73%

In graphics tests, the Arc 140V showed strong performance ahead of the 780M in several areas. However, in Aztec Ruins tests, both Normal Tier and High Tier, the 140V scored 74.9 fps and 50.3 fps respectively, compared to the 780M’s 80.1 fps and 60.9 fps. More comprehensive real-world benchmarks are needed to draw definitive conclusions.

It’s also essential to consider the varying TDPs, with AMD’s Ryzen AI 9 HX 370 APUs configurable from 15W to 54W, while discrete GPUs range from 35-75W. The Arc 140V shares TDP with the Lunar Lake SOC, rated at 17W at PL1 and peaking at 30W at PL2.

High-Level Tests:
– Intel Arc 140V (8GB)
– AMD Radeon 780M

– Aztec Ruins Normal Tier: 4817 Frames (74.9 fps) vs. 5149 Frames (80.1 fps)
– Aztec Ruins High Tier: 3232 Frames (50.3 fps) vs. 3913 Frames (60.9 fps)

The Arc 140V was tested with 8GB of shared system memory, operating at 2000 MHz, likely on an Intel Core Ultra 266V or 268V. This GPU features Xe2-LPG architecture and is part of the mobile-specific Lunar Lake chips, expected to launch on September 3rd.

Intel Lunar Lake “Core Ultra 200V” CPU Lineup:

– Core Ultra 9 288V: 8/8 cores/threads, 12 MB cache, 5.1 / 3.7 GHz P-Core/E-Core boost, Arc 140V @ 2.05 GHz GPU, 30W/30W PL1/PL2, 32 GB LPDDR5X, NPU/XMX (GPU) TOPs 48/67
– Core Ultra 7 268V: 8/8 cores/threads, 12 MB cache, 5.0 / 3.7 GHz P-Core/E-Core boost, Arc 140V @ 2.00 GHz GPU, 17W/30W PL1/PL2, 32 GB LPDDR5X, NPU/XMX (GPU) TOPs 48/66
– Core Ultra 7 266V: 8/8 cores/threads, 12 MB cache, 5.0 / 3.7 GHz P-Core/E-Core boost, Arc 140V @ 2.00 GHz GPU, 17W/30W PL1/PL2, 16 GB LPDDR5X, NPU/XMX (GPU) TOPs 48/66
– Core Ultra 7 258V: 8/8 cores/threads, 12 MB cache, 4.8 / 3.7 GHz P-Core/E-Core boost, Arc 140V @ 1.95 GHz GPU, 17W/30W PL1/PL2, 32 GB LPDDR5X, NPU/XMX (GPU) TOPs 47/64
– Core Ultra 7 256V: 8/8 cores/threads, 12 MB cache, 4.8 / 3.7 GHz P-Core/E-Core boost, Arc 140V @ 1.95 GHz GPU, 17W/30W PL1/PL2, 16 GB LPDDR5X, NPU/XMX (GPU) TOPs 47/64
– Core Ultra 5 238V: 8/8 cores/threads, 8 MB cache, 4.7 / 3.5 GHz P-Core/E-Core boost, Arc 130V @ 1.85 GHz GPU, 17W/30W PL1/PL2, 32 GB LPDDR5X, NPU/XMX (GPU) TOPs 40/53
– Core Ultra 5 236V: 8/8 cores/threads, 8 MB cache, 4.7 / 3.5 GHz P-Core/E-Core boost, Arc 130V @ 1.85 GHz GPU, 17W/30W PL1/PL2, 16 GB LPDDR5X, NPU/XMX (GPU) TOPs 40/53
– Core Ultra 5 228V: 8/8 cores/threads, 8 MB cache, 4.5 / 3.5 GHz P-Core/E-Core boost, Arc 130V @ 1.85 GHz GPU, 17W/30W PL1/PL2, 32 GB LPDDR5X, NPU/XMX (GPU) TOPs 40/53
– Core Ultra 5 226V: 8/8 cores/threads, 8 MB cache, 4.5 / 3.5 GHz P-Core/E-Core boost, Arc 130V @ 1.85 GHz GPU, 17W/30W PL1/PL2, 16 GB LPDDR5X, NPU/XMX (GPU) TOPs 40/53