[PATCH 2/3] drm/amd/pp: Refine get_gpu_power for VI

[Zhu, Rex]

with this patch, there should be no difference between AGT and sysfs.



Best Regards

Rex

________________________________
From: Huang, JinHuiEric
Sent: Thursday, April 5, 2018 12:00 AM
To: Zhu, Rex; amd-gfx at lists.freedesktop.org
Subject: Re: [PATCH 2/3] drm/amd/pp: Refine get_gpu_power for VI


So you didn't compare the result with AGT, but I did. The reality is not like your speculation.


Regards,

Eric

On 2018-04-04 11:50 AM, Zhu, Rex wrote:

If user send message PPSMC_MSG_GetCurrPkgPwr, firmware will return the pkgpwr immediately as current power value.

as no PPSMC_MSG_GetCurrPkgPwr support, so send message  let firmware write pkgpwr  to ixSMU_PM_STATUS_94,

and driver delay 10ms to read ixSMU_PM_STATUS_94.


I don't think there is any problem. otherwise, there is same issue on polaris/vega.


I clean ixSMU_PM_STATUS_94 before let smu write it.

The delay time is enough unless we got 0 from ixSMU_PM_STATUS_94 .



Best Regards

Rex


________________________________
From: amd-gfx <amd-gfx-bounces at lists.freedesktop.org><mailto:amd-gfx-bounces at lists.freedesktop.org> on behalf of Eric Huang <jinhuieric.huang at amd.com><mailto:jinhuieric.huang at amd.com>
Sent: Wednesday, April 4, 2018 11:36 PM
To: amd-gfx at lists.freedesktop.org<mailto:amd-gfx at lists.freedesktop.org>
Subject: Re: [PATCH 2/3] drm/amd/pp: Refine get_gpu_power for VI

Sampling period is too short. The power reading value will be not
aligned with AGT's. It will confuse user that why AMD provides two
different power results.

Regards,
Eric

On 2018-04-04 04:25 AM, Rex Zhu wrote:
> 1. On polaris10/11/12, Sending smu message PPSMC_MSG_GetCurrPkgPwr to
>     read currentpkgpwr directly.
> 2. On Fiji/tonga/bonaire/hawwii, no PPSMC_MSG_GetCurrPkgPwr support.
>     Send PPSMC_MSG_PmStatusLogStart to let smu write currentpkgpwr
>     to ixSMU_PM_STATUS_94. this is asynchronous. need to delay no less
>     than 1ms.
> 3. Clean ixSMU_PM_STATUS_94 immediately when send PPSMC_MSG_PmStatusLogStart
>     to avoid read old value.
> 4. delay 10 ms instand of 20 ms. so the result will more similar to
>     the output of PPSMC_MSG_GetCurrPkgPwr.
> 5. remove max power/vddc/vddci output to keep consistent with vega.
> 6. for vddc/vddci power, we can calculate the average value per
>     [10ms, 4s] in other interface if needed.
>
> Signed-off-by: Rex Zhu <Rex.Zhu at amd.com><mailto:Rex.Zhu at amd.com>
> ---
>   drivers/gpu/drm/amd/powerplay/hwmgr/smu7_hwmgr.c | 45 +++++++++++-------------
>   1 file changed, 21 insertions(+), 24 deletions(-)
>
> diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/smu7_hwmgr.c b/drivers/gpu/drm/amd/powerplay/hwmgr/smu7_hwmgr.c
> index 40f2f87..c0ce672 100644
> --- a/drivers/gpu/drm/amd/powerplay/hwmgr/smu7_hwmgr.c
> +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/smu7_hwmgr.c
> @@ -3363,30 +3363,27 @@ static int smu7_get_pp_table_entry(struct pp_hwmgr *hwmgr,
>   static int smu7_get_gpu_power(struct pp_hwmgr *hwmgr,
>                struct pp_gpu_power *query)
>   {
> -     PP_ASSERT_WITH_CODE(!smum_send_msg_to_smc(hwmgr,
> -                     PPSMC_MSG_PmStatusLogStart),
> -                     "Failed to start pm status log!",
> -                     return -1);
> -
> -     msleep_interruptible(20);
> -
> -     PP_ASSERT_WITH_CODE(!smum_send_msg_to_smc(hwmgr,
> -                     PPSMC_MSG_PmStatusLogSample),
> -                     "Failed to sample pm status log!",
> -                     return -1);
> -
> -     query->vddc_power = cgs_read_ind_register(hwmgr->device,
> -                     CGS_IND_REG__SMC,
> -                     ixSMU_PM_STATUS_40);
> -     query->vddci_power = cgs_read_ind_register(hwmgr->device,
> -                     CGS_IND_REG__SMC,
> -                     ixSMU_PM_STATUS_49);
> -     query->max_gpu_power = cgs_read_ind_register(hwmgr->device,
> -                     CGS_IND_REG__SMC,
> -                     ixSMU_PM_STATUS_94);
> -     query->average_gpu_power = cgs_read_ind_register(hwmgr->device,
> -                     CGS_IND_REG__SMC,
> -                     ixSMU_PM_STATUS_95);
> +     if (!query)
> +             return -EINVAL;
> +
> +     memset(query, 0, sizeof *query);
> +
> +     if (hwmgr->chip_id >= CHIP_POLARIS10) {
> +             smum_send_msg_to_smc(hwmgr, PPSMC_MSG_GetCurrPkgPwr);
> +             query->average_gpu_power = cgs_read_register(hwmgr->device, mmSMC_MSG_ARG_0);
> +     } else {
> +             smum_send_msg_to_smc(hwmgr, PPSMC_MSG_PmStatusLogStart);
> +             cgs_write_ind_register(hwmgr->device, CGS_IND_REG__SMC,
> +                                     ixSMU_PM_STATUS_94, 0);
> +
> +             msleep_interruptible(10);
> +
> +             smum_send_msg_to_smc(hwmgr, PPSMC_MSG_PmStatusLogSample);
> +
> +             query->average_gpu_power = cgs_read_ind_register(hwmgr->device,
> +                                                     CGS_IND_REG__SMC,
> +                                                     ixSMU_PM_STATUS_94);
> +     }
>
>        return 0;
>   }

_______________________________________________
amd-gfx mailing list
amd-gfx at lists.freedesktop.org<mailto:amd-gfx at lists.freedesktop.org>
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
amd-gfx Info Page - freedesktop.org<https://lists.freedesktop.org/mailman/listinfo/amd-gfx>
lists.freedesktop.org
Subscribing to amd-gfx: Subscribe to amd-gfx by filling out the following form. Use of all freedesktop.org lists is subject to our Code of Conduct.




-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/amd-gfx/attachments/20180404/de14ca3d/attachment.html>