i9-11900k video working, cpu governor goofy, Z590 motherboard Ubuntu 20.04 oem kernel

Question

After solving my graphics issue with help from here: Blank (no signal) display on new ASUS Z590 motherboard

And helping someone else out here: UnclaImed display in ubuntu 21.04 with Rocket Lake

I have an issue with my 11900k "powersave" govenor (cpupower) only allowing a max of 4.7ghz. If I manually change it to "performance" using cpupower-gui I can clock up to 5.3ghz and 5.1 on heavy AVX using s-tui just like my motherboard is set for.

Is there a way to "fix" the powersave governor built into this oem kernel to allow max speed in powersave mode? Because under performance mode it's constantly revved up to over 5ghz and running a little hotter. I've searched all over for microcode to no avail. I am clueless about modifying kernels.

Side note: I will ask in a separate thread: iperf is showing my 2.5gb NIC running only at gigabit speed tested against a known 2.5gb connection (works with other PCs at full speed).

Seem like we are needing some actual linux drivers from Intel, what gives?

Update to add inxi output:

CPU: 8-Core 11th Gen Intel Core i9-11900K (-MT MCP-) 
speed/min/max: 3726/800/5300 MHz Kernel: 5.10.0-1023-oem x86_64 Up: 1h     
05m 
Mem: 1485.8/64079.5 MiB (2.3%) Storage: 11.94 TiB (12.7% used) Procs: 
369 
Shell: bash 5.0.17 inxi: 3.0.38

^^^ That storage output is odd as well, only showing my hardware raid array an not my 3 NVMe drives

UPDATE: I installed kernel 5.12. Initially I thought I couldn't boot into it until I moved one of my monitors cables to my discrete GPU. It boots up on the onboard graphics then that just shows a blinking cursor and the gpu connected monitor takes over for a single monitor desktop. Still the same issue with the CPU governor speeds and still no 2.5G ethernet.

Update 2: per Doug Smythies request Before Grub change:

grep . /sys/devices/system/cpu/intel_pstate/*
/sys/devices/system/cpu/intel_pstate/hwp_dynamic_boost:0
/sys/devices/system/cpu/intel_pstate/max_perf_pct:100
/sys/devices/system/cpu/intel_pstate/min_perf_pct:15
/sys/devices/system/cpu/intel_pstate/no_turbo:0
/sys/devices/system/cpu/intel_pstate/num_pstates:46 
/sys/devices/system/cpu/intel_pstate/status:active
/sys/devices/system/cpu/intel_pstate/turbo_pct:40

After Grub change:

chris@pc-01:~$ grep . /sys/devices/system/cpu/cpu*/cpufreq/scaling_m*
/sys/devices/system/cpu/cpu0/cpufreq/scaling_max_freq:5300000
/sys/devices/system/cpu/cpu0/cpufreq/scaling_min_freq:800000
/sys/devices/system/cpu/cpu10/cpufreq/scaling_max_freq:5300000
/sys/devices/system/cpu/cpu10/cpufreq/scaling_min_freq:800000
/sys/devices/system/cpu/cpu11/cpufreq/scaling_max_freq:5300000
/sys/devices/system/cpu/cpu11/cpufreq/scaling_min_freq:800000
/sys/devices/system/cpu/cpu12/cpufreq/scaling_max_freq:5300000
/sys/devices/system/cpu/cpu12/cpufreq/scaling_min_freq:800000
/sys/devices/system/cpu/cpu13/cpufreq/scaling_max_freq:5300000
/sys/devices/system/cpu/cpu13/cpufreq/scaling_min_freq:800000
/sys/devices/system/cpu/cpu14/cpufreq/scaling_max_freq:5300000
/sys/devices/system/cpu/cpu14/cpufreq/scaling_min_freq:800000
/sys/devices/system/cpu/cpu15/cpufreq/scaling_max_freq:5300000
/sys/devices/system/cpu/cpu15/cpufreq/scaling_min_freq:800000
/sys/devices/system/cpu/cpu1/cpufreq/scaling_max_freq:5300000
/sys/devices/system/cpu/cpu1/cpufreq/scaling_min_freq:800000
/sys/devices/system/cpu/cpu2/cpufreq/scaling_max_freq:5300000
/sys/devices/system/cpu/cpu2/cpufreq/scaling_min_freq:800000
/sys/devices/system/cpu/cpu3/cpufreq/scaling_max_freq:5300000
/sys/devices/system/cpu/cpu3/cpufreq/scaling_min_freq:800000
/sys/devices/system/cpu/cpu4/cpufreq/scaling_max_freq:5300000
/sys/devices/system/cpu/cpu4/cpufreq/scaling_min_freq:800000
/sys/devices/system/cpu/cpu5/cpufreq/scaling_max_freq:5300000
/sys/devices/system/cpu/cpu5/cpufreq/scaling_min_freq:800000
/sys/devices/system/cpu/cpu6/cpufreq/scaling_max_freq:5300000 
/sys/devices/system/cpu/cpu6/cpufreq/scaling_min_freq:800000
/sys/devices/system/cpu/cpu7/cpufreq/scaling_max_freq:5300000
/sys/devices/system/cpu/cpu7/cpufreq/scaling_min_freq:800000
/sys/devices/system/cpu/cpu8/cpufreq/scaling_max_freq:5300000
/sys/devices/system/cpu/cpu8/cpufreq/scaling_min_freq:800000
/sys/devices/system/cpu/cpu9/cpufreq/scaling_max_freq:5300000
/sys/devices/system/cpu/cpu9/cpufreq/scaling_min_freq:800000

