Results 1 to 5 of 5

Thread: Wrong Frequency detected (GHz)?

  1. #1
    Join Date
    Oct 2014
    Posts
    8

    Default Wrong Frequency detected (GHz)?

    (I mentioned it in another thread, so here a more detailed report)

    $ phoronix-test-suite system-info

    Phoronix Test Suite v5.2.1
    System Information

    Hardware:
    Processor: Intel Core i7-3930K @ 5.70GHz (12 Cores), Motherboard: ASUS P9X79 PRO, Chipset: Intel Xeon E5/Core, Memory: 32768MB, Disk: 240GB Patriot Pyro SE + 128GB Patriot Torqx 12, Graphics: NVIDIA GeForce GTX 680 2048MB (1084/3004MHz), Audio: Intel C600/X79, Monitor: DELL UP2414Q + Acer B243HL, Network: Intel 82579V Gigabit Connection

    Software:
    OS: Fedora 20, Kernel: 3.16.3-200.fc20.x86_64 (x86_64), Desktop: KDE 4.14.1, Display Server: X Server 1.14.4, Display Driver: NVIDIA 343.22, OpenGL: 4.3.0, Compiler: CUDA 5.0, File-System: ext4, Screen Resolution: 5760x2160

    The detection here is wrong. My CPU is actually not running on 5.7 GHz.

    Here an excerpt from my /proc/cpuinfo:

    ---------------------------- /proc/cpuinfo ----------------------------
    processor : 11
    vendor_id : GenuineIntel
    cpu family : 6
    model : 45
    model name : Intel(R) Core(TM) i7-3930K CPU @ 3.20GHz
    stepping : 7
    microcode : 0x710
    cpu MHz : 3300.125
    cache size : 12288 KB
    physical id : 0
    siblings : 12
    core id : 5
    cpu cores : 6
    apicid : 11
    initial apicid : 11
    fpu : yes
    fpu_exception : yes
    cpuid level : 13
    wp : yes
    flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc aperfmperf eagerfpu pni pclmulqdq dtes64 monitor ds_cpl vmx est tm2 ssse3 cx16 xtpr pdcm pcid dca sse4_1 sse4_2 x2apic popcnt tsc_deadline_timer aes xsave avx lahf_lm ida arat epb xsaveopt pln pts dtherm tpr_shadow vnmi flexpriority ept vpid
    bogomips : 8088.49
    clflush size : 64
    cache_alignment : 64
    address sizes : 46 bits physical, 48 bits virtual
    power management:
    --------------------------------------------------------

    It seems with TURBO et al it is actually faster, but far from 5.2 GHz, here the information from i7z:

    ------------------------ i7z ---------------------------------------------------
    Cpu speed from cpuinfo 4044.00Mhz
    cpuinfo might be wrong if cpufreq is enabled. To guess correctly try estimating via tsc
    Linux's inbuilt cpu_khz code emulated now
    True Frequency (without accounting Turbo) 4043 MHz
    CPU Multiplier 32x || Bus clock frequency (BCLK) 126.34 MHz

    Socket [0] - [physical cores=6, logical cores=12, max online cores ever=6]
    TURBO ENABLED on 6 Cores, Hyper Threading ON
    Max Frequency without considering Turbo 4169.34 MHz (126.34 x [33])
    Max TURBO Multiplier (if Enabled) with 1/2/3/4/5/6 Cores is 57x/57x/57x/57x/57x/57x
    Real Current Frequency 4199.52 MHz [126.34 x 33.24] (Max of below)
    Core [core-id] :Actual Freq (Mult.) C0% Halt(C1)% C3 % C6 % C7 % Temp
    Core 1 [0]: 4167.67 (32.99x) 1.3 1.47 1 0 96.2 33
    Core 2 [1]: 4168.81 (33.00x) 2.69 3.93 1 0 92.3 29
    Core 3 [2]: 4167.62 (32.99x) 1.48 1.29 0 0 97.2 32
    Core 4 [3]: 4199.52 (33.24x) 1 100 0 0 0 47
    Core 5 [4]: 4169.16 (33.00x) 8.87 3.09 0 0 87.8 31
    Core 6 [5]: 4168.21 (32.99x) 1.92 4.83 1 0 92.2 32
    34
    C0 = Processor running without halting
    C1 = Processor running with halts (States >C0 are power saver)
    C3 = Cores running with PLL turned off and core cache turned off
    C6 = Everything in C3 + core state saved to last level cache
    Above values in table are in percentage over the last 1 sec
    [core-id] refers to core-id number in /proc/cpuinfo
    ---------------------------------------------------------------

    A short glance at the /sys/bus/cpu/devices/cpu0/cpufreq/* files indicate that the reason must be
    /sys/bus/cpu/devices/cpu0/cpufreq/scaling_max_freq and cpuinfo_max_freq which contains the value 5.7 GHz (I use intel_pstate/powersave governor).

