Introduce Bandwidth OPPs for interconnects
v4 cover Introduce Bandwidth OPPs for interconnects
Interconnects and interconnect paths quantify their performance levels in terms of bandwidth and not in terms of frequency. So similar to how we have frequency based OPP tables in DT and in the OPP framework, we need bandwidth OPP table support in DT and in the OPP framework.
So with the DT bindings added in this patch series, the DT for a GPU that does bandwidth voting from GPU to Cache and GPU to DDR would look something like this:
gpu_cache_opp_table: gpu_cache_opp_table {
compatible = "operating-points-v2";
gpu_cache_3000: opp-3000 {
opp-peak-KBps = <3000000>;
opp-avg-KBps = <1000000>;
};
gpu_cache_6000: opp-6000 {
opp-peak-KBps = <6000000>;
opp-avg-KBps = <2000000>;
};
gpu_cache_9000: opp-9000 {
opp-peak-KBps = <9000000>;
opp-avg-KBps = <9000000>;
};
};gpu_ddr_opp_table: gpu_ddr_opp_table {
compatible = "operating-points-v2";
gpu_ddr_1525: opp-1525 {
opp-peak-KBps = <1525000>;
opp-avg-KBps = <452000>;
};
gpu_ddr_3051: opp-3051 {
opp-peak-KBps = <3051000>;
opp-avg-KBps = <915000>;
};
gpu_ddr_7500: opp-7500 {
opp-peak-KBps = <7500000>;
opp-avg-KBps = <3000000>;
};
};gpu_opp_table: gpu_opp_table {
compatible = "operating-points-v2";
opp-shared;
opp-200000000 {
opp-hz = /bits/ 64 <200000000>;
};
opp-400000000 {
opp-hz = /bits/ 64 <400000000>;
};
};gpu@7864000 {
...
operating-points-v2 = <&gpu_opp_table>, <&gpu_cache_opp_table>, <&gpu_ddr_opp_table>;
...
};v1 -> v3:
- Lots of patch additions that were later dropped v3 -> v4:
- Fixed typo bugs pointed out by Sibi.
- Fixed bug that incorrectly reset rate to 0 all the time
- Added units documentation
- Dropped interconnect-opp-table property and related changes
Cheers,
Saravana
Saravana Kannan (3):
dt-bindings: opp: Introduce opp-peak-KBps and opp-avg-KBps bindings
OPP: Add support for bandwidth OPP tables
OPP: Add helper function for bandwidth OPP tables
Documentation/devicetree/bindings/opp/opp.txt | 15 ++++-- .../devicetree/bindings/property-units.txt | 4 ++ drivers/opp/core.c | 51 +++++++++++++++++++ drivers/opp/of.c | 41 +++++++++++---- drivers/opp/opp.h | 4 +- include/linux/pm_opp.h | 19 +++++++ 6 files changed, 121 insertions(+), 13 deletions(-)
Signature · 1 line
2.22.0.709.g102302147b-goog
PATCH v4 1/3 dt-bindings: opp: Introduce opp-peak-KBps and opp-avg-KBps bindings
Interconnects often quantify their performance points in terms of bandwidth. So, add opp-peak-KBps (required) and opp-avg-KBps (optional) to allow specifying Bandwidth OPP tables in DT.
opp-peak-KBps is a required property that replace opp-hz for Bandwidth OPP tables.
opp-avg-KBps is an optional property that can be used in Bandwidth OPP tables.
--- Documentation/devicetree/bindings/opp/opp.txt | 15 ++++++++++++--- .../devicetree/bindings/property-units.txt | 4 ++++ 2 files changed, 16 insertions(+), 3 deletions(-)
Documentation/devicetree/bindings/opp/opp.txt +12 -3
@@ -83,9 +83,14 @@ properties. Required properties: - opp-hz: Frequency in Hz, expressed as a 64-bit big-endian integer. This is a - required property for all device nodes but devices like power domains. The - power domain nodes must have another (implementation dependent) property which - uniquely identifies the OPP nodes. + required property for all device nodes but for devices like power domains or + bandwidth opp tables. The power domain nodes must have another (implementation + dependent) property which uniquely identifies the OPP nodes. The interconnect + opps are required to have the opp-peak-KBps property. + +- opp-peak-KBps: Peak bandwidth in kilobytes per second, expressed as a 32-bit + big-endian integer. This is a required property for all devices that don't + have opp-hz. For example, bandwidth OPP tables for interconnect paths. Optional properties: - opp-microvolt: voltage in micro Volts.