Also after grub update:

chris@pc-01:~$ grep . /sys/devices/system/cpu/intel_pstate/*
/sys/devices/system/cpu/intel_pstate/hwp_dynamic_boost:0
/sys/devices/system/cpu/intel_pstate/max_perf_pct:100
/sys/devices/system/cpu/intel_pstate/min_perf_pct:15
/sys/devices/system/cpu/intel_pstate/no_turbo:0
/sys/devices/system/cpu/intel_pstate/num_pstates:46  
/sys/devices/system/cpu/intel_pstate/status:active
/sys/devices/system/cpu/intel_pstate/turbo_pct:40

s-tui no change. I did something with systemctl to force it to boot to performance mode. Disabled ondemand as I recall. Performance mode reports 5000mhz, correct per my bios setting for heavy AVX, powersave still at 4700mhz

Output of cpupower frequency-info

chris@pc-01:~$ sudo cpupower frequency-info
analyzing CPU 0:
driver: intel_pstate
CPUs which run at the same hardware frequency: 0
CPUs which need to have their frequency coordinated by software: 0
maximum transition latency:  Cannot determine or is not supported.
hardware limits: 800 MHz - 5.30 GHz
available cpufreq governors: performance powersave
current policy: frequency should be within 800 MHz and 5.30 GHz.
              The governor "powersave" may decide which speed to use
              within this range.
current CPU frequency: Unable to call hardware
current CPU frequency: 2.20 GHz (asserted by call to kernel)
boost state support:
Supported: yes
Active: yes

Turbostat outputs:

chris@pc-01:~$ sudo turbostat --Summary --quiet --show 
Busy%   Bzy_MHz IRQ PkgTmp  PkgWatt GFXWatt RAMWatt PKG_%
0.37    5176    40647   32  21.14   0.01    0.00    0.00
0.37    5165    40647   33  21.26   0.01    0.00    0.00
0.37    5157    39663   32  21.19   0.01    0.00    0.00
0.36    5163    39501   33  21.27   0.01    0.00    0.00
0.37    5168    41354   32  21.17   0.01    0.00    0.00
0.38    5161    42241   32  21.37   0.01    0.00    0.00
0.39    5167    5869    33  21.51   0.01    0.00    0.00
chris@pc-01:~$ sudo turbostat --Summary --quiet --show
Busy%   Bzy_MHz IRQ PkgTmp  PkgWatt GFXWatt RAMWatt PKG_%
1.07    1393    42633   31  17.42   0.01    0.00    0.00
1.06    1435    43711   32  17.52   0.01    0.00    0.00
1.08    1397    41036   32  17.41   0.01    0.00    0.00
1.08    1389    41864   32  17.40   0.01    0.00    0.00
1.16    1390    19255   33  17.48   0.01    0.00    0.00

So it's costing me more like 4 watts just idling. Not a huge deal but I'd still like to fix it.

Also my turbo_pct should be roughly 51 to be correct. Does not appear to be a way to change it that I can find. Perhaps the kernels are seeing my CPU as a non K version? Like 11900 instead of 11900k yet when I check things like hwinfo it shows 11900k

Powersave turbostat no quiet:

