Intel has been quiet regarding its next generation of Xeon Scalable processors, but samples are beginning to float around. This week, one leaker got their hands on one of the upcoming processors and put it to the test in a range of benchmarks.
Chinese leaker, @yuuki_ans, ran the engineering sample through AIDA64 cache and memory tests and many synthetic benchmarks, including CPU-Z, Cinebench R15, R20, and R23, as well as V-Ray. Looking at his data, this Intel CPU features 48x cores and 96x threads and has a base frequency of 2.3GHz, boosting up to 3.3GHz. The TDP is set at 270W, and there's 90MB of L3 cache, but there's no sign of HBM2e in this engineering sample.
Frequency: 2.30GHz – 3.30GHz
L3 Cache: 90MB
— 结城安穗-YuuKi_AnS🇨🇳 (@yuuki_ans) February 12, 2022
Despite being an engineering sample, the Sapphire Rapids CPU was slightly faster than its predecessor in both single and multi-core workloads. Compared to the EPYC 7773X, the AMD server processor had the upper hand in most multi-core scenarios, as expected due to featuring more cores/threads. However, the Intel chip did shine in Cinebench single-core tests, outperforming the red team CPU by a fair margin, although this trend didn't continue with the CPU-Z 1T-test.
As per the leaker, this could be due to AVX-512 support being a paid feature of the Intel chip. In addition, the leaker suggests the chip's HBM2e functions may also be behind a paywall. Finally, it's also worth mentioning that although the boost clock is specified to be 3.3GHz, the Intel processor could only achieve a maximum single-core clock of 2.9GHz. So there's clearly room for improvement.
Discuss on our Facebook page, HERE.
KitGuru says: These results aren't enough to pinpoint the exact level of performance of the upcoming Sapphire Rapids CPUs, but it is a good starting point. Considering these results are from an engineering sample, performance should improve in launch samples.