@@ -132,6 +137,10 @@ Optional properties: - opp-level: A value representing the performance level of the device, expressed as a 32-bit integer. +- opp-avg-KBps: Average bandwidth in kilobytes per second, expressed as a + 32-bit big-endian integer. This property is only meaningful in OPP tables + where opp-peak-KBps is present. + - clock-latency-ns: Specifies the maximum possible transition latency (in nanoseconds) for switching to this OPP from any other OPP.
Documentation/devicetree/bindings/property-units.txt +4 -0
@@ -41,3 +41,7 @@ Temperature Pressure ---------------------------------------- -kpascal : kilopascal + +Throughput +---------------------------------------- +-KBps : kilobytes per second
Signature · 1 line
2.22.0.709.g102302147b-goog
Stephen Boyd
maintainerRe: [PATCH v4 1/3] dt-bindings: opp: Introduce opp-peak-KBps and opp-avg-KBps bindings
Quoting Saravana Kannan (2019-07-26 16:15:55)
Documentation/devicetree/bindings/opp/opp.txt properties. · line 86
Quoted 12 lines
Documentation/devicetree/bindings/opp/opp.txt +1
-3@@ -83,9 +83,14 @@ properties. Required properties: - opp-hz: Frequency in Hz, expressed as a 64-bit big-endian integer. This is a - required property for all device nodes but devices like power domains. The - power domain nodes must have another (implementation dependent) property which - uniquely identifies the OPP nodes. + required property for all device nodes but for devices like power domains or
s/but/except/
Quoted 11 lines
Documentation/devicetree/bindings/opp/opp.txt +7
-0+ bandwidth opp tables. The power domain nodes must have another (implementation + dependent) property which uniquely identifies the OPP nodes. The interconnect + opps are required to have the opp-peak-KBps property. + +- opp-peak-KBps: Peak bandwidth in kilobytes per second, expressed as a 32-bit + big-endian integer. This is a required property for all devices that don't + have opp-hz. For example, bandwidth OPP tables for interconnect paths. Optional properties: - opp-microvolt: voltage in micro Volts.
Documentation/devicetree/bindings/opp/opp.txt Optional properties: · line 140
Quoted 4 lines
Documentation/devicetree/bindings/opp/opp.txt +1
-0@@ -132,6 +137,10 @@ Optional properties: - opp-level: A value representing the performance level of the device, expressed as a 32-bit integer. +- opp-avg-KBps: Average bandwidth in kilobytes per second, expressed as a
Shouldn't that be a lower case k? kBps?
Quoted 11 lines
Documentation/devicetree/bindings/opp/opp.txt +3
-0+ 32-bit big-endian integer. This property is only meaningful in OPP tables + where opp-peak-KBps is present. + - clock-latency-ns: Specifies the maximum possible transition latency (in nanoseconds) for switching to this OPP from any other OPP.Documentation/devicetree/bindings/property-units.txt +0
-0
Documentation/devicetree/bindings/property-units.txt Temperature · line 47
Quoted 7 lines
Documentation/devicetree/bindings/property-units.txt +4
-0@@ -41,3 +41,7 @@ Temperature Pressure ---------------------------------------- -kpascal : kilopascal + +Throughput +---------------------------------------- +-KBps : kilobytes per second
Same comment.
PATCH v4 2/3 OPP: Add support for bandwidth OPP tables
In reply to Saravana KannanNot all devices quantify their performance points in terms of frequency. Devices like interconnects quantify their performance points in terms of bandwidth. We need a way to represent these bandwidth levels in OPP. So, add support for parsing bandwidth OPPs from DT.