chris@pc-01:~$ sudo turbostat --Summary --show 
Busy%,Bzy_MHz,IRQ,PkgWatt,PkgTmp,RAMWatt,GFXWatt,PKG_% 
turbostat version 20.09.30 - Len Brown <lenb@kernel.org>
CPUID(0): GenuineIntel 0x1b CPUID levels; 0x80000008 xlevels;  
family:model:stepping 0x6:a7:1 (6:167:1)
CPUID(1): SSE3 MONITOR SMX EIST TM2 TSC MSR ACPI-TM HT TM
CPUID(6): APERF, TURBO, DTS, PTM, HWP, HWPnotify, HWPwindow, HWPepp, 
HWPpkg, EPB
cpu4: MSR_IA32_MISC_ENABLE: 0x00850089 (TCC EIST MWAIT PREFETCH TURBO)
CPUID(7): No-SGX
CPUID(0x15): eax_crystal: 2 ebx_tsc: 292 ecx_crystal_hz: 24000000
TSC: 3504 MHz (24000000 Hz * 292 / 2 / 1000000)
CPUID(0x16): base_mhz: 3500 max_mhz: 5300 bus_mhz: 100
cpu4: MSR_MISC_PWR_MGMT: 0x00401cc0 (ENable-EIST_Coordination DISable-
EPB DISable-OOB)
RAPL: 2097 sec. Joule Counter Range, at 125 Watts
cpu4: MSR_PLATFORM_INFO: 0x808083af1812300
8 * 100.0 = 800.0 MHz max efficiency frequency
35 * 100.0 = 3500.0 MHz base frequency
cpu4: MSR_IA32_POWER_CTL: 0x0034005d (C1E auto-promotion: DISabled)
cpu4: MSR_TURBO_RATIO_LIMIT: 0x3232323434353535
50 * 100.0 = 5000.0 MHz max turbo 8 active cores
50 * 100.0 = 5000.0 MHz max turbo 7 active cores
50 * 100.0 = 5000.0 MHz max turbo 6 active cores
52 * 100.0 = 5200.0 MHz max turbo 5 active cores
52 * 100.0 = 5200.0 MHz max turbo 4 active cores
53 * 100.0 = 5300.0 MHz max turbo 3 active cores
53 * 100.0 = 5300.0 MHz max turbo 2 active cores
53 * 100.0 = 5300.0 MHz max turbo 1 active cores
cpu4: MSR_CONFIG_TDP_NOMINAL: 0x00000023 (base_ratio=35)
cpu4: MSR_CONFIG_TDP_LEVEL_1: 0x001e02f8 (PKG_MIN_PWR_LVL1=0 
PKG_MAX_PWR_LVL1=0 LVL1_RATIO=30 PKG_TDP_LVL1=760)
cpu4: MSR_CONFIG_TDP_LEVEL_2: 0x00000000 ()
cpu4: MSR_CONFIG_TDP_CONTROL: 0x00000000 ( lock=0)
cpu4: MSR_TURBO_ACTIVATION_RATIO: 0x00000000 (MAX_NON_TURBO_RATIO=0 
lock=0)
cpu4: MSR_PKG_CST_CONFIG_CONTROL: 0x74000000 (UNdemote-C1, demote-C1, 
UNlocked, pkg-cstate-limit=0 (pc0))
/dev/cpu_dma_latency: 2000000000 usec (default)
current_driver: intel_idle
current_governor: menu
current_governor_ro: menu
cpu4: POLL: CPUIDLE CORE POLL IDLE
cpu4: C1ACPI: ACPI FFH MWAIT 0x0
cpu4: C2ACPI: ACPI FFH MWAIT 0x31
cpu4: C3ACPI: ACPI FFH MWAIT 0x60
cpu4: cpufreq driver: intel_pstate
cpu4: cpufreq governor: powersave
cpufreq intel_pstate no_turbo: 0
cpu4: MSR_MISC_FEATURE_CONTROL: 0x00000000 (L2-Prefetch L2-Prefetch-
