* [PATCH 02/12] tools/power turbostat: update PERF_LIMIT_REASONS decoding
2015-04-19 0:40 ` [PATCH 01/12] tools/power turbostat: simplify default output Len Brown
@ 2015-04-19 0:40 ` Len Brown
2015-04-19 0:40 ` [PATCH 03/12] tools/power turbostat: label base frequency Len Brown
` (10 subsequent siblings)
11 siblings, 0 replies; 16+ messages in thread
From: Len Brown @ 2015-04-19 0:40 UTC (permalink / raw)
To: linux-pm; +Cc: Len Brown
From: Len Brown <len.brown@intel.com>
cosmetic only.
order the decoding of MSR_PERF_LIMIT_REASONS bits
from MSB to LSB -- which you notice when more than 1 bit is set
and you are, say, comparing the output to the documentation...
Signed-off-by: Len Brown <len.brown@intel.com>
---
tools/power/x86/turbostat/turbostat.c | 52 +++++++++++++++++------------------
1 file changed, 26 insertions(+), 26 deletions(-)
diff --git a/tools/power/x86/turbostat/turbostat.c b/tools/power/x86/turbostat/turbostat.c
index 1bc6e6c..50ef857 100644
--- a/tools/power/x86/turbostat/turbostat.c
+++ b/tools/power/x86/turbostat/turbostat.c
@@ -1700,35 +1700,35 @@ int print_perf_limit(struct thread_data *t, struct core_data *c, struct pkg_data
get_msr(cpu, MSR_CORE_PERF_LIMIT_REASONS, &msr);
fprintf(stderr, "cpu%d: MSR_CORE_PERF_LIMIT_REASONS, 0x%08llx", cpu, msr);
fprintf(stderr, " (Active: %s%s%s%s%s%s%s%s%s%s%s%s%s%s)",
- (msr & 1 << 0) ? "PROCHOT, " : "",
- (msr & 1 << 1) ? "ThermStatus, " : "",
- (msr & 1 << 2) ? "bit2, " : "",
- (msr & 1 << 4) ? "Graphics, " : "",
- (msr & 1 << 5) ? "Auto-HWP, " : "",
- (msr & 1 << 6) ? "VR-Therm, " : "",
- (msr & 1 << 8) ? "Amps, " : "",
- (msr & 1 << 9) ? "CorePwr, " : "",
- (msr & 1 << 10) ? "PkgPwrL1, " : "",
- (msr & 1 << 11) ? "PkgPwrL2, " : "",
- (msr & 1 << 12) ? "MultiCoreTurbo, " : "",
- (msr & 1 << 13) ? "Transitions, " : "",
+ (msr & 1 << 15) ? "bit15, " : "",
(msr & 1 << 14) ? "bit14, " : "",
- (msr & 1 << 15) ? "bit15, " : "");
+ (msr & 1 << 13) ? "Transitions, " : "",
+ (msr & 1 << 12) ? "MultiCoreTurbo, " : "",
+ (msr & 1 << 11) ? "PkgPwrL2, " : "",
+ (msr & 1 << 10) ? "PkgPwrL1, " : "",
+ (msr & 1 << 9) ? "CorePwr, " : "",
+ (msr & 1 << 8) ? "Amps, " : "",
+ (msr & 1 << 6) ? "VR-Therm, " : "",
+ (msr & 1 << 5) ? "Auto-HWP, " : "",
+ (msr & 1 << 4) ? "Graphics, " : "",
+ (msr & 1 << 2) ? "bit2, " : "",
+ (msr & 1 << 1) ? "ThermStatus, " : "",
+ (msr & 1 << 0) ? "PROCHOT, " : "");
fprintf(stderr, " (Logged: %s%s%s%s%s%s%s%s%s%s%s%s%s%s)\n",
- (msr & 1 << 16) ? "PROCHOT, " : "",
- (msr & 1 << 17) ? "ThermStatus, " : "",
- (msr & 1 << 18) ? "bit18, " : "",
- (msr & 1 << 20) ? "Graphics, " : "",
- (msr & 1 << 21) ? "Auto-HWP, " : "",
- (msr & 1 << 22) ? "VR-Therm, " : "",
- (msr & 1 << 24) ? "Amps, " : "",
- (msr & 1 << 25) ? "CorePwr, " : "",
- (msr & 1 << 26) ? "PkgPwrL1, " : "",
- (msr & 1 << 27) ? "PkgPwrL2, " : "",
- (msr & 1 << 28) ? "MultiCoreTurbo, " : "",
- (msr & 1 << 29) ? "Transitions, " : "",
+ (msr & 1 << 31) ? "bit31, " : "",
(msr & 1 << 30) ? "bit30, " : "",
- (msr & 1 << 31) ? "bit31, " : "");
+ (msr & 1 << 29) ? "Transitions, " : "",
+ (msr & 1 << 28) ? "MultiCoreTurbo, " : "",
+ (msr & 1 << 27) ? "PkgPwrL2, " : "",
+ (msr & 1 << 26) ? "PkgPwrL1, " : "",
+ (msr & 1 << 25) ? "CorePwr, " : "",
+ (msr & 1 << 24) ? "Amps, " : "",
+ (msr & 1 << 22) ? "VR-Therm, " : "",
+ (msr & 1 << 21) ? "Auto-HWP, " : "",
+ (msr & 1 << 20) ? "Graphics, " : "",
+ (msr & 1 << 18) ? "bit18, " : "",
+ (msr & 1 << 17) ? "ThermStatus, " : "",
+ (msr & 1 << 16) ? "PROCHOT, " : "");
}
if (do_gfx_perf_limit_reasons) {
--
2.4.0.rc1
^ permalink raw reply related [flat|nested] 16+ messages in thread* [PATCH 03/12] tools/power turbostat: label base frequency
2015-04-19 0:40 ` [PATCH 01/12] tools/power turbostat: simplify default output Len Brown
2015-04-19 0:40 ` [PATCH 02/12] tools/power turbostat: update PERF_LIMIT_REASONS decoding Len Brown
@ 2015-04-19 0:40 ` Len Brown
2015-04-19 0:40 ` [PATCH 04/12] x86 msr-index: define MSR_TURBO_RATIO_LIMIT,1,2 Len Brown
` (9 subsequent siblings)
11 siblings, 0 replies; 16+ messages in thread
From: Len Brown @ 2015-04-19 0:40 UTC (permalink / raw)
To: linux-pm; +Cc: Len Brown
From: Len Brown <len.brown@intel.com>
syntax only.
