The alleged early benchmark results of one of Intel’s upcoming Arc Alchemist graphics cards with a GPU featuring 512 vector engines have been published in SiSoftware Sandra’s Benchmark Ranker and discovered by @Tum_Apisak. The overall GPU compute score of Intel’s upcoming discrete graphics processing unit (GPU) is higher than that of Nvidia’s high-end GeForce RTX 3070 Ti. But there’s a catch, so take these scores with a grain of salt for now.
The score for Intel’s presumably flagship Arc Alchemist discrete GPU was submitted by an unknown person(s) to the database. The card’s GPU features 512 enabled vector engines (which correspond to 4096 stream processors), has 4MB of L2 cache, and runs at 2.1GHz. The card is equipped with 12.8 GB of usable memory, which probably means 16 GB of memory in total. See more details on the test setup below.
Intel’s Arc Alchemist Xe-HPG 512VE/4096SP vs. Nvidia’s GeForce RTX 3070 Ti
| Alchemist Bow Xe-HPG 512 (MPix/s) | GeForce RTX 3070 Ti (MPix/s) | |
| FP16 | Half-floating GP | 35,093 | 36,510 |
| FP32 | Single float GP | 20,888 | 27,029 |
| FP64 | Double Float GP | 1,000 | 594 |
| FP128 | four-float GP | 109 | 22 |
| Goal | 9017.52 | 8369.51 |
Assuming the results submitted to SiSoftware Sandra’s Benchmark Ranker are accurate and the drivers are mature enough, the overall score for Intel’s Arc Alchemist graphics card with a 512 vector engine GPU of 9017.52 Mpix/s seems good enough. Surpassing Nvidia’s GeForce RTX 3070 Ti, it reaches 8369.51 MPix/s. But the devil is in the details.
When it comes to FP32 GPU compute workloads – a precision that graphics applications use the most these days – Intel’s Arc Alchemist discrete GPU with 4096 stream processors at 2.10 GHz is at around 23 % behind its competitor GA104 with 6144 stream processors at 1440 MHz – 1850 MHz. That’s not exactly a spectacular result, at least when it comes to the performance of FP32 compute workloads. Of course, we’re not dealing with commercial hardware or software here, so take the information with a grain of salt. Perhaps Intel’s flagship Arc Alchemist will perform considerably better.
Intel’s Arc Alchemist Xe-512 performs on par with Nvidia’s GeForce RTX 3070 Ti in FP16 GPU compute workloads, which aren’t widely used for graphics. Meanwhile, the Xe-HPG GPU outperforms Nvidia’s GA104 by 68% in FP64 GPU compute workloads and nearly 5x (or 395%) in FP128 GPU compute workloads, but neither FP64 nor FP128 are not used by graphics applications and are reserved. for high performance computing.
The result published in SiSoftware Sandra’s Benchmark Ranker is marked as ‘Intel(R) Graphics d gfx-driver-ci-master-10002 DCH Release Internal (4096S 512C SM3.0 2.1GHz, 4MB L2, 12.8GB) (OpenCL),’ so we’re dealing with a sample and the driver for internal testing. Perhaps the most suspicious part of this test of Intel’s next graphics card is that it was performed on Intel’s “CoffeeLake Client Platform CannonLake Client System (Intel CoffeeLake S UDIMM RVP)”, which does not is not a game system at all.
While the Intel Arc Alchemist Xe-512 benchmark results should be taken with discretion, these are the early performance numbers for the part, so we have at least some food for thought.