    Seriously I dont know, why there are those 5.7 GHz in the max_freq-files, but under no load ever I have saw him reaching that.

  2. #2

    Default

    Quote Originally Posted by Flolo View Post
    (I mentioned it in another thread, so here a more detailed report)

    $ phoronix-test-suite system-info

    Phoronix Test Suite v5.2.1
    System Information

    Hardware:
    Processor: Intel Core i7-3930K @ 5.70GHz (12 Cores), Motherboard: ASUS P9X79 PRO, Chipset: Intel Xeon E5/Core, Memory: 32768MB, Disk: 240GB Patriot Pyro SE + 128GB Patriot Torqx 12, Graphics: NVIDIA GeForce GTX 680 2048MB (1084/3004MHz), Audio: Intel C600/X79, Monitor: DELL UP2414Q + Acer B243HL, Network: Intel 82579V Gigabit Connection

    Software:
    OS: Fedora 20, Kernel: 3.16.3-200.fc20.x86_64 (x86_64), Desktop: KDE 4.14.1, Display Server: X Server 1.14.4, Display Driver: NVIDIA 343.22, OpenGL: 4.3.0, Compiler: CUDA 5.0, File-System: ext4, Screen Resolution: 5760x2160

    The detection here is wrong. My CPU is actually not running on 5.7 GHz.

    Here an excerpt from my /proc/cpuinfo:

    ---------------------------- /proc/cpuinfo ----------------------------
    processor : 11
    vendor_id : GenuineIntel
    cpu family : 6
    model : 45
    model name : Intel(R) Core(TM) i7-3930K CPU @ 3.20GHz
    stepping : 7
    microcode : 0x710
    cpu MHz : 3300.125
    cache size : 12288 KB
    physical id : 0
    siblings : 12
    core id : 5
    cpu cores : 6
    apicid : 11
    initial apicid : 11
    fpu : yes
    fpu_exception : yes
    cpuid level : 13
    wp : yes
    flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc aperfmperf eagerfpu pni pclmulqdq dtes64 monitor ds_cpl vmx est tm2 ssse3 cx16 xtpr pdcm pcid dca sse4_1 sse4_2 x2apic popcnt tsc_deadline_timer aes xsave avx lahf_lm ida arat epb xsaveopt pln pts dtherm tpr_shadow vnmi flexpriority ept vpid
    bogomips : 8088.49
    clflush size : 64
    cache_alignment : 64
    address sizes : 46 bits physical, 48 bits virtual
    power management:
    --------------------------------------------------------

    It seems with TURBO et al it is actually faster, but far from 5.2 GHz, here the information from i7z:

    ------------------------ i7z ---------------------------------------------------
    Cpu speed from cpuinfo 4044.00Mhz
    cpuinfo might be wrong if cpufreq is enabled. To guess correctly try estimating via tsc
    Linux's inbuilt cpu_khz code emulated now
    True Frequency (without accounting Turbo) 4043 MHz
    CPU Multiplier 32x || Bus clock frequency (BCLK) 126.34 MHz