The cool kids are now using the phrase "base frequency",
where in the past we used "max non-turbo frequency" or "TSC frequency".
This distinction becomes important when a processor has a TSC
that runs at a different speed than the "base frequency".
Signed-off-by: Len Brown <len.brown@intel.com>
---
tools/power/x86/turbostat/turbostat.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/tools/power/x86/turbostat/turbostat.c b/tools/power/x86/turbostat/turbostat.c
index 50ef857..9bf6ca9 100644
--- a/tools/power/x86/turbostat/turbostat.c
+++ b/tools/power/x86/turbostat/turbostat.c
@@ -1091,11 +1091,11 @@ void dump_system_config_info(void)
fprintf(stderr, "cpu0: MSR_NHM_PLATFORM_INFO: 0x%08llx\n", msr);
ratio = (msr >> 40) & 0xFF;
- fprintf(stderr, "%d * %.0f = %.0f MHz max efficiency\n",
+ fprintf(stderr, "%d * %.0f = %.0f MHz max efficiency frequency\n",
ratio, bclk, ratio * bclk);
ratio = (msr >> 8) & 0xFF;
- fprintf(stderr, "%d * %.0f = %.0f MHz TSC frequency\n",
+ fprintf(stderr, "%d * %.0f = %.0f MHz base frequency\n",
ratio, bclk, ratio * bclk);
get_msr(0, MSR_IA32_POWER_CTL, &msr);
--
2.4.0.rc1
^ permalink raw reply related [flat|nested] 16+ messages in thread* [PATCH 04/12] x86 msr-index: define MSR_TURBO_RATIO_LIMIT,1,2
2015-04-19 0:40 ` [PATCH 01/12] tools/power turbostat: simplify default output Len Brown
2015-04-19 0:40 ` [PATCH 02/12] tools/power turbostat: update PERF_LIMIT_REASONS decoding Len Brown
2015-04-19 0:40 ` [PATCH 03/12] tools/power turbostat: label base frequency Len Brown
@ 2015-04-19 0:40 ` Len Brown
2015-04-19 0:40 ` [PATCH 05/12] tools/power turbostat: use new MSR_TURBO_RATIO_LIMIT names Len Brown
` (8 subsequent siblings)
11 siblings, 0 replies; 16+ messages in thread
From: Len Brown @ 2015-04-19 0:40 UTC (permalink / raw)
To: linux-pm; +Cc: Len Brown, x86
From: Len Brown <len.brown@intel.com>
MSR_TURBO_RATIO_LIMIT has grown into a set of three registers.
Add the documented names for them, in preparation
for deleting the previous ad-hoc names:
+#define MSR_TURBO_RATIO_LIMIT 0x000001ad
+#define MSR_TURBO_RATIO_LIMIT1 0x000001ae
+#define MSR_TURBO_RATIO_LIMIT2 0x000001af
Signed-off-by: Len Brown <len.brown@intel.com>
Cc: x86@kernel.org
---
arch/x86/include/uapi/asm/msr-index.h | 3 +++
1 file changed, 3 insertions(+)
diff --git a/arch/x86/include/uapi/asm/msr-index.h b/arch/x86/include/uapi/asm/msr-index.h
index 3ce0791..c4c7527 100644
--- a/arch/x86/include/uapi/asm/msr-index.h
+++ b/arch/x86/include/uapi/asm/msr-index.h
@@ -61,6 +61,9 @@
#define MSR_OFFCORE_RSP_1 0x000001a7
#define MSR_NHM_TURBO_RATIO_LIMIT 0x000001ad
#define MSR_IVT_TURBO_RATIO_LIMIT 0x000001ae
+#define MSR_TURBO_RATIO_LIMIT 0x000001ad
+#define MSR_TURBO_RATIO_LIMIT1 0x000001ae
+#define MSR_TURBO_RATIO_LIMIT2 0x000001af
#define MSR_LBR_SELECT 0x000001c8
#define MSR_LBR_TOS 0x000001c9
--
2.4.0.rc1
^ permalink raw reply related [flat|nested] 16+ messages in thread* [PATCH 05/12] tools/power turbostat: use new MSR_TURBO_RATIO_LIMIT names
2015-04-19 0:40 ` [PATCH 01/12] tools/power turbostat: simplify default output Len Brown
` (2 preceding siblings ...)
2015-04-19 0:40 ` [PATCH 04/12] x86 msr-index: define MSR_TURBO_RATIO_LIMIT,1,2 Len Brown
@ 2015-04-19 0:40 ` Len Brown
2015-04-19 0:40 ` [PATCH 06/12] tools/power turbostat: dump MSR_TURBO_RATIO_LIMIT2 Len Brown
` (7 subsequent siblings)
11 siblings, 0 replies; 16+ messages in thread
From: Len Brown @ 2015-04-19 0:40 UTC (permalink / raw)
To: linux-pm; +Cc: Len Brown
From: Len Brown <len.brown@intel.com>
s/MSR_NHM_TURBO_RATIO_LIMIT/MSR_TURBO_RATIO_LIMIT/
s/MSR_IVT_TURBO_RATIO_LIMIT/MSR_TURBO_RATIO_LIMIT1/
syntax only -- use the documented strings describing these registers.
Signed-off-by: Len Brown <len.brown@intel.com>
---
tools/power/x86/turbostat/turbostat.c | 8 ++++----
1 file changed, 4 insertions(+), 4 deletions(-)
diff --git a/tools/power/x86/turbostat/turbostat.c b/tools/power/x86/turbostat/turbostat.c
index 9bf6ca9..b699a0e 100644
--- a/tools/power/x86/turbostat/turbostat.c
+++ b/tools/power/x86/turbostat/turbostat.c
@@ -1105,9 +1105,9 @@ void dump_system_config_info(void)
if (!do_ivt_turbo_ratio_limit)
goto print_nhm_turbo_ratio_limits;
- get_msr(0, MSR_IVT_TURBO_RATIO_LIMIT, &msr);
+ get_msr(0, MSR_TURBO_RATIO_LIMIT1, &msr);
- fprintf(stderr, "cpu0: MSR_IVT_TURBO_RATIO_LIMIT: 0x%08llx\n", msr);
+ fprintf(stderr, "cpu0: MSR_TURBO_RATIO_LIMIT1: 0x%08llx\n", msr);
ratio = (msr >> 56) & 0xFF;
if (ratio)
@@ -1169,9 +1169,9 @@ print_nhm_turbo_ratio_limits:
if (!do_nhm_turbo_ratio_limit)
return;
- get_msr(0, MSR_NHM_TURBO_RATIO_LIMIT, &msr);
+ get_msr(0, MSR_TURBO_RATIO_LIMIT, &msr);
- fprintf(stderr, "cpu0: MSR_NHM_TURBO_RATIO_LIMIT: 0x%08llx\n", msr);
+ fprintf(stderr, "cpu0: MSR_TURBO_RATIO_LIMIT: 0x%08llx\n", msr);
ratio = (msr >> 56) & 0xFF;
if (ratio)
--
2.4.0.rc1
^ permalink raw reply related [flat|nested] 16+ messages in thread* [PATCH 06/12] tools/power turbostat: dump MSR_TURBO_RATIO_LIMIT2
2015-04-19 0:40 ` [PATCH 01/12] tools/power turbostat: simplify default output Len Brown
` (3 preceding siblings ...)
