Basically, *you* need to know if: 1. your workload does a LOT of syscalls (kernel entries from userspace) OR
2. your workload does a LOT of VMEXITs (runs on a VM) OR 3. uses Intel SGX (you're on your own in that case, go talk to Intel). Which are directly impacted by the microcode and/or kernel changes. You can also selectively disable KPTI/PTI, L1D flushes (L1TF mitigation), and SSBD mitigation to track down which is not compatible with your workload. That *might* give enough hints on how to change your workload to not suffer as much. https://www.kernel.org/doc/html/latest/admin-guide/l1tf.html and also https://www.kernel.org/doc/html/latest/admin-guide/kernel-parameters.html ( read the documentation for the parameters pti=, spectre_v2=, spec_store_bypass_disable= ) You can disable kernel microcode updating (which will revert to the microcode in your BIOS/UEFI) using the dis_ucode_ldr parameter on the *bootloader* (grub, etc). Instructions are in intel-microcode's README file in /usr/share/doc/intel-microcode You can just install an older version of the intel-microcode package and reboot, if you want to switch to an earlier microcode update **provided that your system firmware (BIOS/UEFI) does not have a newer microcode**. -- You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. https://bugs.launchpad.net/bugs/1789719 Title: severe performance issue after intel-microcode update To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu/+source/intel-microcode/+bug/1789719/+subscriptions -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