    Socket [0] - [physical cores=6, logical cores=12, max online cores ever=6]
    TURBO ENABLED on 6 Cores, Hyper Threading ON
    Max Frequency without considering Turbo 4169.34 MHz (126.34 x [33])
    Max TURBO Multiplier (if Enabled) with 1/2/3/4/5/6 Cores is 57x/57x/57x/57x/57x/57x
    Real Current Frequency 4199.52 MHz [126.34 x 33.24] (Max of below)
    Core [core-id] :Actual Freq (Mult.) C0% Halt(C1)% C3 % C6 % C7 % Temp
    Core 1 [0]: 4167.67 (32.99x) 1.3 1.47 1 0 96.2 33
    Core 2 [1]: 4168.81 (33.00x) 2.69 3.93 1 0 92.3 29
    Core 3 [2]: 4167.62 (32.99x) 1.48 1.29 0 0 97.2 32
    Core 4 [3]: 4199.52 (33.24x) 1 100 0 0 0 47
    Core 5 [4]: 4169.16 (33.00x) 8.87 3.09 0 0 87.8 31
    Core 6 [5]: 4168.21 (32.99x) 1.92 4.83 1 0 92.2 32
    34
    C0 = Processor running without halting
    C1 = Processor running with halts (States >C0 are power saver)
    C3 = Cores running with PLL turned off and core cache turned off
    C6 = Everything in C3 + core state saved to last level cache
    Above values in table are in percentage over the last 1 sec
    [core-id] refers to core-id number in /proc/cpuinfo
    ---------------------------------------------------------------

    A short glance at the /sys/bus/cpu/devices/cpu0/cpufreq/* files indicate that the reason must be
    /sys/bus/cpu/devices/cpu0/cpufreq/scaling_max_freq and cpuinfo_max_freq which contains the value 5.7 GHz (I use intel_pstate/powersave governor).

    Seriously I dont know, why there are those 5.7 GHz in the max_freq-files, but under no load ever I have saw him reaching that.
    Wow, strange, haven't see such high/wrong values indicated by the P-State driver... But yes, that's where the Phoronix Test Suite is reading that frequency from. Would you be interested in booting the kernel once with intel_pstate=disable to see what the Phoronix Test Suite / CPUfreq reports? Just trying to figure out if it's a P-State driver bug or some hardware errata.

  3. #3
    Join Date
    Oct 2014
    Posts
    8

    Default

    Quote Originally Posted by Michael View Post
    Wow, strange, haven't see such high/wrong values indicated by the P-State driver... But yes, that's where the Phoronix Test Suite is reading that frequency from. Would you be interested in booting the kernel once with intel_pstate=disable to see what the Phoronix Test Suite / CPUfreq reports? Just trying to figure out if it's a P-State driver bug or some hardware errata.
    Running now without P-State, and there I got right/sane numbers:
    ------------------------------------
    $ phoronix-test-suite system-info

    Phoronix Test Suite v5.2.1
    System Information

    Hardware:
    Processor: Intel Core i7-3930K @ 3.20GHz (12 Cores), Motherboard: ASUS P9X79 PRO, Chipset: Intel Xeon E5/Core, Memory: 32768MB, Disk: 240GB Patriot Pyro SE + 128GB Patriot Torqx 12, Graphics: NVIDIA GeForce GTX 680 2048MB (1006/3004MHz), Audio: Intel C600/X79, Monitor: DELL UP2414Q + Acer B243HL, Network: Intel 82579V Gigabit Connection

    Software:
    OS: Fedora 20, Kernel: 3.16.3-200.fc20.x86_64 (x86_64), Desktop: KDE 4.14.1, Display Server: X Server 1.14.4, Display Driver: NVIDIA 343.22, OpenGL: 4.3.0, Compiler: CUDA 5.0, File-System: ext4, Screen Resolution: 5760x2160
    ---------------------------------------------------------

    3.2 GHz - That are actually the right numbers (and in cpuinfo/scaling_max_freq they are of course also now at 3.2GHz).

    So it seems to be related to the intel_pstate driver (acpi_cpufreq seems to set the values different).

    Btw: As I just saw the recognized software: They is also an error: I use xfce as DE. (

  4. #4
    Join Date
    Oct 2014
    Posts
    8

    Default

    Quote Originally Posted by Michael View Post
    Wow, strange, haven't see such high/wrong values indicated by the P-State driver... But yes, that's where the Phoronix Test Suite is reading that frequency from. Would you be interested in booting the kernel once with intel_pstate=disable to see what the Phoronix Test Suite / CPUfreq reports? Just trying to figure out if it's a P-State driver bug or some hardware errata.
    Running now without P-State, and there I got right/sane numbers:
    ------------------------------------
    $ phoronix-test-suite system-info

