On 10/17/2024 5:16 AM, Feng, Kenneth wrote:
> [AMD Official Use Only - AMD Internal Distribution Only]
>
> [AMD Official Use Only - AMD Internal Distribution Only]
>
> The logic is that when entering the compute workload, enable deep sleep.
> When exiting the compute workload, disable deep sleep.
> It's not required to do the deep sleep setting in other workload type.
> Thanks.
>
This patch looks good. Please name temp_profile_mode as current_profile
for clarity and may be comment about smu->power_profile_mode as the
new/target mode.
With those -
Reviewed-by: Lijo Lazar <lijo.lazar@xxxxxxx>
Thanks,
Lijo
>
> -----Original Message-----
> From: Wang, Yang(Kevin) <KevinYang.Wang@xxxxxxx>
> Sent: Wednesday, October 16, 2024 8:38 PM
> To: Feng, Kenneth <Kenneth.Feng@xxxxxxx>; amd-gfx@xxxxxxxxxxxxxxxxxxxxx
> Cc: Feng, Kenneth <Kenneth.Feng@xxxxxxx>
> Subject: RE: [PATCH] drm/amd/pm: update deep sleep status on smu v14.0.2/3
>
> [AMD Official Use Only - AMD Internal Distribution Only]
>
> -----Original Message-----
> From: Kenneth Feng <kenneth.feng@xxxxxxx>
> Sent: Wednesday, October 16, 2024 3:19 PM
> To: amd-gfx@xxxxxxxxxxxxxxxxxxxxx
> Cc: Wang, Yang(Kevin) <KevinYang.Wang@xxxxxxx>; Feng, Kenneth <Kenneth.Feng@xxxxxxx>
> Subject: [PATCH] drm/amd/pm: update deep sleep status on smu v14.0.2/3
>
> disable deep sleep during the compute workload for the potential performance loss on smu v14.0.2/3
>
> Signed-off-by: Kenneth Feng <kenneth.feng@xxxxxxx>
> ---
> drivers/gpu/drm/amd/pm/swsmu/smu14/smu_v14_0_2_ppt.c | 7 ++++++-
> 1 file changed, 6 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu14/smu_v14_0_2_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu14/smu_v14_0_2_ppt.c
> index 27f4e0ce447d..ecdd11a87311 100644
> --- a/drivers/gpu/drm/amd/pm/swsmu/smu14/smu_v14_0_2_ppt.c
> +++ b/drivers/gpu/drm/amd/pm/swsmu/smu14/smu_v14_0_2_ppt.c
> @@ -1784,7 +1784,7 @@ static int smu_v14_0_2_set_power_profile_mode(struct smu_context *smu,
> DpmActivityMonitorCoeffInt_t *activity_monitor =
> &(activity_monitor_external.DpmActivityMonitorCoeffInt);
> int workload_type, ret = 0;
> -
> + uint32_t temp_profile_mode = smu->power_profile_mode;
> smu->power_profile_mode = input[size];
>
> if (smu->power_profile_mode >= PP_SMC_POWER_PROFILE_COUNT) { @@ -1842,6 +1842,11 @@ static int smu_v14_0_2_set_power_profile_mode(struct smu_context *smu,
> }
> }
>
> + if (smu->power_profile_mode == PP_SMC_POWER_PROFILE_COMPUTE)
> + smu_v14_0_deep_sleep_control(smu, false);
> + else if (temp_profile_mode == PP_SMC_POWER_PROFILE_COMPUTE)
> + smu_v14_0_deep_sleep_control(smu, true);
> [Kevin]:
>
> This code logic doesn't seem easy to understand, If I understand correctly, can I modify it to the following writing style:
>
> If (new profile != current profile) {
> smu_v14_0_deep_sleep_control(smu, new profile == compute profile ? false : true);
> smu->power_profile_mode = new profile.
> }
>
> Best Regards,
> Kevin
> /* conv PP_SMC_POWER_PROFILE* to WORKLOAD_PPLIB_*_BIT */
> workload_type = smu_cmn_to_asic_specific_index(smu,
> CMN2ASIC_MAPPING_WORKLOAD,
> --
> 2.34.1
>
>