2015-04-19 0:40 ` [PATCH 05/12] tools/power turbostat: use new MSR_TURBO_RATIO_LIMIT names Len Brown
@ 2015-04-19 0:40 ` Len Brown
2015-04-19 0:40 ` [PATCH 07/12] tools/power turbostat: modprobe msr, if needed Len Brown
` (6 subsequent siblings)
11 siblings, 0 replies; 16+ messages in thread
From: Len Brown @ 2015-04-19 0:40 UTC (permalink / raw)
To: linux-pm; +Cc: Len Brown
From: Len Brown <len.brown@intel.com>
and up to 18 cores of turbo ratio limit
when using the turbostat --debug option.
Signed-off-by: Len Brown <len.brown@intel.com>
---
tools/power/x86/turbostat/turbostat.c | 141 +++++++++++++++++++++++++---------
1 file changed, 105 insertions(+), 36 deletions(-)
diff --git a/tools/power/x86/turbostat/turbostat.c b/tools/power/x86/turbostat/turbostat.c
index b699a0e..f05a3f7 100644
--- a/tools/power/x86/turbostat/turbostat.c
+++ b/tools/power/x86/turbostat/turbostat.c
@@ -65,8 +65,6 @@ unsigned int units = 1000000; /* MHz etc */
unsigned int genuine_intel;
unsigned int has_invariant_tsc;
unsigned int do_nhm_platform_info;
-unsigned int do_nhm_turbo_ratio_limit;
-unsigned int do_ivt_turbo_ratio_limit;
unsigned int extra_msr_offset32;
unsigned int extra_msr_offset64;
unsigned int extra_delta_offset32;
@@ -1078,14 +1076,12 @@ int slv_pkg_cstate_limits[8] = {PCL__0, PCL__1, PCLRSV, PCLRSV, PCL__4, PCLRSV,
int amt_pkg_cstate_limits[8] = {PCL__0, PCL__1, PCL__2, PCLRSV, PCLRSV, PCLRSV, PCL__6, PCL__7};
int phi_pkg_cstate_limits[8] = {PCL__0, PCL__2, PCL_6N, PCL_6R, PCLRSV, PCLRSV, PCLRSV, PCLUNL};
-void dump_system_config_info(void)
+static void
+dump_nhm_platform_info(void)
{
unsigned long long msr;
unsigned int ratio;
- if (!do_nhm_platform_info)
- return;
-
get_msr(0, MSR_NHM_PLATFORM_INFO, &msr);
fprintf(stderr, "cpu0: MSR_NHM_PLATFORM_INFO: 0x%08llx\n", msr);
@@ -1102,8 +1098,36 @@ void dump_system_config_info(void)
fprintf(stderr, "cpu0: MSR_IA32_POWER_CTL: 0x%08llx (C1E auto-promotion: %sabled)\n",
msr, msr & 0x2 ? "EN" : "DIS");
- if (!do_ivt_turbo_ratio_limit)
- goto print_nhm_turbo_ratio_limits;
+ return;
+}
+
+static void
+dump_hsw_turbo_ratio_limits(void)
+{
+ unsigned long long msr;
+ unsigned int ratio;
+
+ get_msr(0, MSR_TURBO_RATIO_LIMIT2, &msr);
+
+ fprintf(stderr, "cpu0: MSR_TURBO_RATIO_LIMIT2: 0x%08llx\n", msr);
+
+ ratio = (msr >> 8) & 0xFF;
+ if (ratio)
+ fprintf(stderr, "%d * %.0f = %.0f MHz max turbo 18 active cores\n",
+ ratio, bclk, ratio * bclk);
+
+ ratio = (msr >> 0) & 0xFF;
+ if (ratio)
+ fprintf(stderr, "%d * %.0f = %.0f MHz max turbo 17 active cores\n",
+ ratio, bclk, ratio * bclk);
+ return;
+}
+
+static void
+dump_ivt_turbo_ratio_limits(void)
+{
+ unsigned long long msr;
+ unsigned int ratio;
get_msr(0, MSR_TURBO_RATIO_LIMIT1, &msr);
@@ -1148,26 +1172,14 @@ void dump_system_config_info(void)
if (ratio)
fprintf(stderr, "%d * %.0f = %.0f MHz max turbo 9 active cores\n",
ratio, bclk, ratio * bclk);
+ return;
+}
-print_nhm_turbo_ratio_limits:
- get_msr(0, MSR_NHM_SNB_PKG_CST_CFG_CTL, &msr);
-
-#define SNB_C1_AUTO_UNDEMOTE (1UL << 27)
-#define SNB_C3_AUTO_UNDEMOTE (1UL << 28)
-
- fprintf(stderr, "cpu0: MSR_NHM_SNB_PKG_CST_CFG_CTL: 0x%08llx", msr);
-
- fprintf(stderr, " (%s%s%s%s%slocked: pkg-cstate-limit=%d: %s)\n",
- (msr & SNB_C3_AUTO_UNDEMOTE) ? "UNdemote-C3, " : "",
- (msr & SNB_C1_AUTO_UNDEMOTE) ? "UNdemote-C1, " : "",
- (msr & NHM_C3_AUTO_DEMOTE) ? "demote-C3, " : "",
- (msr & NHM_C1_AUTO_DEMOTE) ? "demote-C1, " : "",
- (msr & (1 << 15)) ? "" : "UN",
- (unsigned int)msr & 7,
- pkg_cstate_limit_strings[pkg_cstate_limit]);
-
- if (!do_nhm_turbo_ratio_limit)
- return;
+static void
+dump_nhm_turbo_ratio_limits(void)
+{
+ unsigned long long msr;
+ unsigned int ratio;
get_msr(0, MSR_TURBO_RATIO_LIMIT, &msr);
@@ -1212,7 +1224,30 @@ print_nhm_turbo_ratio_limits:
if (ratio)
fprintf(stderr, "%d * %.0f = %.0f MHz max turbo 1 active cores\n",
ratio, bclk, ratio * bclk);
+ return;
+}
+static void
+dump_nhm_cst_cfg(void)
+{
+ unsigned long long msr;
+
+ get_msr(0, MSR_NHM_SNB_PKG_CST_CFG_CTL, &msr);
+
+#define SNB_C1_AUTO_UNDEMOTE (1UL << 27)
+#define SNB_C3_AUTO_UNDEMOTE (1UL << 28)
+
+ fprintf(stderr, "cpu0: MSR_NHM_SNB_PKG_CST_CFG_CTL: 0x%08llx", msr);
+
+ fprintf(stderr, " (%s%s%s%s%slocked: pkg-cstate-limit=%d: %s)\n",
+ (msr & SNB_C3_AUTO_UNDEMOTE) ? "UNdemote-C3, " : "",
+ (msr & SNB_C1_AUTO_UNDEMOTE) ? "UNdemote-C1, " : "",
+ (msr & NHM_C3_AUTO_DEMOTE) ? "demote-C3, " : "",
+ (msr & NHM_C1_AUTO_DEMOTE) ? "demote-C1, " : "",
