Intel vs AMD: EPYC Siena rivals Sapphire Rapids with less power

AMD EPYC Siena Performance

Both of AMD’s new Zen 4c equipped EPYC 8324P and 8324PN (Siena) server CPUs were recently tested in Linux by Phoronix, and the results showed that both chips deliver exceptional performance when used in server applications. Phoronix’s tests revealed that both of AMD’s Zen 4c processors were capable of matching the performance of an Intel Xeon Gold 6421N 32-core CPU in most tasks, although using a far lower amount of power than the Intel processor.

AMD’s EPYC most recent range of EPYC server CPUs, the 8004 series, focuses on power economy rather than sheer CPU performance. This is reflected in the 8324P and 8324PN models of the EPYC server CPUs. The new Zen 4c core, which is included into all of AMD’s 8004 processors, is 35% more compact than its predecessor, Zen 4, and boasts improved power efficiency compared to Zen 4’s bigger version. The raw performance of an AMD Zen 4c core is lower than that of a normal Zen 4 core. This is the single drawback of the Zen 4c core.

The 8324P and PN are located squarely in the centre of the CPU stack for AMD’s 8004 series. They include 32 cores, 128 MB of L3 cache, six DDR5 memory channels, and 96 PCIe Gen 5 lanes. They are distinct from one another in terms of clock speeds and power ratings; for example, the 8324P has a base clock of 2.65 GHz, a boost clock of 3 GHz, and a TDP of 180W. The 8324PN, on the other hand, has a base speed of 2.05 GHz, a boost frequency of 3 GHz, and a TDP that is significantly lower at 130 watts.

This is due to the fact that the PN series of 8004 CPUs were developed specifically for installations that must comply with the Network Equipment Building System (NEBS), which calls for systems that are able to survive higher operational temperature ranges than ordinary chips.

Low Power Draw AMD EPYC Siena

Phoronix evaluated the performance of the two chips in a variety of different benchmarks, which together equate to seven distinct categories. These categories include artificial intelligence workloads, video encoding, code compiling, file compression/decompression, high-performance computing, and other tasks.

When both processors were set to their highest power settings, the performance of the two AMD CPUs was superior to that of the Intel Xeon Gold 6421N 32 core. The Xeon 6421N was 2.7% faster than the 8324P when it was configured with its default settings, 2.5% slower when it was configured with its lowest adjustable TDP of 155W, and 5% faster when it was configured with the chip’s performance determinism mode.

It should not come as a surprise that the 8324PN ran at a little slower speed due to the lower power rating and base clock parameters. The Intel Xeon was 11.8% faster than the 8324PN in its stock configuration; however, the 8324PN was still able to outperform the chip in its performance determinism mode, achieving a 4.8% advantage over the Intel CPU.

AMD EPYC Siena Optimization

Even with performance determinism mode turned on, the 8324P and 8324PN were able to provide average power consumption ratings that were considerably below the 100W threshold. This was one of the many ways in which AMD prioritized power efficiency. On the other hand, Intel’s counterpart to Sapphire Rapids, which had an average power consumption rating of 137 watts, was unable to come even near to matching AMD’s level of power efficiency.

Phoronix’s testing has shown that AMD’s most recent Zen 4c core is quite powerful and is capable of producing performance that can compete with the absolute best that Intel has to offer right now. Although Intel may have beaten AMD to the E-core game with the release of Alder Lake, the most recent benchmarks demonstrate that AMD is capable of producing a delectable efficiency core that is capable of competing with Intel’s high-performance cores under certain conditions.

Source