    Phoronix Test Suite v5.2.1
    System Information

    Hardware:
    Processor: Intel Core i7-3930K @ 3.20GHz (12 Cores), Motherboard: ASUS P9X79 PRO, Chipset: Intel Xeon E5/Core, Memory: 32768MB, Disk: 240GB Patriot Pyro SE + 128GB Patriot Torqx 12, Graphics: NVIDIA GeForce GTX 680 2048MB (1006/3004MHz), Audio: Intel C600/X79, Monitor: DELL UP2414Q + Acer B243HL, Network: Intel 82579V Gigabit Connection

    Software:
    OS: Fedora 20, Kernel: 3.16.3-200.fc20.x86_64 (x86_64), Desktop: KDE 4.14.1, Display Server: X Server 1.14.4, Display Driver: NVIDIA 343.22, OpenGL: 4.3.0, Compiler: CUDA 5.0, File-System: ext4, Screen Resolution: 5760x2160
    ---------------------------------------------------------

    3.2 GHz - That are actually the right numbers (and in cpuinfo/scaling_max_freq they are of course also now at 3.2GHz).

    So it seems to be related to the intel_pstate driver (acpi_cpufreq seems to set the values different).

    Btw: As I just saw the recognized software: They is also an error: I use xfce as DE. (

    ---------------EDIT----:
    Ok, I digged a little more. Reading out the MSRs of my CPU, the multipliers of the cores for TURBO are at 57. With the base clock of 100 MHz, this indeed yields a topmost frequency of 5.7 GHz.
    But a) why does the actual used multiplier never exceed 33 (I checked it, TURBO is enabled) and b) why does p-state compute the topmost freq as 5.7GHz, and acpi_cpufreq use the highest non-TURBO multiplier of 32 (resulting in 3.2GHz)?

  5. #5

    Default

    Quote Originally Posted by Flolo View Post
    Running now without P-State, and there I got right/sane numbers:
    ------------------------------------
    $ phoronix-test-suite system-info

    Phoronix Test Suite v5.2.1
    System Information

    Hardware:
    Processor: Intel Core i7-3930K @ 3.20GHz (12 Cores), Motherboard: ASUS P9X79 PRO, Chipset: Intel Xeon E5/Core, Memory: 32768MB, Disk: 240GB Patriot Pyro SE + 128GB Patriot Torqx 12, Graphics: NVIDIA GeForce GTX 680 2048MB (1006/3004MHz), Audio: Intel C600/X79, Monitor: DELL UP2414Q + Acer B243HL, Network: Intel 82579V Gigabit Connection

    Software:
    OS: Fedora 20, Kernel: 3.16.3-200.fc20.x86_64 (x86_64), Desktop: KDE 4.14.1, Display Server: X Server 1.14.4, Display Driver: NVIDIA 343.22, OpenGL: 4.3.0, Compiler: CUDA 5.0, File-System: ext4, Screen Resolution: 5760x2160
    ---------------------------------------------------------

    3.2 GHz - That are actually the right numbers (and in cpuinfo/scaling_max_freq they are of course also now at 3.2GHz).

    So it seems to be related to the intel_pstate driver (acpi_cpufreq seems to set the values different).

    Btw: As I just saw the recognized software: They is also an error: I use xfce as DE. (

    ---------------EDIT----:
    Ok, I digged a little more. Reading out the MSRs of my CPU, the multipliers of the cores for TURBO are at 57. With the base clock of 100 MHz, this indeed yields a topmost frequency of 5.7 GHz.
    But a) why does the actual used multiplier never exceed 33 (I checked it, TURBO is enabled) and b) why does p-state compute the topmost freq as 5.7GHz, and acpi_cpufreq use the highest non-TURBO multiplier of 32 (resulting in 3.2GHz)?
    Traditionally ACPI CPUfreq reports Intel base clock frequencies where as Intel P-State reports the turbo frequency -- while Turbo does work with CPUFreq, it's just a reporting semantic difference. That's just how it's always has been between the two drivers.

    In terms of the MSRs reporting a turbo of 57, wow, that's really odd.... I've never seen this happen so I'm afraid I can't shed much more light on that issue.

    In terms of it detecting KDE, do you have a kde4/kded process running on the system? That's the process it's checking for to trying to decide if KDE4 is the desktop. The Xfce checks that come afterwards are for xfce4-session || xfce-mcs-manager.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •