[PATCH] drm/amd/swsmu: update smu v14_0_0 driver if version and metrics table

Ma, Li Li.Ma at amd.com
Mon Dec 4 08:51:41 UTC 2023


[Public]

Hi Alex,

Sorry for the late reply. Only smu14 used this gpu_metrics_v3_0 struct. And the patch has upstream. As far as l know, umr used gpu_metrics_v3_0 and I will submit a patch to umr.
Does this struct need to be back compatible currently? If yes, I will revert this patch and add a new gpu_metrics_v3_1.

Best Regards,
Li

-----Original Message-----
From: Deucher, Alexander <Alexander.Deucher at amd.com>
Sent: Tuesday, November 28, 2023 4:47 AM
To: Ma, Li <Li.Ma at amd.com>; amd-gfx at lists.freedesktop.org
Cc: Koenig, Christian <Christian.Koenig at amd.com>; Zhang, Yifan <Yifan1.Zhang at amd.com>; Yu, Lang <Lang.Yu at amd.com>
Subject: RE: [PATCH] drm/amd/swsmu: update smu v14_0_0 driver if version and metrics table

[Public]

> -----Original Message-----
> From: Ma, Li <Li.Ma at amd.com>
> Sent: Thursday, November 23, 2023 5:07 AM
> To: amd-gfx at lists.freedesktop.org
> Cc: Deucher, Alexander <Alexander.Deucher at amd.com>; Koenig, Christian
> <Christian.Koenig at amd.com>; Zhang, Yifan <Yifan1.Zhang at amd.com>; Yu,
> Lang <Lang.Yu at amd.com>; Ma, Li <Li.Ma at amd.com>
> Subject: [PATCH] drm/amd/swsmu: update smu v14_0_0 driver if version
> and metrics table
>
> Increment the driver if version and add new mems to the mertics table.
>
> Signed-off-by: Li Ma <li.ma at amd.com>
> ---
>  .../gpu/drm/amd/include/kgd_pp_interface.h    | 17 ++++
>  drivers/gpu/drm/amd/pm/swsmu/inc/amdgpu_smu.h | 10 +++
>  .../inc/pmfw_if/smu14_driver_if_v14_0_0.h     | 77 +++++++++++--------
>  .../drm/amd/pm/swsmu/smu14/smu_v14_0_0_ppt.c  | 46 ++++++++++-
>  4 files changed, 115 insertions(+), 35 deletions(-)
>
> diff --git a/drivers/gpu/drm/amd/include/kgd_pp_interface.h
> b/drivers/gpu/drm/amd/include/kgd_pp_interface.h
> index 8ebba87f4289..eaea1c65e526 100644
> --- a/drivers/gpu/drm/amd/include/kgd_pp_interface.h
> +++ b/drivers/gpu/drm/amd/include/kgd_pp_interface.h
> @@ -1086,6 +1086,10 @@ struct gpu_metrics_v3_0 {
>       uint16_t                        average_dram_reads;
>       /* time filtered DRAM write bandwidth [MB/sec] */
>       uint16_t                        average_dram_writes;
> +     /* time filtered IPU read bandwidth [MB/sec] */
> +     uint16_t                        average_ipu_reads;
> +     /* time filtered IPU write bandwidth [MB/sec] */
> +     uint16_t                        average_ipu_writes;
>
>       /* Driver attached timestamp (in ns) */
>       uint64_t                        system_clock_counter;
> @@ -1105,6 +1109,8 @@ struct gpu_metrics_v3_0 {
>       uint32_t                        average_all_core_power;
>       /* calculated core power [mW] */
>       uint16_t                        average_core_power[16];
> +     /* time filtered total system power [mW] */
> +     uint16_t                        average_sys_power;
>       /* maximum IRM defined STAPM power limit [mW] */
>       uint16_t                        stapm_power_limit;
>       /* time filtered STAPM power limit [mW] */ @@ -1117,6 +1123,8 @@
> struct gpu_metrics_v3_0 {
>       uint16_t                        average_ipuclk_frequency;
>       uint16_t                        average_fclk_frequency;
>       uint16_t                        average_vclk_frequency;
> +     uint16_t                        average_uclk_frequency;
> +     uint16_t                        average_mpipu_frequency;
>
>       /* Current clocks */
>       /* target core frequency [MHz] */ @@ -1126,6 +1134,15 @@ struct
> gpu_metrics_v3_0 {
>       /* GFXCLK frequency limit enforced on GFX [MHz] */
>       uint16_t                        current_gfx_maxfreq;
>
> +     /* Throttle Residency (ASIC dependent) */
> +     uint32_t throttle_residency_prochot;
> +     uint32_t throttle_residency_spl;
> +     uint32_t throttle_residency_fppt;
> +     uint32_t throttle_residency_sppt;
> +     uint32_t throttle_residency_thm_core;
> +     uint32_t throttle_residency_thm_gfx;
> +     uint32_t throttle_residency_thm_soc;
> +
>       /* Metrics table alpha filter time constant [us] */
>       uint32_t                        time_filter_alphavalue;
>  };

Is anything else besides smu14 using v3 of this struct?  If so, we can't change the layout otherwise it will break existing tools.  If so, bump the version minor and append the new items to the end.

Alex


> diff --git a/drivers/gpu/drm/amd/pm/swsmu/inc/amdgpu_smu.h
> b/drivers/gpu/drm/amd/pm/swsmu/inc/amdgpu_smu.h
> index c125253df20b..c2265e027ca8 100644
> --- a/drivers/gpu/drm/amd/pm/swsmu/inc/amdgpu_smu.h
> +++ b/drivers/gpu/drm/amd/pm/swsmu/inc/amdgpu_smu.h
> @@ -1418,6 +1418,16 @@ typedef enum {
>       METRICS_PCIE_WIDTH,
>       METRICS_CURR_FANPWM,
>       METRICS_CURR_SOCKETPOWER,
> +     METRICS_AVERAGE_VPECLK,
> +     METRICS_AVERAGE_IPUCLK,
> +     METRICS_AVERAGE_MPIPUCLK,
> +     METRICS_THROTTLER_RESIDENCY_PROCHOT,
> +     METRICS_THROTTLER_RESIDENCY_SPL,
> +     METRICS_THROTTLER_RESIDENCY_FPPT,
> +     METRICS_THROTTLER_RESIDENCY_SPPT,
> +     METRICS_THROTTLER_RESIDENCY_THM_CORE,
> +     METRICS_THROTTLER_RESIDENCY_THM_GFX,
> +     METRICS_THROTTLER_RESIDENCY_THM_SOC,
>  } MetricsMember_t;
>
>  enum smu_cmn2asic_mapping_type {
> diff --git
> a/drivers/gpu/drm/amd/pm/swsmu/inc/pmfw_if/smu14_driver_if_v14_0_0
> .h
> b/drivers/gpu/drm/amd/pm/swsmu/inc/pmfw_if/smu14_driver_if_v14_0_0
> .h
> index 22f88842a7fd..8f42771e1f0a 100644
> ---
> a/drivers/gpu/drm/amd/pm/swsmu/inc/pmfw_if/smu14_driver_if_v14_0_0
> .h
> +++
> b/drivers/gpu/drm/amd/pm/swsmu/inc/pmfw_if/smu14_driver_if_v14_0_0
> .h
> @@ -27,7 +27,7 @@
>  // *** IMPORTANT ***
>  // SMU TEAM: Always increment the interface version if  // any
> structure is changed in this file -#define PMFW_DRIVER_IF_VERSION 6
> +#define PMFW_DRIVER_IF_VERSION 7
>
>  typedef struct {
>    int32_t value;
> @@ -150,37 +150,50 @@ typedef struct {  } DpmClocks_t;
>
>  typedef struct {
> -  uint16_t CoreFrequency[16];        //Target core frequency [MHz]
> -  uint16_t CorePower[16];            //CAC calculated core power [mW]
> -  uint16_t CoreTemperature[16];      //TSEN measured core temperature
> [centi-C]
> -  uint16_t GfxTemperature;           //TSEN measured GFX temperature [centi-C]
> -  uint16_t SocTemperature;           //TSEN measured SOC temperature [centi-C]
> -  uint16_t StapmOpnLimit;            //Maximum IRM defined STAPM power limit
> [mW]
> -  uint16_t StapmCurrentLimit;        //Time filtered STAPM power limit [mW]
> -  uint16_t InfrastructureCpuMaxFreq; //CCLK frequency limit enforced
> on classic cores [MHz]
> -  uint16_t InfrastructureGfxMaxFreq; //GFXCLK frequency limit
> enforced on GFX [MHz]
> -  uint16_t SkinTemp;                 //Maximum skin temperature reported by APU
> and HS2 chassis sensors [centi-C]
> -  uint16_t GfxclkFrequency;          //Time filtered target GFXCLK frequency
> [MHz]
> -  uint16_t FclkFrequency;            //Time filtered target FCLK frequency [MHz]
> -  uint16_t GfxActivity;              //Time filtered GFX busy % [0-100]
> -  uint16_t SocclkFrequency;          //Time filtered target SOCCLK frequency
> [MHz]
> -  uint16_t VclkFrequency;            //Time filtered target VCLK frequency [MHz]
> -  uint16_t VcnActivity;              //Time filtered VCN busy % [0-100]
> -  uint16_t VpeclkFrequency;          //Time filtered target VPECLK frequency
> [MHz]
> -  uint16_t IpuclkFrequency;          //Time filtered target IPUCLK frequency
> [MHz]
> -  uint16_t IpuBusy[8];               //Time filtered IPU per-column busy % [0-100]
> -  uint16_t DRAMReads;                //Time filtered DRAM read bandwidth
> [MB/sec]
> -  uint16_t DRAMWrites;               //Time filtered DRAM write bandwidth
> [MB/sec]
> -  uint16_t CoreC0Residency[16];      //Time filtered per-core C0 residency %
> [0-100]
> -  uint16_t IpuPower;                 //Time filtered IPU power [mW]
> -  uint32_t ApuPower;                 //Time filtered APU power [mW]
> -  uint32_t GfxPower;                 //Time filtered GFX power [mW]
> -  uint32_t dGpuPower;                //Time filtered dGPU power [mW]
> -  uint32_t SocketPower;              //Time filtered power used for PPT/STAPM
> [APU+dGPU] [mW]
> -  uint32_t AllCorePower;             //Time filtered sum of core power across all
> cores in the socket [mW]
> -  uint32_t FilterAlphaValue;         //Metrics table alpha filter time constant [us]
> -  uint32_t MetricsCounter;           //Counter that is incremented on every
> metrics table update [PM_TIMER cycles]
> -  uint32_t spare[16];
> +  uint16_t CoreFrequency[16];          //Target core frequency [MHz]
> +  uint16_t CorePower[16];              //CAC calculated core power [mW]
> +  uint16_t CoreTemperature[16];        //TSEN measured core temperature
> [centi-C]
> +  uint16_t GfxTemperature;             //TSEN measured GFX temperature [centi-
> C]
> +  uint16_t SocTemperature;             //TSEN measured SOC temperature [centi-
> C]
> +  uint16_t StapmOpnLimit;              //Maximum IRM defined STAPM power
> limit [mW]
> +  uint16_t StapmCurrentLimit;          //Time filtered STAPM power limit [mW]
> +  uint16_t InfrastructureCpuMaxFreq;   //CCLK frequency limit enforced on
> classic cores [MHz]
> +  uint16_t InfrastructureGfxMaxFreq;   //GFXCLK frequency limit enforced on
> GFX [MHz]
> +  uint16_t SkinTemp;                   //Maximum skin temperature reported by APU
> and HS2 chassis sensors [centi-C]
> +  uint16_t GfxclkFrequency;            //Time filtered target GFXCLK frequency
> [MHz]
> +  uint16_t FclkFrequency;              //Time filtered target FCLK frequency [MHz]
> +  uint16_t GfxActivity;                //Time filtered GFX busy % [0-100]
> +  uint16_t SocclkFrequency;            //Time filtered target SOCCLK frequency
> [MHz]
> +  uint16_t VclkFrequency;              //Time filtered target VCLK frequency [MHz]
> +  uint16_t VcnActivity;                //Time filtered VCN busy % [0-100]
> +  uint16_t VpeclkFrequency;            //Time filtered target VPECLK frequency
> [MHz]
> +  uint16_t IpuclkFrequency;            //Time filtered target IPUCLK frequency
> [MHz]
> +  uint16_t IpuBusy[8];                 //Time filtered IPU per-column busy % [0-
> 100]
> +  uint16_t DRAMReads;                  //Time filtered DRAM read bandwidth
> [MB/sec]
> +  uint16_t DRAMWrites;                 //Time filtered DRAM write bandwidth
> [MB/sec]
> +  uint16_t CoreC0Residency[16];        //Time filtered per-core C0 residency %
> [0-100]
> +  uint16_t IpuPower;                   //Time filtered IPU power [mW]
> +  uint32_t ApuPower;                   //Time filtered APU power [mW]
> +  uint32_t GfxPower;                   //Time filtered GFX power [mW]
> +  uint32_t dGpuPower;                  //Time filtered dGPU power [mW]
> +  uint32_t SocketPower;                //Time filtered power used for PPT/STAPM
> [APU+dGPU] [mW]
> +  uint32_t AllCorePower;               //Time filtered sum of core power across all
> cores in the socket [mW]