+ (msr & (1 << 15)) ? "" : "UN",
+ (unsigned int)msr & 7,
+ pkg_cstate_limit_strings[pkg_cstate_limit]);
+ return;
}
void free_all_buffers(void)
@@ -1625,12 +1660,49 @@ int has_ivt_turbo_ratio_limit(unsigned int family, unsigned int model)
switch (model) {
case 0x3E: /* IVB Xeon */
+ case 0x3F: /* HSW Xeon */
+ return 1;
+ default:
+ return 0;
+ }
+}
+int has_hsw_turbo_ratio_limit(unsigned int family, unsigned int model)
+{
+ if (!genuine_intel)
+ return 0;
+
+ if (family != 6)
+ return 0;
+
+ switch (model) {
+ case 0x3F: /* HSW Xeon */
return 1;
default:
return 0;
}
}
+static void
+dump_cstate_pstate_config_info(family, model)
+{
+ if (!do_nhm_platform_info)
+ return;
+
+ dump_nhm_platform_info();
+
+ if (has_hsw_turbo_ratio_limit(family, model))
+ dump_hsw_turbo_ratio_limits();
+
+ if (has_ivt_turbo_ratio_limit(family, model))
+ dump_ivt_turbo_ratio_limits();
+
+ if (has_nhm_turbo_ratio_limit(family, model))
+ dump_nhm_turbo_ratio_limits();
+
+ dump_nhm_cst_cfg();
+}
+
+
/*
* print_epb()
* Decode the ENERGY_PERF_BIAS MSR
@@ -2238,7 +2310,7 @@ guess:
return 0;
}
-void check_cpuid()
+void process_cpuid()
{
unsigned int eax, ebx, ecx, edx, max_level;
unsigned int fms, family, model, stepping;
@@ -2314,15 +2386,15 @@ void check_cpuid()
do_slm_cstates = is_slm(family, model);
bclk = discover_bclk(family, model);
- do_nhm_turbo_ratio_limit = do_nhm_platform_info && has_nhm_turbo_ratio_limit(family, model);
- do_ivt_turbo_ratio_limit = has_ivt_turbo_ratio_limit(family, model);
rapl_probe(family, model);
perf_limit_reasons_probe(family, model);
+ if (debug)
+ dump_cstate_pstate_config_info();
+
return;
}
-
void help()
{
fprintf(stderr,
@@ -2560,14 +2632,11 @@ void turbostat_init()
{
check_dev_msr();
check_permissions();
- check_cpuid();
+ process_cpuid();
setup_all_buffers();
if (debug)
- dump_system_config_info();
-
- if (debug)
for_all_cpus(print_epb, ODD_COUNTERS);
if (debug)
@@ -2644,7 +2713,7 @@ int get_and_dump_counters(void)
}
void print_version() {
- fprintf(stderr, "turbostat version 4.1 10-Feb, 2015"
+ fprintf(stderr, "turbostat version 4.2 23 Mar, 2015"
" - Len Brown <lenb@kernel.org>\n");
}
--
2.4.0.rc1
^ permalink raw reply related [flat|nested] 16+ messages in thread* [PATCH 07/12] tools/power turbostat: modprobe msr, if needed
2015-04-19 0:40 ` [PATCH 01/12] tools/power turbostat: simplify default output Len Brown
` (4 preceding siblings ...)
2015-04-19 0:40 ` [PATCH 06/12] tools/power turbostat: dump MSR_TURBO_RATIO_LIMIT2 Len Brown
@ 2015-04-19 0:40 ` Len Brown
2015-04-19 14:29 ` Doug Smythies
2015-04-19 0:40 ` [PATCH 08/12] tools/power turbostat: Use $(CURDIR) instead of $(PWD) and add support for O= option in Makefile Len Brown
` (5 subsequent siblings)
11 siblings, 1 reply; 16+ messages in thread
From: Len Brown @ 2015-04-19 0:40 UTC (permalink / raw)
To: linux-pm; +Cc: Len Brown
From: Len Brown <len.brown@intel.com>
Some distros (Ubuntu) ship the msr driver as a module.
If turbosat is run as root on those systems, and discovers
that there is no /dev/cpu/cpu0/msr, it will now "modprobe msr"
for the user.
If not root, the modprobe attempt will fail, and turbostat will exit as before:
turbostat: no /dev/cpu/0/msr, Try "# modprobe msr" : No such file or directory
Signed-off-by: Len Brown <len.brown@intel.com>
---
tools/power/x86/turbostat/turbostat.c | 5 +++--
1 file changed, 3 insertions(+), 2 deletions(-)
diff --git a/tools/power/x86/turbostat/turbostat.c b/tools/power/x86/turbostat/turbostat.c
index f05a3f7..50341a3 100644
--- a/tools/power/x86/turbostat/turbostat.c
+++ b/tools/power/x86/turbostat/turbostat.c
@@ -1528,7 +1528,8 @@ void check_dev_msr()
struct stat sb;
if (stat("/dev/cpu/0/msr", &sb))
- err(-5, "no /dev/cpu/0/msr, Try \"# modprobe msr\" ");
+ if (system("/sbin/modprobe msr > /dev/null 2>&1"))
+ err(-5, "no /dev/cpu/0/msr, Try \"# modprobe msr\" ");
}
void check_permissions()
@@ -2713,7 +2714,7 @@ int get_and_dump_counters(void)
}
void print_version() {
- fprintf(stderr, "turbostat version 4.2 23 Mar, 2015"
+ fprintf(stderr, "turbostat version 4.3 24 Mar, 2015"
" - Len Brown <lenb@kernel.org>\n");
}
--
2.4.0.rc1
^ permalink raw reply related [flat|nested] 16+ messages in thread* [PATCH 08/12] tools/power turbostat: Use $(CURDIR) instead of $(PWD) and add support for O= option in Makefile
2015-04-19 0:40 ` [PATCH 01/12] tools/power turbostat: simplify default output Len Brown
` (5 preceding siblings ...)