pair L1-Prefetch L1-IP-Prefetch)
cpu0: MSR_PM_ENABLE: 0x00000001 (HWP)
cpu0: MSR_HWP_CAPABILITIES: 0x010d2335 (high 53 guar 35 eff 13 low 1)
cpu0: MSR_HWP_REQUEST: 0x80003508 (min 8 max 53 des 0 epp 0x80 window 
0x0 
pkg 0x0)
cpu0: MSR_HWP_REQUEST_PKG: 0x8000ff01 (min 1 max 255 des 0 epp 0x80 
window 
0x0)
cpu0: MSR_HWP_INTERRUPT: 0x00000000 (Dis_Guaranteed_Perf_Change, 
Dis_Excursion_Min)
cpu0: MSR_HWP_STATUS: 0x00000004 (No-Guaranteed_Perf_Change, No-
Excursion_Min)
cpu0: MSR_IA32_ENERGY_PERF_BIAS: 0x00000006 (balanced)
cpu0: MSR_RAPL_POWER_UNIT: 0x000a0e03 (0.125000 Watts, 0.000061 
Joules,   
0.000977 sec.)
cpu0: MSR_PKG_POWER_INFO: 0x000003e8 (125 W TDP, RAPL 0 - 0 W, 
0.000000 
sec.)
cpu0: MSR_PKG_POWER_LIMIT: 0x42fff800e5fff8 (UNlocked)
cpu0: PKG Limit #1: ENabled (4095.000000 Watts, 448.000000 sec, clamp 
ENabled)
cpu0: PKG Limit #2: ENabled (4095.000000 Watts, 0.002441* sec, clamp 
DISabled)
cpu0: MSR_DRAM_POWER_LIMIT: 0x5400de00000000 (UNlocked)
cpu0: DRAM Limit: DISabled (0.000000 Watts, 0.000977 sec, clamp 
DISabled)
cpu0: MSR_PP0_POLICY: 0
cpu0: MSR_PP0_POWER_LIMIT: 0x00000000 (UNlocked)
cpu0: Cores Limit: DISabled (0.000000 Watts, 0.000977 sec, clamp 
DISabled)
cpu0: MSR_PP1_POLICY: 0
cpu0: MSR_PP1_POWER_LIMIT: 0x00000000 (UNlocked)
cpu0: GFX Limit: DISabled (0.000000 Watts, 0.000977 sec, clamp 
DISabled)
cpu4: MSR_IA32_TEMPERATURE_TARGET: 0x00640000 (100 C) (100 default - 0   
offset)
cpu0: MSR_IA32_PACKAGE_THERM_STATUS: 0x88450000 (31 C)
cpu0: MSR_IA32_PACKAGE_THERM_INTERRUPT: 0x00000003 (100 C, 100 C)
cpu4: MSR_PKGC3_IRTL: 0x00000000 (NOTvalid, 0 ns)
cpu4: MSR_PKGC6_IRTL: 0x00000000 (NOTvalid, 0 ns)
cpu4: MSR_PKGC7_IRTL: 0x00000000 (NOTvalid, 0 ns)
cpu4: MSR_PKGC8_IRTL: 0x00000000 (NOTvalid, 0 ns)
cpu4: MSR_PKGC9_IRTL: 0x00000000 (NOTvalid, 0 ns)
cpu4: MSR_PKGC10_IRTL: 0x00000000 (NOTvalid, 0 ns)
Busy%   Bzy_MHz IRQ PkgTmp  PkgWatt GFXWatt RAMWatt PKG_%
0.76    1268    6511    30  16.87   0.00    0.00    0.00
0.89    983 9705    31  17.39   0.03    0.00    0.00
0.43    851 7473    31  17.03   0.00    0.00    0.00
0.36    800 5721    30  16.85   0.00    0.00    0.00
0.33    800 5427    30  16.80   0.00    0.00    0.00
0.34    800 5414    31  16.77   0.00    0.00    0.00
0.46    800 5744    30  16.61   0.01    0.00    0.00
0.48    800 2881    31  16.68   0.00    0.00    0.00