> +  uint32_t FilterAlphaValue;           //Metrics table alpha filter time constant
> [us]
> +  uint32_t MetricsCounter;             //Counter that is incremented on every
> metrics table update [PM_TIMER cycles]
> +  uint16_t MemclkFrequency;            //Time filtered target MEMCLK frequency
> [MHz]
> +  uint16_t MpipuclkFrequency;          //Time filtered target MPIPUCLK
> frequency [MHz]
> +  uint16_t IpuReads;                   //Time filtered IPU read bandwidth [MB/sec]
> +  uint16_t IpuWrites;                  //Time filtered IPU write bandwidth [MB/sec]
> +  uint32_t ThrottleResidency_PROCHOT;  //Counter that is incremented
> + on
> every metrics table update when PROCHOT was engaged [PM_TIMER cycles]
> +  uint32_t ThrottleResidency_SPL;      //Counter that is incremented on every
> metrics table update when SPL was engaged [PM_TIMER cycles]
> +  uint32_t ThrottleResidency_FPPT;     //Counter that is incremented on every
> metrics table update when fast PPT was engaged [PM_TIMER cycles]
> +  uint32_t ThrottleResidency_SPPT;     //Counter that is incremented on every
> metrics table update when slow PPT was engaged [PM_TIMER cycles]
> +  uint32_t ThrottleResidency_THM_CORE; //Counter that is incremented
> + on every metrics table update when CORE thermal throttling was
> + engaged [PM_TIMER cycles]  uint32_t ThrottleResidency_THM_GFX;
> + //Counter that
> is incremented on every metrics table update when GFX thermal
> throttling was engaged [PM_TIMER cycles]  uint32_t
> ThrottleResidency_THM_SOC; //Counter that is incremented on every
> metrics table update when SOC thermal throttling was engaged [PM_TIMER
> cycles]
> +  uint16_t Psys;                       //Time filtered Psys power [mW]
> +  uint16_t spare1;
> +  uint32_t spare[6];
>  } SmuMetrics_t;
>
>  //ISP tile definitions
> diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu14/smu_v14_0_0_ppt.c
> b/drivers/gpu/drm/amd/pm/swsmu/smu14/smu_v14_0_0_ppt.c
> index 03b38c3a9968..94ccdbfd7090 100644
> --- a/drivers/gpu/drm/amd/pm/swsmu/smu14/smu_v14_0_0_ppt.c
> +++ b/drivers/gpu/drm/amd/pm/swsmu/smu14/smu_v14_0_0_ppt.c
> @@ -246,11 +246,20 @@ static int
> smu_v14_0_0_get_smu_metrics_data(struct smu_context *smu,
>               *value = 0;
>               break;
>       case METRICS_AVERAGE_UCLK:
> -             *value = 0;
> +             *value = metrics->MemclkFrequency;
>               break;
>       case METRICS_AVERAGE_FCLK:
>               *value = metrics->FclkFrequency;
>               break;
> +     case METRICS_AVERAGE_VPECLK:
> +             *value = metrics->VpeclkFrequency;
> +             break;
> +     case METRICS_AVERAGE_IPUCLK:
> +             *value = metrics->IpuclkFrequency;
> +             break;
> +     case METRICS_AVERAGE_MPIPUCLK:
> +             *value = metrics->MpipuclkFrequency;
> +             break;
>       case METRICS_AVERAGE_GFXACTIVITY:
>               *value = metrics->GfxActivity / 100;
>               break;
> @@ -270,8 +279,26 @@ static int
> smu_v14_0_0_get_smu_metrics_data(struct smu_context *smu,
>               *value = metrics->SocTemperature / 100 *
>               SMU_TEMPERATURE_UNITS_PER_CENTIGRADES;
>               break;
> -     case METRICS_THROTTLER_STATUS:
> -             *value = 0;
> +     case METRICS_THROTTLER_RESIDENCY_PROCHOT:
> +             *value = metrics->ThrottleResidency_PROCHOT;
> +             break;
