From e0999e4c56bd20f6524b05a3b1aa876a5788a8b4 Mon Sep 17 00:00:00 2001 From: Nicole Rappe Date: Wed, 22 Jan 2025 03:59:50 -0700 Subject: [PATCH] Update blog/posts/01-22-2025 - Windows Power Profiles Causing Notable CPU Performance Loss.md --- ...ndows Power Profiles Causing Notable CPU Performance Loss.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/blog/posts/01-22-2025 - Windows Power Profiles Causing Notable CPU Performance Loss.md b/blog/posts/01-22-2025 - Windows Power Profiles Causing Notable CPU Performance Loss.md index 87c334d..b6bfb40 100644 --- a/blog/posts/01-22-2025 - Windows Power Profiles Causing Notable CPU Performance Loss.md +++ b/blog/posts/01-22-2025 - Windows Power Profiles Causing Notable CPU Performance Loss.md @@ -20,7 +20,7 @@ So I've been noticing a trend recently regarding something I never really took m The general idea is that Windows devices (Workstations & Servers) have what are called power "**profiles**". These profiles, by default, are set to "**Balanced**". Which in basic terms means that the operating system will artificially limit the CPU speed to below 2.0GHz at all times. This means if the CPU is capable of 4GHz, it will be limited to 2GHz no-matter-what. This is a huge problem since it leaves performance just sitting on the table. ## Observations & Actions Taken -When I learned of the above, I began to audit every Windows-based server and workstation (Physical and Virtual) in my homelab. The virtual machines seemed unaffected by this issue, but I still configured them to "**High Performance** power profiles regardless. However, every single physical host (`VIRT-NODE-01`, `VIRT-NODE-02`, and `LAB-DRAAS-01`), all saw notable performance improvements ranging from 32% to 41%, on average going from 1.75GHz to 2.6GHz on the virtualization hosts, and 1.9GHz to 3.2GHz on the backup server. +When I learned of the above, I began to audit every Windows-based server and workstation (Physical and Virtual) in my homelab. The virtual machines seemed unaffected by this issue, but I still configured them to "**High Performance**" power profiles regardless. However, every single physical host (`VIRT-NODE-01`, `VIRT-NODE-02`, and `LAB-DRAAS-01`), all saw notable performance improvements ranging from 32% to 41%, on average going from 1.75GHz to 2.6GHz on the virtualization hosts, and 1.9GHz to 3.2GHz on the backup server. ## Final Thoughts I am so upset that for years, no, decades, it never occured to me that the power profiles applied to server operating systems. I always just assumed they ran in "**High Performance**" power profiles all the time. I discovered I had non-trivial amounts of performance loss because of this simple checkbox setting in the OS.