Intel’s been having a tough go of it lately with some of their silicon. First their Atom SoCs were causing some Cisco gear to brick back in February. Now comes this news of issues with HyperThreading on Skylake and Kaby Lake CPUs. This seems limited to a relatively specific workloads, but has a wide range of effected processors. Most desktop CPUs in the last couple years, and recent Xeon E3s are subject to the error.
AMD Epyc sounds pretty epic, with epoch-defining memory, I/O, and even cores of a dual-socket server in a single socket. And that’s something to get excited about, especially considering that the Zen cores inside these chips are almost at IPC parity with Intel’s latest, and can handle dual threads like Intel, too.
The race for desktop CPU dominance has proceeded in fits and starts for three decades, with improvements in architecture, manufacturing process, and clock speed ratcheting up performance. Now Intel is announcing high-end desktop CPU’s with many, many cores, including the long-awaited Core i9 series.
Ever since Microsoft and Intel declared that the combination of Windows and Nehalem could deliver over a million iSCSI IOPS, I’ve been curious about just how they did it. What black magic could push that many I/Os over a single Ethernet connection? And what was on the other end? Now Intel has revealed all in a whitepaper, and the results are surprising!
In case you’re wondering—OK, so it’s probably only geeks like me that wonder about these sorts of things—putting a Xeon 5500-based server into a cluster with older servers and using EVC to ensure VMotion compatibility won’t impact the performance benefits of the Xeon 5500.
Cisco today announced a new Unified Computing System (UCS) server form factor: The C-Series rack-mount server. The C-Series features the same features found in the existing B-Series UCS blades but starts much smaller and cheaper. Cisco hopes to extend the UCS vision into small businesses, remote offices, and other locations where a blade server chassis would not make sense.