2015-04-19 0:40 ` [PATCH 07/12] tools/power turbostat: modprobe msr, if needed Len Brown
@ 2015-04-19 0:40 ` Len Brown
2015-04-19 0:40 ` [PATCH 09/12] tools/power turbostat: Initial Skylake support Len Brown
` (4 subsequent siblings)
11 siblings, 0 replies; 16+ messages in thread
From: Len Brown @ 2015-04-19 0:40 UTC (permalink / raw)
To: linux-pm; +Cc: Thomas D, Mark Asselstine, Len Brown
From: Thomas D <whissi@whissi.de>
Since commit ee0778a30153
("tools/power: turbostat: make Makefile a bit more capable")
turbostat's Makefile is using
[...]
BUILD_OUTPUT := $(PWD)
[...]
which obviously causes trouble when building "turbostat" with
make -C /usr/src/linux/tools/power/x86/turbostat ARCH=x86 turbostat
because GNU make does not update nor guarantee that $PWD is set.
This patch changes the Makefile to use $CURDIR instead, which GNU make
guarantees to set and update (i.e. when using "make -C ...") and also
adds support for the O= option (see "make help" in your root of your
kernel source tree for more details).
Link: https://bugs.gentoo.org/show_bug.cgi?id=533918
Fixes: ee0778a30153 ("tools/power: turbostat: make Makefile a bit more capable")
Signed-off-by: Thomas D. <whissi@whissi.de>
Cc: Mark Asselstine <mark.asselstine@windriver.com>
Signed-off-by: Len Brown <len.brown@intel.com>
---
tools/power/x86/turbostat/Makefile | 6 +++++-
1 file changed, 5 insertions(+), 1 deletion(-)
diff --git a/tools/power/x86/turbostat/Makefile b/tools/power/x86/turbostat/Makefile
index d1b3a36..4039854 100644
--- a/tools/power/x86/turbostat/Makefile
+++ b/tools/power/x86/turbostat/Makefile
@@ -1,8 +1,12 @@
CC = $(CROSS_COMPILE)gcc
-BUILD_OUTPUT := $(PWD)
+BUILD_OUTPUT := $(CURDIR)
PREFIX := /usr
DESTDIR :=
+ifeq ("$(origin O)", "command line")
+ BUILD_OUTPUT := $(O)
+endif
+
turbostat : turbostat.c
CFLAGS += -Wall
CFLAGS += -DMSRHEADER='"../../../../arch/x86/include/uapi/asm/msr-index.h"'
--
2.4.0.rc1
^ permalink raw reply related [flat|nested] 16+ messages in thread* [PATCH 09/12] tools/power turbostat: Initial Skylake support
2015-04-19 0:40 ` [PATCH 01/12] tools/power turbostat: simplify default output Len Brown
` (6 preceding siblings ...)
2015-04-19 0:40 ` [PATCH 08/12] tools/power turbostat: Use $(CURDIR) instead of $(PWD) and add support for O= option in Makefile Len Brown
@ 2015-04-19 0:40 ` Len Brown
2015-04-19 0:40 ` [PATCH 10/12] tools/power turbostat: correct DRAM RAPL units on recent Xeon processors Len Brown
` (3 subsequent siblings)
11 siblings, 0 replies; 16+ messages in thread
From: Len Brown @ 2015-04-19 0:40 UTC (permalink / raw)
To: linux-pm; +Cc: Len Brown
From: Len Brown <len.brown@intel.com>
Skylake adds some additional residency counters.
Skylake supports a different mix of RAPL registers
from any previous product.
In most other ways, Skylake is like Broadwell.
Signed-off-by: Len Brown <len.brown@intel.com>
---
arch/x86/include/uapi/asm/msr-index.h | 5 ++
tools/power/x86/turbostat/turbostat.c | 124 ++++++++++++++++++++++++++++++----
2 files changed, 116 insertions(+), 13 deletions(-)
diff --git a/arch/x86/include/uapi/asm/msr-index.h b/arch/x86/include/uapi/asm/msr-index.h
index c4c7527..fe01b0a 100644
--- a/arch/x86/include/uapi/asm/msr-index.h
+++ b/arch/x86/include/uapi/asm/msr-index.h
@@ -150,6 +150,11 @@
#define MSR_PP1_ENERGY_STATUS 0x00000641
#define MSR_PP1_POLICY 0x00000642
+#define MSR_PKG_WEIGHTED_CORE_C0_RES 0x00000658
+#define MSR_PKG_ANY_CORE_C0_RES 0x00000659
+#define MSR_PKG_ANY_GFXE_C0_RES 0x0000065A
+#define MSR_PKG_BOTH_CORE_GFXE_C0_RES 0x0000065B
+
#define MSR_CORE_C1_RES 0x00000660
#define MSR_CC6_DEMOTION_POLICY_CONFIG 0x00000668
diff --git a/tools/power/x86/turbostat/turbostat.c b/tools/power/x86/turbostat/turbostat.c
index 50341a3..ad56889 100644
--- a/tools/power/x86/turbostat/turbostat.c
+++ b/tools/power/x86/turbostat/turbostat.c
@@ -57,6 +57,7 @@ unsigned int do_pc3;
unsigned int do_pc6;
unsigned int do_pc7;
unsigned int do_c8_c9_c10;
+unsigned int do_skl_residency;
unsigned int do_slm_cstates;
unsigned int use_c1_residency_msr;
unsigned int has_aperf;
@@ -99,18 +100,18 @@ unsigned int do_ring_perf_limit_reasons;
#define RAPL_DRAM (1 << 3)
/* 0x618 MSR_DRAM_POWER_LIMIT */
/* 0x619 MSR_DRAM_ENERGY_STATUS */
- /* 0x61c MSR_DRAM_POWER_INFO */
#define RAPL_DRAM_PERF_STATUS (1 << 4)
/* 0x61b MSR_DRAM_PERF_STATUS */
+#define RAPL_DRAM_POWER_INFO (1 << 5)
+ /* 0x61c MSR_DRAM_POWER_INFO */
-#define RAPL_CORES (1 << 5)
+#define RAPL_CORES (1 << 6)
/* 0x638 MSR_PP0_POWER_LIMIT */
/* 0x639 MSR_PP0_ENERGY_STATUS */
-#define RAPL_CORE_POLICY (1 << 6)
+#define RAPL_CORE_POLICY (1 << 7)
/* 0x63a MSR_PP0_POLICY */
-
-#define RAPL_GFX (1 << 7)
+#define RAPL_GFX (1 << 8)
/* 0x640 MSR_PP1_POWER_LIMIT */
/* 0x641 MSR_PP1_ENERGY_STATUS */
/* 0x642 MSR_PP1_POLICY */
@@ -157,6 +158,10 @@ struct pkg_data {
unsigned long long pc8;
unsigned long long pc9;
unsigned long long pc10;
+ unsigned long long pkg_wtd_core_c0;
+ unsigned long long pkg_any_core_c0;
+ unsigned long long pkg_any_gfxe_c0;
+ unsigned long long pkg_both_core_gfxe_c0;
unsigned int package_id;
unsigned int energy_pkg; /* MSR_PKG_ENERGY_STATUS */
unsigned int energy_dram; /* MSR_DRAM_ENERGY_STATUS */
@@ -320,6 +325,13 @@ void print_header(void)
if (do_ptm)
outp += sprintf(outp, " PkgTmp");
+ if (do_skl_residency) {
+ outp += sprintf(outp, " Totl%%C0");
+ outp += sprintf(outp, " Any%%C0");
+ outp += sprintf(outp, " GFX%%C0");
+ outp += sprintf(outp, " CPUGFX%%");
+ }
+
if (do_pc2)
outp += sprintf(outp, " Pkg%%pc2");
if (do_pc3)
@@ -401,6 +413,12 @@ int dump_counters(struct thread_data *t, struct core_data *c,
if (p) {
outp += sprintf(outp, "package: %d\n", p->package_id);
+
+ outp += sprintf(outp, "Weighted cores: %016llX\n", p->pkg_wtd_core_c0);
+ outp += sprintf(outp, "Any cores: %016llX\n", p->pkg_any_core_c0);
+ outp += sprintf(outp, "Any GFX: %016llX\n", p->pkg_any_gfxe_c0);
+ outp += sprintf(outp, "CPU + GFX: %016llX\n", p->pkg_both_core_gfxe_c0);
+
outp += sprintf(outp, "pc2: %016llX\n", p->pc2);
if (do_pc3)
outp += sprintf(outp, "pc3: %016llX\n", p->pc3);
@@ -539,9 +557,18 @@ int format_counters(struct thread_data *t, struct core_data *c,
if (!(t->flags & CPU_IS_FIRST_CORE_IN_PACKAGE))
goto done;
+ /* PkgTmp */
if (do_ptm)
outp += sprintf(outp, "%8d", p->pkg_temp_c);
+ /* Totl%C0, Any%C0 GFX%C0 CPUGFX% */
+ if (do_skl_residency) {
+ outp += sprintf(outp, "%8.2f", 100.0 * p->pkg_wtd_core_c0/t->tsc);
+ outp += sprintf(outp, "%8.2f", 100.0 * p->pkg_any_core_c0/t->tsc);
+ outp += sprintf(outp, "%8.2f", 100.0 * p->pkg_any_gfxe_c0/t->tsc);
+ outp += sprintf(outp, "%8.2f", 100.0 * p->pkg_both_core_gfxe_c0/t->tsc);
+ }
+
if (do_pc2)
outp += sprintf(outp, "%8.2f", 100.0 * p->pc2/t->tsc);
if (do_pc3)
@@ -644,6 +671,13 @@ void format_all_counters(struct thread_data *t, struct core_data *c, struct pkg_
void
delta_package(struct pkg_data *new, struct pkg_data *old)
{
+
+ if (do_skl_residency) {
+ old->pkg_wtd_core_c0 = new->pkg_wtd_core_c0 - old->pkg_wtd_core_c0;
+ old->pkg_any_core_c0 = new->pkg_any_core_c0 - old->pkg_any_core_c0;
+ old->pkg_any_gfxe_c0 = new->pkg_any_gfxe_c0 - old->pkg_any_gfxe_c0;
+ old->pkg_both_core_gfxe_c0 = new->pkg_both_core_gfxe_c0 - old->pkg_both_core_gfxe_c0;
+ }
old->pc2 = new->pc2 - old->pc2;
if (do_pc3)
old->pc3 = new->pc3 - old->pc3;
@@ -790,6 +824,11 @@ void clear_counters(struct thread_data *t, struct core_data *c, struct pkg_data
c->c7 = 0;
c->core_temp_c = 0;
+ p->pkg_wtd_core_c0 = 0;
+ p->pkg_any_core_c0 = 0;
+ p->pkg_any_gfxe_c0 = 0;
+ p->pkg_both_core_gfxe_c0 = 0;
+
p->pc2 = 0;
if (do_pc3)
p->pc3 = 0;
@@ -834,6 +873,13 @@ int sum_counters(struct thread_data *t, struct core_data *c,
if (!(t->flags & CPU_IS_FIRST_CORE_IN_PACKAGE))
return 0;
+ if (do_skl_residency) {
+ average.packages.pkg_wtd_core_c0 += p->pkg_wtd_core_c0;
+ average.packages.pkg_any_core_c0 += p->pkg_any_core_c0;
+ average.packages.pkg_any_gfxe_c0 += p->pkg_any_gfxe_c0;
+ average.packages.pkg_both_core_gfxe_c0 += p->pkg_both_core_gfxe_c0;
+ }
+
average.packages.pc2 += p->pc2;
if (do_pc3)
average.packages.pc3 += p->pc3;
@@ -881,6 +927,13 @@ void compute_average(struct thread_data *t, struct core_data *c,
average.cores.c6 /= topo.num_cores;
average.cores.c7 /= topo.num_cores;
+ if (do_skl_residency) {
+ average.packages.pkg_wtd_core_c0 /= topo.num_packages;
+ average.packages.pkg_any_core_c0 /= topo.num_packages;
+ average.packages.pkg_any_gfxe_c0 /= topo.num_packages;
+ average.packages.pkg_both_core_gfxe_c0 /= topo.num_packages;
+ }
+
average.packages.pc2 /= topo.num_packages;
if (do_pc3)
average.packages.pc3 /= topo.num_packages;