--- drivers/opp/of.c | 41 ++++++++++++++++++++++++++++++++--------- drivers/opp/opp.h | 4 +++- 2 files changed, 35 insertions(+), 10 deletions(-)
drivers/opp/of.c +32 -9
@@ -523,6 +523,35 @@ void dev_pm_opp_of_remove_table(struct device *dev) }EXPORT_SYMBOL_GPL (dev_pm_opp_of_remove_table); +static int _read_opp_key (struct dev_pm_opp *new_opp,struct device_node *np) +{ +int ret; +u64 rate; +u32 bw; + + ret =of_property_read_u64 (np,"opp-hz" , &rate); +if (!ret) { +/* +* Rate is defined as an unsigned long in clk API, and so +* casting explicitly to its type. Must be fixed once rate is 64 +* bit guaranteed in clk API. +*/ + new_opp->rate = (unsigned long )rate; +return 0 ; + } + + ret =of_property_read_u32 (np,"opp-peak-KBps" , &bw); +if (ret) +return ret; + new_opp->rate = (unsigned long ) bw; + + ret =of_property_read_u32 (np,"opp-avg-KBps" , &bw); +if (!ret) + new_opp->avg_bw = (unsigned long ) bw; + +return 0 ; +} +/** * _opp_add_static_v2() - Allocate static OPPs (As per 'v2' DT bindings) * @opp_table: OPP table
@@ -560,22 +589,16 @@ static struct dev_pm_opp *_opp_add_static_v2(struct opp_table *opp_table,if (!new_opp)return ERR_PTR (-ENOMEM ); - ret =of_property_read_u64 (np,"opp-hz" , &rate); + ret =_read_opp_key (new_opp, np);if (ret <0 ) {/* "opp-hz" is optional for devices like power domains. */ if (!opp_table->is_genpd) { -dev_err (dev,"%s: opp-hz not found\n" , __func__); +dev_err (dev,"%s: opp-hz or opp-peak-bw not found\n" , + __func__); goto free_opp; } rate_not_available = true; - }else { -/* -* Rate is defined as an unsigned long in clk API, and so -* casting explicitly to its type. Must be fixed once rate is 64 -* bit guaranteed in clk API. -*/ - new_opp->rate = (unsigned long )rate; }of_property_read_u32 (np,"opp-level" , &new_opp->level);
drivers/opp/opp.h +3 -1
@@ -56,7 +56,8 @@ extern struct list_head opp_tables;* @turbo: true if turbo (boost) OPP * @suspend: true if suspend OPP * @pstate: Device's power domain's performance state. -* @rate: Frequency in hertz +* @rate: Frequency in hertz OR Peak bandwidth in kilobytes per second +* @avg_bw: Average bandwidth in kilobytes per second * @level: Performance level * @supplies: Power supplies voltage/current values * @clock_latency_ns: Latency (in nanoseconds) of switching to this OPP's
@@ -78,6 +79,7 @@ struct dev_pm_opp {bool suspend;unsigned int pstate;unsigned long rate; +unsigned long avg_bw;unsigned int level;struct dev_pm_opp_supply *supplies;
Signature · 1 line
2.22.0.709.g102302147b-goog
Re: [PATCH v4 2/3] OPP: Add support for bandwidth OPP tables
Hi Saravana,
Quoted 24 lines · Saravana Kannan
Not all devices quantify their performance points in terms of frequency. Devices like interconnects quantify their performance points in terms of bandwidth. We need a way to represent these bandwidth levels in OPP. So, add support for parsing bandwidth OPPs from DT.
Signed-off-by Saravana Kannan <[email protected]>--- drivers/opp/of.c | 41 ++++++++++++++++++++++++++++++++--------- drivers/opp/opp.h | 4 +++- 2 files changed, 35 insertions(+), 10 deletions(-)drivers/opp/of.c +5
-0@@ -523,6 +523,35 @@ void dev_pm_opp_of_remove_table(struct device *dev) }EXPORT_SYMBOL_GPL (dev_pm_opp_of_remove_table); +static int _read_opp_key (struct dev_pm_opp *new_opp,struct device_node *np) +{ +int ret; +u64 rate; +u32 bw;
drivers/opp/of.c void dev_pm_opp_of_remove_table(struct device *dev) · line 540
Quoted 10 lines
drivers/opp/of.c +10
-0+ + ret =of_property_read_u64 (np,"opp-hz" , &rate); +if (!ret) { +/* +* Rate is defined as an unsigned long in clk API, and so +* casting explicitly to its type. Must be fixed once rate is 64 +* bit guaranteed in clk API. +*/ + new_opp->rate = (unsigned long )rate; +return 0 ;
So we can't have a single OPP table with both frequency and bandwidth?