> +     case METRICS_THROTTLER_RESIDENCY_SPL:
> +             *value = metrics->ThrottleResidency_SPL;
> +             break;
> +     case METRICS_THROTTLER_RESIDENCY_FPPT:
> +             *value = metrics->ThrottleResidency_FPPT;
> +             break;
> +     case METRICS_THROTTLER_RESIDENCY_SPPT:
> +             *value = metrics->ThrottleResidency_SPPT;
> +             break;
> +     case METRICS_THROTTLER_RESIDENCY_THM_CORE:
> +             *value = metrics->ThrottleResidency_THM_CORE;
> +             break;
> +     case METRICS_THROTTLER_RESIDENCY_THM_GFX:
> +             *value = metrics->ThrottleResidency_THM_GFX;
> +             break;
> +     case METRICS_THROTTLER_RESIDENCY_THM_SOC:
> +             *value = metrics->ThrottleResidency_THM_SOC;
>               break;
>       case METRICS_VOLTAGE_VDDGFX:
>               *value = 0;
> @@ -498,6 +525,8 @@ static ssize_t smu_v14_0_0_get_gpu_metrics(struct
> smu_context *smu,
>               sizeof(uint16_t) * 16);
>       gpu_metrics->average_dram_reads = metrics.DRAMReads;
>       gpu_metrics->average_dram_writes = metrics.DRAMWrites;
> +     gpu_metrics->average_ipu_reads = metrics.IpuReads;
> +     gpu_metrics->average_ipu_writes = metrics.IpuWrites;
>
>       gpu_metrics->average_socket_power = metrics.SocketPower;
>       gpu_metrics->average_ipu_power = metrics.IpuPower; @@ -505,6
> +534,7 @@ static ssize_t smu_v14_0_0_get_gpu_metrics(struct
> +smu_context
> *smu,
>       gpu_metrics->average_gfx_power = metrics.GfxPower;
>       gpu_metrics->average_dgpu_power = metrics.dGpuPower;
>       gpu_metrics->average_all_core_power = metrics.AllCorePower;
> +     gpu_metrics->average_sys_power = metrics.Psys;
>       memcpy(&gpu_metrics->average_core_power[0],
>               &metrics.CorePower[0],
>               sizeof(uint16_t) * 16);
> @@ -515,6 +545,8 @@ static ssize_t smu_v14_0_0_get_gpu_metrics(struct
> smu_context *smu,
>       gpu_metrics->average_fclk_frequency = metrics.FclkFrequency;
>       gpu_metrics->average_vclk_frequency = metrics.VclkFrequency;
>       gpu_metrics->average_ipuclk_frequency = metrics.IpuclkFrequency;
> +     gpu_metrics->average_uclk_frequency = metrics.MemclkFrequency;
> +     gpu_metrics->average_mpipu_frequency =
> metrics.MpipuclkFrequency;
>
>       memcpy(&gpu_metrics->current_coreclk[0],
>               &metrics.CoreFrequency[0], @@ -522,6 +554,14 @@ static
> ssize_t smu_v14_0_0_get_gpu_metrics(struct
> smu_context *smu,
>       gpu_metrics->current_core_maxfreq =
> metrics.InfrastructureCpuMaxFreq;
>       gpu_metrics->current_gfx_maxfreq =
> metrics.InfrastructureGfxMaxFreq;
>
> +     gpu_metrics->throttle_residency_prochot =
> metrics.ThrottleResidency_PROCHOT;
> +     gpu_metrics->throttle_residency_spl = metrics.ThrottleResidency_SPL;
> +     gpu_metrics->throttle_residency_fppt =
> metrics.ThrottleResidency_FPPT;
> +     gpu_metrics->throttle_residency_sppt =
> metrics.ThrottleResidency_SPPT;
> +     gpu_metrics->throttle_residency_thm_core =
> metrics.ThrottleResidency_THM_CORE;
> +     gpu_metrics->throttle_residency_thm_gfx =
> metrics.ThrottleResidency_THM_GFX;
> +     gpu_metrics->throttle_residency_thm_soc =
> +metrics.ThrottleResidency_THM_SOC;
> +
>       gpu_metrics->time_filter_alphavalue = metrics.FilterAlphaValue;
>       gpu_metrics->system_clock_counter = ktime_get_boottime_ns();
>
> --
> 2.25.1




More information about the amd-gfx mailing list