performance turbostat no quiet

chris@pc-01:~$ sudo turbostat --Summary --show 
Busy%,Bzy_MHz,IRQ,PkgWatt,PkgTmp,RAMWatt,GFXWatt,PKG_% 
turbostat version 20.09.30 - Len Brown <lenb@kernel.org>
CPUID(0): GenuineIntel 0x1b CPUID levels; 0x80000008 xlevels; 
family:model:stepping 0x6:a7:1 (6:167:1)
CPUID(1): SSE3 MONITOR SMX EIST TM2 TSC MSR ACPI-TM HT TM
CPUID(6): APERF, TURBO, DTS, PTM, HWP, HWPnotify, HWPwindow, HWPepp, 
HWPpkg, EPB
cpu4: MSR_IA32_MISC_ENABLE: 0x00850089 (TCC EIST MWAIT PREFETCH TURBO)
CPUID(7): No-SGX
CPUID(0x15): eax_crystal: 2 ebx_tsc: 292 ecx_crystal_hz: 24000000
TSC: 3504 MHz (24000000 Hz * 292 / 2 / 1000000)
CPUID(0x16): base_mhz: 3500 max_mhz: 5300 bus_mhz: 100
cpu4: MSR_MISC_PWR_MGMT: 0x00401cc0 (ENable-EIST_Coordination DISable-
EPB DISable-OOB)
RAPL: 2097 sec. Joule Counter Range, at 125 Watts
cpu4: MSR_PLATFORM_INFO: 0x808083af1812300
8 * 100.0 = 800.0 MHz max efficiency frequency
35 * 100.0 = 3500.0 MHz base frequency
cpu4: MSR_IA32_POWER_CTL: 0x0034005d (C1E auto-promotion: DISabled)
cpu4: MSR_TURBO_RATIO_LIMIT: 0x3232323434353535
50 * 100.0 = 5000.0 MHz max turbo 8 active cores
50 * 100.0 = 5000.0 MHz max turbo 7 active cores
50 * 100.0 = 5000.0 MHz max turbo 6 active cores
52 * 100.0 = 5200.0 MHz max turbo 5 active cores
52 * 100.0 = 5200.0 MHz max turbo 4 active cores
53 * 100.0 = 5300.0 MHz max turbo 3 active cores
53 * 100.0 = 5300.0 MHz max turbo 2 active cores
53 * 100.0 = 5300.0 MHz max turbo 1 active cores
cpu4: MSR_CONFIG_TDP_NOMINAL: 0x00000023 (base_ratio=35)
cpu4: MSR_CONFIG_TDP_LEVEL_1: 0x001e02f8 (PKG_MIN_PWR_LVL1=0 
PKG_MAX_PWR_LVL1=0 LVL1_RATIO=30 PKG_TDP_LVL1=760)
cpu4: MSR_CONFIG_TDP_LEVEL_2: 0x00000000 ()
cpu4: MSR_CONFIG_TDP_CONTROL: 0x00000000 ( lock=0)
cpu4: MSR_TURBO_ACTIVATION_RATIO: 0x00000000 (MAX_NON_TURBO_RATIO=0 
lock=0)
cpu4: MSR_PKG_CST_CONFIG_CONTROL: 0x74000000 (UNdemote-C1, demote-C1, 
UNlocked, pkg-cstate-limit=0 (pc0))
/dev/cpu_dma_latency: 2000000000 usec (default)
current_driver: intel_idle
current_governor: menu
current_governor_ro: menu
cpu4: POLL: CPUIDLE CORE POLL IDLE
cpu4: C1ACPI: ACPI FFH MWAIT 0x0
cpu4: C2ACPI: ACPI FFH MWAIT 0x31
cpu4: C3ACPI: ACPI FFH MWAIT 0x60
cpu4: cpufreq driver: intel_pstate
cpu4: cpufreq governor: performance
cpufreq intel_pstate no_turbo: 0
cpu4: MSR_MISC_FEATURE_CONTROL: 0x00000000 (L2-Prefetch L2-Prefetch-
pair L1-Prefetch L1-IP-Prefetch)
cpu0: MSR_PM_ENABLE: 0x00000001 (HWP)
cpu0: MSR_HWP_CAPABILITIES: 0x010d2335 (high 53 guar 35 eff 13 low 1)
cpu0: MSR_HWP_REQUEST: 0x00003535 (min 53 max 53 des 0 epp 0x0 window 
0x0 pkg 0x0)
cpu0: MSR_HWP_REQUEST_PKG: 0x8000ff01 (min 1 max 255 des 0 epp 0x80 
window 0x0)
cpu0: MSR_HWP_INTERRUPT: 0x00000000 (Dis_Guaranteed_Perf_Change, 
Dis_Excursion_Min)
cpu0: MSR_HWP_STATUS: 0x00000004 (No-Guaranteed_Perf_Change, No-
Excursion_Min)
cpu0: MSR_IA32_ENERGY_PERF_BIAS: 0x00000006 (balanced)
cpu0: MSR_RAPL_POWER_UNIT: 0x000a0e03 (0.125000 Watts, 0.000061 
Joules, 0.000977 sec.)
cpu0: MSR_PKG_POWER_INFO: 0x000003e8 (125 W TDP, RAPL 0 - 0 W, 
0.000000 sec.)
cpu0: MSR_PKG_POWER_LIMIT: 0x42fff800e5fff8 (UNlocked)
cpu0: PKG Limit #1: ENabled (4095.000000 Watts, 448.000000 sec, clamp 
ENabled)
cpu0: PKG Limit #2: ENabled (4095.000000 Watts, 0.002441* sec, clamp 
DISabled)
cpu0: MSR_DRAM_POWER_LIMIT: 0x5400de00000000 (UNlocked)
cpu0: DRAM Limit: DISabled (0.000000 Watts, 0.000977 sec, clamp 
DISabled)
cpu0: MSR_PP0_POLICY: 0
cpu0: MSR_PP0_POWER_LIMIT: 0x00000000 (UNlocked)
cpu0: Cores Limit: DISabled (0.000000 Watts, 0.000977 sec, clamp 
DISabled)
cpu0: MSR_PP1_POLICY: 0
cpu0: MSR_PP1_POWER_LIMIT: 0x00000000 (UNlocked)
cpu0: GFX Limit: DISabled (0.000000 Watts, 0.000977 sec, clamp 
DISabled)
cpu4: MSR_IA32_TEMPERATURE_TARGET: 0x00640000 (100 C) (100 default - 0 
offset)
cpu0: MSR_IA32_PACKAGE_THERM_STATUS: 0x88410000 (35 C)
cpu0: MSR_IA32_PACKAGE_THERM_INTERRUPT: 0x00000003 (100 C, 100 C)
cpu4: MSR_PKGC3_IRTL: 0x00000000 (NOTvalid, 0 ns)
cpu4: MSR_PKGC6_IRTL: 0x00000000 (NOTvalid, 0 ns)
cpu4: MSR_PKGC7_IRTL: 0x00000000 (NOTvalid, 0 ns)
cpu4: MSR_PKGC8_IRTL: 0x00000000 (NOTvalid, 0 ns)
cpu4: MSR_PKGC9_IRTL: 0x00000000 (NOTvalid, 0 ns)
cpu4: MSR_PKGC10_IRTL: 0x00000000 (NOTvalid, 0 ns)
Busy%   Bzy_MHz IRQ PkgTmp  PkgWatt GFXWatt RAMWatt PKG_%
0.10    5295    2016    32  19.00   0.00    0.00    0.00
0.10    5297    2001    31  19.00   0.00    0.00    0.00
0.09    5298    1944    31  18.93   0.00    0.00    0.00
0.09    5297    1913    31  18.95   0.00    0.00    0.00
0.09    5298    1907    32  18.88   0.00    0.00    0.00
0.10    5287    1778    31  19.18   0.00    0.00    0.00