Quoted 18 lines
drivers/opp/of.c +14
-0+ } + + ret =of_property_read_u32 (np,"opp-peak-KBps" , &bw); +if (ret) +return ret; + new_opp->rate = (unsigned long ) bw; + + ret =of_property_read_u32 (np,"opp-avg-KBps" , &bw); +if (!ret) + new_opp->avg_bw = (unsigned long ) bw; + +return 0 ; +} +/** * _opp_add_static_v2() - Allocate static OPPs (As per 'v2' DT bindings) * @opp_table: OPP table
drivers/opp/of.c static struct dev_pm_opp *_opp_add_static_v2(struct opp_table *opp_table, · line 596
Quoted 10 lines
drivers/opp/of.c +2
-2@@ -560,22 +589,16 @@ static struct dev_pm_opp *_opp_add_static_v2(struct opp_table *opp_table,if (!new_opp)return ERR_PTR (-ENOMEM ); - ret =of_property_read_u64 (np,"opp-hz" , &rate); + ret =_read_opp_key (new_opp, np);if (ret <0 ) {/* "opp-hz" is optional for devices like power domains. */ if (!opp_table->is_genpd) { -dev_err (dev,"%s: opp-hz not found\n" , __func__); + dev_err(dev, "%s: opp-hz or opp-peak-bw not found\n",
s/opp-peak-bw/opp-peak-kBps/
Thanks,
Georgi
Re: [PATCH v4 2/3] OPP: Add support for bandwidth OPP tables
Quoted 32 lines · Georgi Djakov
Hi Saravana,
Quoted 28 lines · Saravana Kannan
Not all devices quantify their performance points in terms of frequency. Devices like interconnects quantify their performance points in terms of bandwidth. We need a way to represent these bandwidth levels in OPP. So, add support for parsing bandwidth OPPs from DT.
Signed-off-by Saravana Kannan <[email protected]>--- drivers/opp/of.c | 41 ++++++++++++++++++++++++++++++++--------- drivers/opp/opp.h | 4 +++- 2 files changed, 35 insertions(+), 10 deletions(-)drivers/opp/of.c +9
-0@@ -523,6 +523,35 @@ void dev_pm_opp_of_remove_table(struct device *dev) }EXPORT_SYMBOL_GPL (dev_pm_opp_of_remove_table); +static int _read_opp_key (struct dev_pm_opp *new_opp,struct device_node *np) +{ +int ret; +u64 rate; +u32 bw; + + ret =of_property_read_u64 (np,"opp-hz" , &rate); +if (!ret) { +/*
Quoted 8 lines
drivers/opp/of.c void dev_pm_opp_of_remove_table(struct device *dev) · line 540
Quoted 6 lines
drivers/opp/of.c +6
-0+* Rate is defined as an unsigned long in clk API, and so +* casting explicitly to its type. Must be fixed once rate is 64 +* bit guaranteed in clk API. +*/ + new_opp->rate = (unsigned long )rate; +return 0 ;So we can't have a single OPP table with both frequency and bandwidth?
Right, because we can have only 1 "key" for the OPP table. Having more than one "key" for an OPP table makes a lot of things pretty messy. Most of the helper functions need to be rewritten to say which key is being referred to when searching. A lot of the error checking when creating OPP tables becomes convoluted -- can we allow more than one OPP entry with the same frequency just because the opp-peak-kBps is different? Etc. Seems like a lot of code change for something that I don't think is a very useful.
Also, an OPP table is either going to represent performance levels of a clock domain (opp-hz) or the performance levels of an interconnect path (opp-peak-kBps) or an OPP table for genpd. Mixing them all up is just going to make it convoluted with not enough benefit or use case that can't be handled as is (separate BW and freq OPP tables).
Quoted 21 lines
Quoted 21 lines
drivers/opp/of.c +14
-0+ } + + ret =of_property_read_u32 (np,"opp-peak-KBps" , &bw); +if (ret) +return ret; + new_opp->rate = (unsigned long ) bw; + + ret =of_property_read_u32 (np,"opp-avg-KBps" , &bw); +if (!ret) + new_opp->avg_bw = (unsigned long ) bw; + +return 0 ; +} +/** * _opp_add_static_v2() - Allocate static OPPs (As per 'v2' DT bindings) * @opp_table: OPP table @@ -560,22 +589,16 @@ static struct dev_pm_opp *_opp_add_static_v2(struct opp_table *opp_table,if (!new_opp)return ERR_PTR (-ENOMEM );
Quoted 9 lines
drivers/opp/of.c static struct dev_pm_opp *_opp_add_static_v2(struct opp_table *opp_table, · line 596
Quoted 7 lines
drivers/opp/of.c +2
-2- ret =of_property_read_u64 (np,"opp-hz" , &rate); + ret =_read_opp_key (new_opp, np);if (ret <0 ) {/* "opp-hz" is optional for devices like power domains. */ if (!opp_table->is_genpd) { -dev_err (dev,"%s: opp-hz not found\n" , __func__); + dev_err(dev, "%s: opp-hz or opp-peak-bw not found\n",s/opp-peak-bw/opp-peak-kBps/
Thanks. Will fix.
-Saravana
PATCH v4 3/3 OPP: Add helper function for bandwidth OPP tables
In reply to Saravana KannanThe frequency OPP tables have helper functions to search for entries in the table based on frequency and get the frequency values for a given (or suspend) OPP entry.
Add similar helper functions for bandwidth OPP tables to search for entries in the table based on peak bandwidth and to get the peak and average bandwidth for a given (or suspend) OPP entry.
--- drivers/opp/core.c | 51 ++++++++++++++++++++++++++++++++++++++++++ include/linux/pm_opp.h | 19 ++++++++++++++++ 2 files changed, 70 insertions(+)
drivers/opp/core.c +51 -0
@@ -127,6 +127,29 @@ unsigned long dev_pm_opp_get_freq(struct dev_pm_opp *opp) }EXPORT_SYMBOL_GPL (dev_pm_opp_get_freq); +/** +* dev_pm_opp_get_bw() - Gets the bandwidth corresponding to an available opp +* @opp: opp for which frequency has to be returned for +* @avg_bw: Pointer where the corresponding average bandwidth is stored. +* Can be NULL. +* +* Return: Peak bandwidth in KBps corresponding to the opp, else +* return 0 +*/ +unsigned long dev_pm_opp_get_bw (struct dev_pm_opp *opp,unsigned long *avg_bw) +{ +if (IS_ERR_OR_NULL (opp) || !opp->available) { +pr_err ("%s: Invalid parameters\n" , __func__); +return 0 ; + } + +if (avg_bw) + *avg_bw = opp->avg_bw; + +return opp->rate; +} +EXPORT_SYMBOL_GPL (dev_pm_opp_get_bw); +/** * dev_pm_opp_get_level() - Gets the level corresponding to an available opp * @opp: opp for which level value has to be returned for
@@ -299,6 +322,34 @@ unsigned long dev_pm_opp_get_suspend_opp_freq(struct device *dev) }EXPORT_SYMBOL_GPL (dev_pm_opp_get_suspend_opp_freq); +/** +* dev_pm_opp_get_suspend_opp_bw() - Get peak bandwidth of suspend opp in KBps +* @dev: device for which we do this operation +* @avg_bw: Pointer where the corresponding average bandwidth is stored. +* Can be NULL. +* +* Return: This function returns the peak bandwidth of the OPP marked as +* suspend_opp if one is available, else returns 0; +*/ +unsigned long dev_pm_opp_get_suspend_opp_bw (struct device *dev, +unsigned long *avg_bw) +{ +struct opp_table *opp_table; +unsigned long peak_bw =0 ; + + opp_table =_find_opp_table (dev); +if (IS_ERR (opp_table)) +return 0 ; + +if (opp_table->suspend_opp && opp_table->suspend_opp->available) + peak_bw =dev_pm_opp_get_bw (opp_table->suspend_opp, avg_bw); + +dev_pm_opp_put_opp_table (opp_table); + +return peak_bw; +} +EXPORT_SYMBOL_GPL (dev_pm_opp_get_suspend_opp_bw); +int _get_opp_count (struct opp_table *opp_table) {struct dev_pm_opp *opp;
include/linux/pm_opp.h +19 -0
@@ -82,6 +82,7 @@ void dev_pm_opp_put_opp_table(struct opp_table *opp_table);unsigned long dev_pm_opp_get_voltage (struct dev_pm_opp *opp);unsigned long dev_pm_opp_get_freq (struct dev_pm_opp *opp); +unsigned long dev_pm_opp_get_bw (struct dev_pm_opp *opp,unsigned long *avg_bw);unsigned int dev_pm_opp_get_level (struct dev_pm_opp *opp);
@@ -92,6 +93,8 @@ unsigned long dev_pm_opp_get_max_clock_latency(struct device *dev);unsigned long dev_pm_opp_get_max_volt_latency (struct device *dev);unsigned long dev_pm_opp_get_max_transition_latency (struct device *dev);unsigned long dev_pm_opp_get_suspend_opp_freq (struct device *dev); +unsigned long dev_pm_opp_get_suspend_opp_bw (struct device *dev, +unsigned long *avg_bw); struct dev_pm_opp *dev_pm_opp_find_freq_exact(struct device *dev, unsigned long freq,
@@ -158,6 +161,11 @@ static inline unsigned long dev_pm_opp_get_freq(struct dev_pm_opp *opp) {return 0 ; } +static inline unsigned long dev_pm_opp_get_bw (struct dev_pm_opp *opp, +unsigned long *avg_bw) +{ +return 0 ; +}static inline unsigned int dev_pm_opp_get_level (struct dev_pm_opp *opp) {
@@ -194,6 +202,12 @@ static inline unsigned long dev_pm_opp_get_suspend_opp_freq(struct device *dev)return 0 ; } +static inline unsigned long dev_pm_opp_get_suspend_opp_bw (struct device *dev, +unsigned long *avg_bw) +{ +return 0 ; +} +static inline struct dev_pm_opp *dev_pm_opp_find_freq_exact (struct device *dev,unsigned long freq,bool available) {
@@ -329,6 +343,11 @@ static inline void dev_pm_opp_cpumask_remove_table(const struct cpumask *cpumask#endif /* CONFIG_PM_OPP */ +#define dev_pm_opp_find_peak_bw_exact dev_pm_opp_find_freq_exact +#define dev_pm_opp_find_peak_bw_floor dev_pm_opp_find_freq_floor +#define dev_pm_opp_find_peak_bw_ceil_by_volt dev_pm_opp_find_freq_ceil_by_volt +#define dev_pm_opp_find_peak_bw_ceil dev_pm_opp_find_freq_ceil +#if defined(CONFIG_PM_OPP ) && defined(CONFIG_OF )int dev_pm_opp_of_add_table (struct device *dev);int dev_pm_opp_of_add_table_indexed (struct device *dev,int index);
Signature · 1 line
2.22.0.709.g102302147b-goog
Quoted 55 lines · Saravana Kannan
Interconnects and interconnect paths quantify their performance levels in terms of bandwidth and not in terms of frequency. So similar to how we have frequency based OPP tables in DT and in the OPP framework, we need bandwidth OPP table support in DT and in the OPP framework.
So with the DT bindings added in this patch series, the DT for a GPU that does bandwidth voting from GPU to Cache and GPU to DDR would look something like this:
gpu_cache_opp_table: gpu_cache_opp_table { compatible = "operating-points-v2"; gpu_cache_3000: opp-3000 { opp-peak-KBps = <3000000>; opp-avg-KBps = <1000000>; }; gpu_cache_6000: opp-6000 { opp-peak-KBps = <6000000>; opp-avg-KBps = <2000000>; }; gpu_cache_9000: opp-9000 { opp-peak-KBps = <9000000>; opp-avg-KBps = <9000000>; }; };gpu_ddr_opp_table: gpu_ddr_opp_table { compatible = "operating-points-v2"; gpu_ddr_1525: opp-1525 { opp-peak-KBps = <1525000>; opp-avg-KBps = <452000>; }; gpu_ddr_3051: opp-3051 { opp-peak-KBps = <3051000>; opp-avg-KBps = <915000>; }; gpu_ddr_7500: opp-7500 { opp-peak-KBps = <7500000>; opp-avg-KBps = <3000000>; }; };gpu_opp_table: gpu_opp_table { compatible = "operating-points-v2"; opp-shared; opp-200000000 { opp-hz = /bits/ 64 <200000000>; }; opp-400000000 { opp-hz = /bits/ 64 <400000000>; }; };
Quoted 5 lines
gpu@7864000 {
...
operating-points-v2 = <&gpu_opp_table>, <&gpu_cache_opp_table>, <&gpu_ddr_opp_table>;
...
};One feedback I missed giving earlier. Will it be possible to get some user code merged along with this ? I want to make sure anything we add ends up getting used.
That also helps understanding the problems you are facing in a better way, i.e. with real examples.
Signature · 1 line
viresh
Quoted 63 lines · Viresh Kumar
Quoted 55 lines · Saravana Kannan
Interconnects and interconnect paths quantify their performance levels in terms of bandwidth and not in terms of frequency. So similar to how we have frequency based OPP tables in DT and in the OPP framework, we need bandwidth OPP table support in DT and in the OPP framework.
So with the DT bindings added in this patch series, the DT for a GPU that does bandwidth voting from GPU to Cache and GPU to DDR would look something like this:
gpu_cache_opp_table: gpu_cache_opp_table { compatible = "operating-points-v2"; gpu_cache_3000: opp-3000 { opp-peak-KBps = <3000000>; opp-avg-KBps = <1000000>; }; gpu_cache_6000: opp-6000 { opp-peak-KBps = <6000000>; opp-avg-KBps = <2000000>; }; gpu_cache_9000: opp-9000 { opp-peak-KBps = <9000000>; opp-avg-KBps = <9000000>; }; };gpu_ddr_opp_table: gpu_ddr_opp_table { compatible = "operating-points-v2"; gpu_ddr_1525: opp-1525 { opp-peak-KBps = <1525000>; opp-avg-KBps = <452000>; }; gpu_ddr_3051: opp-3051 { opp-peak-KBps = <3051000>; opp-avg-KBps = <915000>; }; gpu_ddr_7500: opp-7500 { opp-peak-KBps = <7500000>; opp-avg-KBps = <3000000>; }; };gpu_opp_table: gpu_opp_table { compatible = "operating-points-v2"; opp-shared; opp-200000000 { opp-hz = /bits/ 64 <200000000>; }; opp-400000000 { opp-hz = /bits/ 64 <400000000>; }; };Quoted 5 lines
gpu@7864000 { ... operating-points-v2 = <&gpu_opp_table>, <&gpu_cache_opp_table>, <&gpu_ddr_opp_table>; ... };
Quoted 3 lines
One feedback I missed giving earlier. Will it be possible to get some user code merged along with this ? I want to make sure anything we add ends up getting used.
Sibi might be working on doing that for the SDM845 CPUfreq driver. Georgi could also change his GPU driver use case to use this BW OPP table and required-opps.
The problem is that people don't want to start using this until we decide on the DT representation. So it's like a chicken and egg situation.
-Saravana
Quoted 7 lines · Saravana Kannan
Sibi might be working on doing that for the SDM845 CPUfreq driver. Georgi could also change his GPU driver use case to use this BW OPP table and required-opps.
The problem is that people don't want to start using this until we decide on the DT representation. So it's like a chicken and egg situation.
Yeah, I agree to that.
@Georgi and @Sibi: This is your chance to speak up about the proposal from Saravana and if you find anything wrong with them. And specially that it is mostly about interconnects here, I would like to have an explicit Ack from Georgi on this.
And if you guys are all okay about this then please at least commit that you will convert your stuff based on this in coming days.
Signature · 1 line
viresh
Hey Viresh,
Quoted 11 lines · Viresh Kumar
Quoted 7 lines · Saravana Kannan
Sibi might be working on doing that for the SDM845 CPUfreq driver. Georgi could also change his GPU driver use case to use this BW OPP table and required-opps.
The problem is that people don't want to start using this until we decide on the DT representation. So it's like a chicken and egg situation.
Yeah, I agree to that.
Quoted 7 lines
@Georgi and @Sibi: This is your chance to speak up about the proposal from Saravana and if you find anything wrong with them. And specially that it is mostly about interconnects here, I would like to have an explicit Ack from Georgi on this.
And if you guys are all okay about this then please at least commit that you will convert your stuff based on this in coming days.
I've been using both Saravana's and Georgi's series for a while now to scale DDR and L3 on SDM845. There is currently no consensus as to where the votes are to be actuated from, hence couldn't post anything out.
DCVS based on Saravana's series + passive governor:
https://github.com/QuinAsura/linux/tree/lnext-072619-SK-series
DCVS based on Georgi's series: (I had already posted this out)
https://github.com/QuinAsura/linux/tree/lnext-072619-GJ-series
Signature · 3 lines
Qualcomm Innovation Center, Inc. Qualcomm Innovation Center, Inc, is a member of Code Aurora Forum, a Linux Foundation Collaborative Project
Quoted 23 lines · Sibi Sankar
Hey Viresh,
Quoted 11 lines · Viresh Kumar
Quoted 7 lines · Saravana Kannan
Sibi might be working on doing that for the SDM845 CPUfreq driver. Georgi could also change his GPU driver use case to use this BW OPP table and required-opps.
