At CoolNation server specs aren’t our bread and butter. Enthusiast-grade is. We want the best performance for gaming, not for webservers or 3rd film rendering. Yet the two aren’t so different, and in the CPU wars Intel is firing a shot just before the year ends.
According to Intel, a server running two third-gen 32-core Xeon Scalable ‘Ice Lake’ CPUs clocked at 2.20 GHz and paired with 256 GB of DDR4-3400 memory is 20% ~ 30% faster in select benchmarks (LAMMPS, NAMD STMV, Monte Carlo) when compared to a server powered by two 64-core AMD EPYC 7742 processors clocked at 2.25 GHz and equipped with 256 GB of DDR4-3200. That’s despite the fact that Intel’s system has half the number of cores.
To put this comparison into perspective, Intel’s CPUs have always supported better memory throughput. AMD’s strategy has been to focus on sheer processing power, which is why the past Ryzen 3990X and now current Ryzen 5950X are so crazy-fast. That’s in part why AMD absolutely destroyed Intel’s double-the-cost CPU earlier this year.
So Intel claiming, and confirming to the public, that the next gen Xeon processors are showing a similar performance jump to AMD’s from Ryzen 3000 to Ryzen 5000? The next 2-3 years of CPU tech are going to be very interesting, as Intel moves to 7nm and tries to reclaim territory lost to AMD.