** Description changed: [SRU Justification] == Impact == See original description. The goal is to adjust the behavior of the balanced governor for certain Xeon CPUs (those are CPUs rather seen in servers so there should be no impact on laptop users). While the initial request was Jammy/5.15 only we should do this to Kinetic and Lunar as well so the experience is the same regardless of release. == Fix == - This picks one change from v6.3-rc1 which tweaks the intel_idle driver for a specific CPU model. + This picks one change from v6.3-rc1 which tweaks the intel_idle driver for a specific CPU model. For Jammy/5.15 this requires one additional change + which introduces the EPP tweak for Alderlake. == Test case == <TBD one of the artificial workloads to compare before and after> == Regression Potential == The change modifies how quickly CPUs scale up (and probably down) depending on workload. Power usage unlikely is of that much concern in server space. Maybe increased heat would be observed. + For Jammy there is the additional risk of havind a different power usage + on Alderlake which is a Laptop CPU. However that change is part of v5.17 + and thus would already be in effect with current 22.04 HWE. == Original description == [Feature Description] Ubuntu uses powersave governor as the default.While the powersave governor has much lower power, the performance is lower than +25% for several workloads compared to performance governor. Report is published: www.Phoronix.com showing difference of 37% The goal here is to keep mean performance delta of powersave governor from performance governor around 10% to 12% by running wide variety of server workloads. For some bursty workload, this delta can be still large, as ramp up of frequency will still lag with powersave governor irrespective of EPP setting. The performance governor always requests maximum frequency. Based on experiments, EPP of 0x00, 0x10, 0x20, the performance delta for powersave governor is around 12%. But the EPP 0x20 has 18% lower average power. Also experiments are done by raising intel_pstate sysfs min_perf_pct as high as 50%. This didn't bring in any additional improvements compared to just changing EPP. Target Kernel: 6.3 Target Release: 22.04 (5.15 kernel) [HW/SW Information] Sapphire Rapids [Business Justification] Performance Improvements Upstream: Merged 6.3 Commit ID: 60675225ebeecea248035fd3a0efc82ae9038a98
-- You received this bug notification because you are a member of Kernel Packages, which is subscribed to linux in Ubuntu. https://bugs.launchpad.net/bugs/2008519 Title: cpufreq: intel_pstate: Update Balance performance EPP for Sapphire Rapids Status in intel: Fix Released Status in linux package in Ubuntu: In Progress Status in linux source package in Jammy: In Progress Status in linux source package in Kinetic: In Progress Status in linux source package in Lunar: In Progress Bug description: [SRU Justification] == Impact == See original description. The goal is to adjust the behavior of the balanced governor for certain Xeon CPUs (those are CPUs rather seen in servers so there should be no impact on laptop users). While the initial request was Jammy/5.15 only we should do this to Kinetic and Lunar as well so the experience is the same regardless of release. == Fix == This picks one change from v6.3-rc1 which tweaks the intel_idle driver for a specific CPU model. For Jammy/5.15 this requires one additional change which introduces the EPP tweak for Alderlake. == Test case == <TBD one of the artificial workloads to compare before and after> == Regression Potential == The change modifies how quickly CPUs scale up (and probably down) depending on workload. Power usage unlikely is of that much concern in server space. Maybe increased heat would be observed. For Jammy there is the additional risk of havind a different power usage on Alderlake which is a Laptop CPU. However that change is part of v5.17 and thus would already be in effect with current 22.04 HWE. == Original description == [Feature Description] Ubuntu uses powersave governor as the default.While the powersave governor has much lower power, the performance is lower than +25% for several workloads compared to performance governor. Report is published: www.Phoronix.com showing difference of 37% The goal here is to keep mean performance delta of powersave governor from performance governor around 10% to 12% by running wide variety of server workloads. For some bursty workload, this delta can be still large, as ramp up of frequency will still lag with powersave governor irrespective of EPP setting. The performance governor always requests maximum frequency. Based on experiments, EPP of 0x00, 0x10, 0x20, the performance delta for powersave governor is around 12%. But the EPP 0x20 has 18% lower average power. Also experiments are done by raising intel_pstate sysfs min_perf_pct as high as 50%. This didn't bring in any additional improvements compared to just changing EPP. Target Kernel: 6.3 Target Release: 22.04 (5.15 kernel) [HW/SW Information] Sapphire Rapids [Business Justification] Performance Improvements Upstream: Merged 6.3 Commit ID: 60675225ebeecea248035fd3a0efc82ae9038a98 To manage notifications about this bug go to: https://bugs.launchpad.net/intel/+bug/2008519/+subscriptions -- Mailing list: https://launchpad.net/~kernel-packages Post to : kernel-packages@lists.launchpad.net Unsubscribe : https://launchpad.net/~kernel-packages More help : https://help.launchpad.net/ListHelp