The problem is that people don't want to start using this until we decide on the DT representation. So it's like a chicken and egg situation.
Yeah, I agree to that.
Quoted 7 lines
@Georgi and @Sibi: This is your chance to speak up about the proposal from Saravana and if you find anything wrong with them. And specially that it is mostly about interconnects here, I would like to have an explicit Ack from Georgi on this.
And if you guys are all okay about this then please at least commit that you will convert your stuff based on this in coming days.
Quoted 7 lines
I've been using both Saravana's and Georgi's series for a while now to scale DDR and L3 on SDM845. There is currently no consensus as to where the votes are to be actuated from, hence couldn't post anything out.
DCVS based on Saravana's series + passive governor:
https://github.com/QuinAsura/linux/tree/lnext-072619-SK-series
Thanks Sibi! You might want to convert your patches so that until the passive governor is ready, you just look up the required opps and vote for BW directly from the cpufreq driver. Once devfreq governor is ready, you can switch to it.
-Saravana
Quoted 8 lines
DCVS based on Georgi's series: (I had already posted this out)
https://github.com/QuinAsura/linux/tree/lnext-072619-GJ-series--
Qualcomm Innovation Center, Inc.
Qualcomm Innovation Center, Inc, is a member of Code Aurora Forum, a Linux Foundation Collaborative Project
Quoted 29 lines · Saravana Kannan
Quoted 28 lines · Sibi Sankar
Hey Viresh,
Quoted 11 lines · Viresh Kumar
Quoted 7 lines · Saravana Kannan
Sibi might be working on doing that for the SDM845 CPUfreq driver. Georgi could also change his GPU driver use case to use this BW OPP table and required-opps.
The problem is that people don't want to start using this until we decide on the DT representation. So it's like a chicken and egg situation.
Yeah, I agree to that.
Quoted 7 lines
@Georgi and @Sibi: This is your chance to speak up about the proposal from Saravana and if you find anything wrong with them. And specially that it is mostly about interconnects here, I would like to have an explicit Ack from Georgi on this.
And if you guys are all okay about this then please at least commit that you will convert your stuff based on this in coming days.
I've been using both Saravana's and Georgi's series for a while now to scale DDR and L3 on SDM845. There is currently no consensus as to where the votes are to be actuated from, hence couldn't post anything out.
Quoted 7 lines
Quoted 2 lines
DCVS based on Saravana's series + passive governor:
https://github.com/QuinAsura/linux/tree/lnext-072619-SK-seriesThanks Sibi! You might want to convert your patches so that until the passive governor is ready, you just look up the required opps and vote for BW directly from the cpufreq driver. Once devfreq governor is ready, you can switch to it.
Sure I'll do that.
Quoted 11 lines
-Saravana
Quoted 8 lines
DCVS based on Georgi's series: (I had already posted this out)
https://github.com/QuinAsura/linux/tree/lnext-072619-GJ-series--
Qualcomm Innovation Center, Inc.
Qualcomm Innovation Center, Inc, is a member of Code Aurora Forum, a Linux Foundation Collaborative Project
Signature · 3 lines
Qualcomm Innovation Center, Inc. Qualcomm Innovation Center, Inc, is a member of Code Aurora Forum, a Linux Foundation Collaborative Project
Quoted 11 lines · Viresh Kumar
Quoted 7 lines · Saravana Kannan
Sibi might be working on doing that for the SDM845 CPUfreq driver. Georgi could also change his GPU driver use case to use this BW OPP table and required-opps.
The problem is that people don't want to start using this until we decide on the DT representation. So it's like a chicken and egg situation.
Yeah, I agree to that.
Quoted 7 lines
@Georgi and @Sibi: This is your chance to speak up about the proposal from Saravana and if you find anything wrong with them. And specially that it is mostly about interconnects here, I would like to have an explicit Ack from Georgi on this.
And if you guys are all okay about this then please at least commit that you will convert your stuff based on this in coming days.
Looks fine to me. I am already doing some testing with this patchset. However, as Stephen already pointed out, we should s/KBps/kBps/.
Thanks,
Georgi