JCC Erratum Impact On Skylake Xeon Scalable Plus The Patched Assembler

Last week ago we provided a number of benchmarks looking at the performance impact from Intel's Jump Conditional Code (JCC) Erratum that required a CPU microcode update to mitigate but that comes with a performance hit. At least Intel has pending GNU Assembler patches to help offset that performance hit. In time for last week's articles I didn't have a chance to perform Skylake Xeon Scalable (1st Gen) benchmarks but now here are some metrics alongside Cascade Lake.

Last week's JCC Erratum performance investigation featured the latest-generation Xeon Scalable "Cascade Lake" processors while today are some results side-by-side with previous-generation Xeon Scalable CPUs. For this testing it was two Xeon Gold 6138 processors, based upon the limited Xeon CPUs I have available for testing.

Tests were done with Intel's Clear Linux for seeing the microcode impact as well as the improvement from their patched GNU Assembler they have been shipping since before this vulnerability was published. Via the Phoronix Test Suite a number of benchmarks were carried out.



The Skylake outcome was similar to Cascade Lake as was expected... The microcode update to address the JCC Erratum does provide a performance hit, but at least for some of those benchmarks the updated assembler does help recover that performance loss. Unfortunately though that assembler work isn't yet mainlined let alone in most Linux distributions outside of Clear Linux.



These tests are looking just at the JCC Erratum issue and not to be confused with the TSX Async Abort (TAA) vulnerability also disclosed last Tuesday.





That's the latest in our benchmarking look at the JCC Erratum among the other performance-related security vulnerabilities disclosed last week.