@@ -987,6 +1040,16 @@ int get_counters(struct thread_data *t, struct core_data *c, struct pkg_data *p)
if (!(t->flags & CPU_IS_FIRST_CORE_IN_PACKAGE))
return 0;
+ if (do_skl_residency) {
+ if (get_msr(cpu, MSR_PKG_WEIGHTED_CORE_C0_RES, &p->pkg_wtd_core_c0))
+ return -10;
+ if (get_msr(cpu, MSR_PKG_ANY_CORE_C0_RES, &p->pkg_any_core_c0))
+ return -11;
+ if (get_msr(cpu, MSR_PKG_ANY_GFXE_C0_RES, &p->pkg_any_gfxe_c0))
+ return -12;
+ if (get_msr(cpu, MSR_PKG_BOTH_CORE_GFXE_C0_RES, &p->pkg_both_core_gfxe_c0))
+ return -13;
+ }
if (do_pc3)
if (get_msr(cpu, MSR_PKG_C3_RESIDENCY, &p->pc3))
return -9;
@@ -1063,15 +1126,18 @@ int get_counters(struct thread_data *t, struct core_data *c, struct pkg_data *p)
#define PCL_6R 9 /* PC6 Retention */
#define PCL__7 10 /* PC7 */
#define PCL_7S 11 /* PC7 Shrink */
-#define PCLUNL 12 /* Unlimited */
+#define PCL__8 12 /* PC8 */
+#define PCL__9 13 /* PC9 */
+#define PCLUNL 14 /* Unlimited */
int pkg_cstate_limit = PCLUKN;
char *pkg_cstate_limit_strings[] = { "reserved", "unknown", "pc0", "pc1", "pc2",
- "pc3", "pc4", "pc6", "pc6n", "pc6r", "pc7", "pc7s", "unlimited"};
+ "pc3", "pc4", "pc6", "pc6n", "pc6r", "pc7", "pc7s", "pc8", "pc9", "unlimited"};
int nhm_pkg_cstate_limits[8] = {PCL__0, PCL__1, PCL__3, PCL__6, PCL__7, PCLRSV, PCLRSV, PCLUNL};
int snb_pkg_cstate_limits[8] = {PCL__0, PCL__2, PCL_6N, PCL_6R, PCL__7, PCL_7S, PCLRSV, PCLUNL};
int hsw_pkg_cstate_limits[8] = {PCL__0, PCL__2, PCL__3, PCL__6, PCL__7, PCL_7S, PCLRSV, PCLUNL};
+int skl_pkg_cstate_limits[8] = {PCL__0, PCL__2, PCL__3, PCL__6, PCL__7, PCL_7S, PCL__8, PCL__9};
int slv_pkg_cstate_limits[8] = {PCL__0, PCL__1, PCLRSV, PCLRSV, PCL__4, PCLRSV, PCL__6, PCL__7};
int amt_pkg_cstate_limits[8] = {PCL__0, PCL__1, PCL__2, PCLRSV, PCLRSV, PCLRSV, PCL__6, PCL__7};
int phi_pkg_cstate_limits[8] = {PCL__0, PCL__2, PCL_6N, PCL_6R, PCLRSV, PCLRSV, PCLRSV, PCLUNL};
@@ -1619,6 +1685,8 @@ int probe_nhm_msrs(unsigned int family, unsigned int model)
case 0x47: /* BDW */
case 0x4F: /* BDX */
case 0x56: /* BDX-DE */
+ case 0x4E: /* SKL */
+ case 0x5E: /* SKL */
pkg_cstate_limits = hsw_pkg_cstate_limits;
break;
case 0x37: /* BYT */
@@ -1895,14 +1963,18 @@ void rapl_probe(unsigned int family, unsigned int model)
case 0x47: /* BDW */
do_rapl = RAPL_PKG | RAPL_CORES | RAPL_CORE_POLICY | RAPL_GFX | RAPL_PKG_POWER_INFO;
break;
+ case 0x4E: /* SKL */
+ case 0x5E: /* SKL */
+ do_rapl = RAPL_PKG | RAPL_DRAM | RAPL_DRAM_PERF_STATUS | RAPL_PKG_PERF_STATUS | RAPL_PKG_POWER_INFO;
+ break;
case 0x3F: /* HSX */
case 0x4F: /* BDX */
case 0x56: /* BDX-DE */
- do_rapl = RAPL_PKG | RAPL_DRAM | RAPL_DRAM_PERF_STATUS | RAPL_PKG_PERF_STATUS | RAPL_PKG_POWER_INFO;
+ do_rapl = RAPL_PKG | RAPL_DRAM | RAPL_DRAM_POWER_INFO | RAPL_DRAM_PERF_STATUS | RAPL_PKG_PERF_STATUS | RAPL_PKG_POWER_INFO;
break;
case 0x2D:
case 0x3E:
- do_rapl = RAPL_PKG | RAPL_CORES | RAPL_CORE_POLICY | RAPL_DRAM | RAPL_PKG_PERF_STATUS | RAPL_DRAM_PERF_STATUS | RAPL_PKG_POWER_INFO;
+ do_rapl = RAPL_PKG | RAPL_CORES | RAPL_CORE_POLICY | RAPL_DRAM | RAPL_DRAM_POWER_INFO | RAPL_PKG_PERF_STATUS | RAPL_DRAM_PERF_STATUS | RAPL_PKG_POWER_INFO;
break;
case 0x37: /* BYT */
case 0x4D: /* AVN */
@@ -2092,19 +2164,18 @@ int print_rapl(struct thread_data *t, struct core_data *c, struct pkg_data *p)
((msr >> 48) & 1) ? "EN" : "DIS");
}
- if (do_rapl & RAPL_DRAM) {
+ if (do_rapl & RAPL_DRAM_POWER_INFO) {
if (get_msr(cpu, MSR_DRAM_POWER_INFO, &msr))
return -6;
-
fprintf(stderr, "cpu%d: MSR_DRAM_POWER_INFO,: 0x%08llx (%.0f W TDP, RAPL %.0f - %.0f W, %f sec.)\n",
cpu, msr,
((msr >> 0) & RAPL_POWER_GRANULARITY) * rapl_power_units,
((msr >> 16) & RAPL_POWER_GRANULARITY) * rapl_power_units,
((msr >> 32) & RAPL_POWER_GRANULARITY) * rapl_power_units,
((msr >> 48) & RAPL_TIME_GRANULARITY) * rapl_time_units);
-
-
+ }
+ if (do_rapl & RAPL_DRAM) {
if (get_msr(cpu, MSR_DRAM_POWER_LIMIT, &msr))
return -9;
fprintf(stderr, "cpu%d: MSR_DRAM_POWER_LIMIT: 0x%08llx (%slocked)\n",
@@ -2173,6 +2244,8 @@ int has_snb_msrs(unsigned int family, unsigned int model)
case 0x47: /* BDW */
case 0x4F: /* BDX */
case 0x56: /* BDX-DE */
+ case 0x4E: /* SKL */
+ case 0x5E: /* SKL */
return 1;
}
return 0;
@@ -2193,12 +2266,36 @@ int has_hsw_msrs(unsigned int family, unsigned int model)
switch (model) {
case 0x45: /* HSW */