end

Show us more information when using the powersave governor, like what the limits are set to and what test you did that makes you think it should have ramped up to max CPU frequency. Provide: grep . /sys/devices/system/cpu/intel_pstate/* and grep . /sys/devices/system/cpu/cpu*/cpufreq/scaling_m*. Just as a test try disabling HWP, intel_pstate=no_hwp on grub command line in addition to what is already there, GRUB_CMDLINE_LINUX_DEFAULT — Doug Smythies, Apr 27 '21 at 20:32
I updated the post with the info you asked for Doug Smythies — rsr911, Apr 28 '21 at 01:26
I have Conky running in it's own window that constantly monitors my system. It's reporting current frequency. When in performance mode it's running over 5000 all the time. In powersave mode it will idle down. — rsr911, Apr 28 '21 at 01:31
The workload used by s-tui, which I gather is via "stress", isn't clear to me. Do a simple test: taskset -c 2 yes >/dev/null to put a 100% load on CPU 2, and what CPU frequency do you get? I only do primitives, so do it this way: grep . /sys/devices/system/cpu/cpu*/cpufreq/scaling_cur_freq. Hmmm... the turbo percent number doesn't seem correct. — Doug Smythies, Apr 28 '21 at 02:49
@DougSmythies when i do that all the cores come up in speed but CPU2 is at 4700mhz
/sys/devices/system/cpu/cpu10/cpufreq/scaling_cur_freq:4700000 — rsr911, Apr 28 '21 at 03:05
BTW I know how to put code in answers and initial posts but I can't figure out how to do it in these comments. Also the output of grep . /sys/devices/system/cpu/cpu*/cpufreq/scaling_cur_freq is too large for a comment. But all CPUS were around 4.5-4.6ghz with just the one pegged at 4.7 flat — rsr911, Apr 28 '21 at 03:11

Doug Smythies · Answer 1 · 2021-04-28T14:25:11.870

Note that running using the performance governor doesn't typically cause a lot of extra processor package heat dissipation. Yes, the CPUs run at higher frequencies, and the energy used is highly non-linear in the turbo region, but they also finish their work faster and therefore spend more time in deep idle states, consuming almost no energy. Examples, using a i5-10600K:

Powersave (HWP disabled):

doug@s19:~$ sudo turbostat --Summary --quiet --show Busy%,Bzy_MHz,IRQ,PkgWatt,PkgTmp,RAMWatt,GFXWatt,PKG_% --interval 15
Busy%   Bzy_MHz IRQ     PkgTmp  PkgWatt GFXWatt RAMWatt PKG_%
0.08    1447    889     36      1.51    0.00    0.89    0.00
0.12    865     779     36      1.42    0.00    0.89    0.00
0.13    800     733     36      1.45    0.00    0.89    0.00
0.12    805     759     36      1.39    0.00    0.89    0.00

And Performance (HWP disabled):

doug@s19:~$ sudo turbostat --Summary --quiet --show Busy%,Bzy_MHz,IRQ,PkgWatt,PkgTmp,RAMWatt,GFXWatt,PKG_% --interval 15
Busy%   Bzy_MHz IRQ     PkgTmp  PkgWatt GFXWatt RAMWatt PKG_%
0.03    4100    792     34      1.48    0.00    0.89    0.00
0.03    4100    658     34      1.52    0.00    0.89    0.00
0.03    4100    656     35      1.51    0.00    0.89    0.00
0.03    4100    749     35      1.62    0.00    0.89    0.00

Observe the cost was about 0.1 watts for an idle system.

And for not so idle (but still fairly idle), similar interrupts / sample period:

Powersave (HWP disabled):

doug@s19:~$ sudo turbostat --Summary --quiet --show Busy%,Bzy_MHz,IRQ,PkgWatt,PkgTmp,RAMWatt,GFXWatt,PKG_%
Busy%   Bzy_MHz IRQ     PkgTmp  PkgWatt GFXWatt RAMWatt PKG_%
1.85    937     46458   33      1.90    0.00    0.89    0.00
2.02    815     46394   33      1.73    0.00    0.89    0.00
2.18    800     46647   33      1.74    0.00    0.89    0.00
2.23    800     46809   33      1.87    0.00    0.89    0.00
2.42    800     47187   33      1.79    0.00    0.89    0.00
2.60    800     48461   33      1.79    0.00    0.89    0.00

