Intel Sapphire Rapids processor benchmark scores leaked

Sapphire Rapids is Intel’s next-generation Xeon processors that are divided into two categories, the difference being whether they are equipped with HBM2e memory. It is said that the Sapphire Rapids processor configured with memory will use four groups of HBM2e, each with a capacity of 16GB, a total of 64GB of memory, and a peak bandwidth between 1.432 TB/s and 1.640 TB/s, and will share the socket with the regular version of the Sapphire Rapids processor.

According to Wccftech reports, some netizens recently exposed the benchmark test results of the normal version of the Sapphire Rapids processor. The information shows that this engineering sample is a specification of 48 cores and 96 threads, the base clock is 2.3 GHz, the boost clock is 3.3 GHz, it is equipped with 90MB of L3 cache, the TDP is 270W, and it is not equipped with HBM2e memory. This is not the top-of-the-line Sapphire Rapids processor. According to Intel’s plan, it can be equipped with up to 56 cores, 112 threads, and a TDP of 350W.

Benchmarks were conducted on a two-socket platform, and also provided test scores for AMD EPYC 7773X (Milan-X/64 cores, 128 threads) and Intel Xeon Platinum 8380 processors (Ice Lake-SP/40 cores, 80 threads). In the AIDA64 cache and memory benchmark, thanks to the 3D vertical cache (3D V-Cache) technology, the EPYC 7773X processor has a total of 768MB of L3 cache to exert its power and has better L2/L3 cache speed. The Sapphire Rapids processor leads its L1 cache speed with 8-channel DDR5 memory.

In the Cinebench R15 test, the Sapphire Rapids processor maintained its single-core performance advantage, but the multi-core performance still trailed the EPYC 7773X processor.


It is understood that the Sapphire Rapids processor of the Eagle Stream platform uses the core of the Golden Cove architecture and is manufactured using the 10nm Enhanced SuperFin process. The new platform also supports PCIe Gen5, CXL 1.1 (Compute Express Link), and eight-channel DDR5 memory, while continuing Intel’s built-in AI acceleration strategy and supporting Intel Advanced Matrix Extensions (AMX).