Performance (HWP disabled):

doug@s19:~$ sudo turbostat --Summary --quiet --show Busy%,Bzy_MHz,IRQ,PkgWatt,PkgTmp,RAMWatt,GFXWatt,PKG_%
Busy%   Bzy_MHz IRQ     PkgTmp  PkgWatt GFXWatt RAMWatt PKG_%
0.72    2404    46360   34      13.98   0.00    0.89    0.00
0.74    2478    46804   35      14.16   0.00    0.89    0.00
0.71    2752    47188   35      14.88   0.00    0.89    0.00
0.87    2458    47454   35      14.12   0.00    0.89    0.00
0.90    2511    48373   35      14.93   0.00    0.89    0.00

So, the cost is ~13 watts, a lot.

Powersave (HWP enabled):

doug@s19:~$ sudo turbostat --Summary --quiet --show Busy%,Bzy_MHz,IRQ,PkgWatt,PkgTmp,RAMWatt,GFXWatt,PKG_%
Busy%   Bzy_MHz IRQ     PkgTmp  PkgWatt GFXWatt RAMWatt PKG_%
1.13    1604    40861   37      10.55   0.00    0.89    0.00
1.13    1592    40709   37      10.56   0.00    0.89    0.00
1.14    1569    40826   37      10.56   0.00    0.89    0.00
1.15    1547    40959   37      10.50   0.00    0.89    0.00
1.16    1555    40936   37      10.32   0.00    0.89    0.00
1.00    1733    41003   37      11.39   0.00    0.89    0.00

Performance (HWP enabled):

doug@s19:~$ sudo turbostat --Summary --quiet --show Busy%,Bzy_MHz,IRQ,PkgWatt,PkgTmp,RAMWatt,GFXWatt,PKG_%
Busy%   Bzy_MHz IRQ     PkgTmp  PkgWatt GFXWatt RAMWatt PKG_%
0.57    3484    43276   37      15.18   0.00    0.89    0.00
0.62    2749    44102   37      14.99   0.00    0.89    0.00
0.65    2493    44177   38      15.24   0.00    0.89    0.00
0.65    2476    44166   37      15.07   0.00    0.89    0.00
0.65    2474    44163   37      15.14   0.00    0.89    0.00

About 4.5 watts.

EDITED: Now, your intel_pstate parameters are interesting, but my understanding of the definition of the numbers was incorrect:

From the documentation:

Sysfs will show :
    max_perf_pct:100, which corresponds to 1 core ratio
    min_perf_pct:24, max_efficiency_ratio / max 1 Core ratio
    no_turbo:0, turbo is not disabled
    num_pstates:26 = (max 1 Core ratio - Max Efficiency Ratio + 1)
    turbo_pct:39 = (max 1 core ratio - max non turbo ratio) / num_pstates

From the code:

static ssize_t show_turbo_pct(struct kobject *kobj,
                                struct kobj_attribute *attr, char *buf)
{
        struct cpudata *cpu;
        int total, no_turbo, turbo_pct;
        uint32_t turbo_fp;
    mutex_lock(&amp;intel_pstate_driver_lock);

    if (!intel_pstate_driver) {
            mutex_unlock(&amp;intel_pstate_driver_lock);
            return -EAGAIN;
    }

    cpu = all_cpu_data[0];

    total = cpu-&gt;pstate.turbo_pstate - cpu-&gt;pstate.min_pstate + 1;
    no_turbo = cpu-&gt;pstate.max_pstate - cpu-&gt;pstate.min_pstate + 1;
    turbo_fp = div_fp(no_turbo, total);
    turbo_pct = 100 - fp_toint(mul_fp(turbo_fp, int_tofp(100)));

    mutex_unlock(&amp;intel_pstate_driver_lock);

    return sprintf(buf, &quot;%u\n&quot;, turbo_pct);

}

So, for your processor, the specs are 3.5 GHz base frequency, 0.8 GHz min, 5.3 GHz max.

(53 - 35) / 46 = 39% for your case. Close enough, and doing it the exact code way, gives 40%.

EDIT 2: Please see this article. Most relevant extract:

What this means is that, if all 8 cores are loaded, TB2 means that it will run at 4.7 GHz. If there is power budget and thermal budget, it will attempt 4.8 GHz. If there is more power budget and thermal budget available, it will go to 4.9 GHz, then 5.0 GHz, then 5.1 GHz. The frequency will float as long as it has enough of those budgets to play with, and it will increase/decrease as necessary. This is important as different instructions cause different amounts of power draw and such.

Yep, I just caught that as well reading your prior comment and went in search of changing it. (turbo_pct) no luck so far. — rsr911, Apr 28 '21 at 03:36
I edited the question to add my turbostat outputs. Costing me 4 watts roughly. — rsr911, Apr 28 '21 at 03:45
@rsr911 : I volunteer in the linux kernel power management group, intel_pstate and idle. I'll look into this issue upstream and in the e-mail archives. It might take days, but I'll get back to you here. I might want to read some MSRs along the way. No, you can not change the turbo %, it is read only. Your processor consumes a lot of power for the % work it is doing, and a lot of interrupts. — Doug Smythies, Apr 28 '21 at 04:00
Thank you very much for all the help. I will look forward to what you find. You are correct, it is very power hungry. I can clock it to 5.4ghz (5.1 under heavy AVX) and running Handbrake even my 280mm radiator with Noctua fans can't keep it cool. It can pull over 300 watts like that. AT 5.3/5.0 like I have it set now I can keep it under control. While I see what you can come up with I'll go in search of fixing the 2.5gb ethernet issue and dig more into the onboard graphics issue. — rsr911, Apr 28 '21 at 04:08
@rsr911 : I made a mistake in my interpretation of turbo %. Yours is O.K. — Doug Smythies, Apr 28 '21 at 05:38
Interesting. But how does the powersave governor interact (seemingly) with ABT while the performance governor runs as high as I let it overclock? I'll need to read the article more later (at work now), perhaps turning off ABT in my bios is a worthwhile experiment? — rsr911, Apr 28 '21 at 16:00
well I disabled ABT in the bios, no change. Played around with TVB settings, no change. Clocked it up to 5.4, still no change. The powersave governor seems to ignore any bios settings I make. When I clocked up to 5.4 cpupower showed 800 - 5400 mhz so at least that is being reported correctly. If I switch to performance governor it will run at my max bios settings but also will not throttle down. As for thermals and power I'm running a 280mm AIO which cools fairly well. PSU is 750 watts. CPU can draw over 275 watts in performance mode. Seems like this is the governor. — rsr911, Apr 28 '21 at 21:58
With a nominal TDP of 125 watts, myself I would not allow it to go to 275 watts, and am surprised you are not hitting electrical current limit throttling. My nominal TDP is 125 watts and I have only let power limit 2 go to 136 watts, and left power limit 1 as is at 125 watts. (Use turbostat without the --quiet option to see details). Not sure why you don't get higher CPU frequency with powersave when only 1 CPU loaded. — Doug Smythies, Apr 28 '21 at 22:41
Posted the no quiet turbostat readings. I did some testing using Handbrake because it seems to put the highest load on the CPU. 275w at 5000mhz, 85C using performance. Powersave gave me 185w at under 4700mhz and 65C, was clearly thottling to stay below 70C it seemed. The max t-junction temp for this CPU is 105C. Handbrake gets the CPU hotter than anything else I run.
My motherboard is more than capable of supplying the power I need. Asus Rog Z590-e Gaming. I'm not concerned with watts when I am really using it as it's performance that matters to me. I could tweak the power to cut back the W — rsr911, Apr 29 '21 at 00:07
"(53 - 35) / 46 = 39% for your case. Close enough, and doing it the exact code way, gives 40%." Wouldn't the 46 be a 45? 5300-800=4500. (53-35)/46 = 40%
It seems like the powersave governor has some sort of thermal/power throttling that the performance governor does not. That's about all that makes sense to me. Especially when I run Handbrake which put a heavy real world load on the CPU, it appears to throttle to stay below 70C and 185W, does not hit the 4.7ghz I see running s-tui (stress). In performance it does not throttle, running right at the 5ghz it's set for for AVX. — rsr911, Apr 29 '21 at 01:19
No, it is inclusive so 53 - 8 + 1 = 46 pstates. I am not familiar with "Handbrake". I use prime95 - torture test - maximum heat test 2 as the most brutal load I have ever found. Oh, I see you do not have power limits set. — Doug Smythies, Apr 29 '21 at 03:10
Nope no power limits. Auto voltages though which could be tweaked down for less power drawn only running thermal limits. I'm not sure we've solved the "problem" lol but I'm going to give your answer an acceptance and upvote. With cpupower-gui I can change the governor in seconds and that's good enough for me. — rsr911, Apr 29 '21 at 03:34

score 1 · Accepted Answer · answered May 01 '21 at 03:01

Turns out this was a bios issue with the CPU frequency. I was running an updated version (0704), updated to 0707 that was released 4/28/2021 and now my CPU will run at full speed under load.

Here's a link to the latest bios from Asus: https://rog.asus.com/us/motherboards/rog-strix/rog-strix-z590-e-gaming-wifi-model/helpdesk_bios

Now I will check my networking and see if anything changed there. Still will not boot into the GUI on mainline kernels or stock kernels without an add in graphics card.

Networking checked with iperf. Still running at 1gb instead of 2.5

i9-11900k video working, cpu governor goofy, Z590 motherboard Ubuntu 20.04 oem kernel

2 Answers2