case 0x3D: /* BDW */
+ case 0x4E: /* SKL */
+ case 0x5E: /* SKL */
+ return 1;
+ }
+ return 0;
+}
+
+/*
+ * SKL adds support for additional MSRS:
+ *
+ * MSR_PKG_WEIGHTED_CORE_C0_RES 0x00000658
+ * MSR_PKG_ANY_CORE_C0_RES 0x00000659
+ * MSR_PKG_ANY_GFXE_C0_RES 0x0000065A
+ * MSR_PKG_BOTH_CORE_GFXE_C0_RES 0x0000065B
+ */
+int has_skl_msrs(unsigned int family, unsigned int model)
+{
+ if (!genuine_intel)
+ return 0;
+
+ switch (model) {
+ case 0x4E: /* SKL */
+ case 0x5E: /* SKL */
return 1;
}
return 0;
}
+
int is_slm(unsigned int family, unsigned int model)
{
if (!genuine_intel)
@@ -2384,6 +2481,7 @@ void process_cpuid()
do_pc6 = (pkg_cstate_limit >= PCL__6);
do_pc7 = do_snb_cstates && (pkg_cstate_limit >= PCL__7);
do_c8_c9_c10 = has_hsw_msrs(family, model);
+ do_skl_residency = has_skl_msrs(family, model);
do_slm_cstates = is_slm(family, model);
bclk = discover_bclk(family, model);
--
2.4.0.rc1
^ permalink raw reply related [flat|nested] 16+ messages in thread* [PATCH 10/12] tools/power turbostat: correct DRAM RAPL units on recent Xeon processors
2015-04-19 0:40 ` [PATCH 01/12] tools/power turbostat: simplify default output Len Brown
` (7 preceding siblings ...)
2015-04-19 0:40 ` [PATCH 09/12] tools/power turbostat: Initial Skylake support Len Brown
@ 2015-04-19 0:40 ` Len Brown
2015-04-19 0:40 ` [PATCH 11/12] tools/power turbostat: calculate TSC frequency from CPUID(0x15) on SKL Len Brown
` (2 subsequent siblings)
11 siblings, 0 replies; 16+ messages in thread
From: Len Brown @ 2015-04-19 0:40 UTC (permalink / raw)
To: linux-pm; +Cc: Andrey Semin, Len Brown
From: Andrey Semin <andrey.semin@intel.com>
While not yet documented in the Software Developer's Manual,
the data-sheet for modern Xeon states that DRAM RAPL ENERGY units
are fixed at 15.3 uJ, rather than being discovered via MSR.
Before this patch, DRAM energy on these products is over-stated by turbostat
because the RAPL units are 4x larger.
ref: "Xeon E5-2600 v3/E5-1600 v3 Datasheet Volume 2"
http://www.intel.com/content/dam/www/public/us/en/documents/datasheets/xeon-e5-v3-datasheet-vol-2.pdf
Signed-off-by: Andrey Semin <andrey.semin@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
---
tools/power/x86/turbostat/turbostat.c | 28 +++++++++++++++++++++++++---
1 file changed, 25 insertions(+), 3 deletions(-)
diff --git a/tools/power/x86/turbostat/turbostat.c b/tools/power/x86/turbostat/turbostat.c
index ad56889..064749d 100644
--- a/tools/power/x86/turbostat/turbostat.c
+++ b/tools/power/x86/turbostat/turbostat.c
@@ -83,7 +83,8 @@ unsigned int do_dts;
unsigned int do_ptm;
unsigned int tcc_activation_temp;
unsigned int tcc_activation_temp_override;
-double rapl_power_units, rapl_energy_units, rapl_time_units;
+double rapl_power_units, rapl_time_units;
+double rapl_dram_energy_units, rapl_energy_units;
double rapl_joule_counter_range;
unsigned int do_core_perf_limit_reasons;
unsigned int do_gfx_perf_limit_reasons;
@@ -600,7 +601,7 @@ int format_counters(struct thread_data *t, struct core_data *c,
if (do_rapl & RAPL_GFX)
outp += sprintf(outp, fmt8, p->energy_gfx * rapl_energy_units / interval_float);
if (do_rapl & RAPL_DRAM)
- outp += sprintf(outp, fmt8, p->energy_dram * rapl_energy_units / interval_float);
+ outp += sprintf(outp, fmt8, p->energy_dram * rapl_dram_energy_units / interval_float);
if (do_rapl & RAPL_PKG_PERF_STATUS)
outp += sprintf(outp, fmt8, 100.0 * p->rapl_pkg_perf_status * rapl_time_units / interval_float);
if (do_rapl & RAPL_DRAM_PERF_STATUS)
@@ -617,7 +618,7 @@ int format_counters(struct thread_data *t, struct core_data *c,
p->energy_gfx * rapl_energy_units);
if (do_rapl & RAPL_DRAM)
outp += sprintf(outp, fmt8,
- p->energy_dram * rapl_energy_units);
+ p->energy_dram * rapl_dram_energy_units);
if (do_rapl & RAPL_PKG_PERF_STATUS)
outp += sprintf(outp, fmt8, 100.0 * p->rapl_pkg_perf_status * rapl_time_units / interval_float);
if (do_rapl & RAPL_DRAM_PERF_STATUS)
@@ -1935,6 +1936,25 @@ double get_tdp(model)
}
}
+/*
+ * rapl_dram_energy_units_probe()
+ * Energy units are either hard-coded, or come from RAPL Energy Unit MSR.
+ */
+static double
+rapl_dram_energy_units_probe(int model, double rapl_energy_units)
+{
+ /* only called for genuine_intel, family 6 */
+
+ switch (model) {
+ case 0x3F: /* HSX */
+ case 0x4F: /* BDX */
+ case 0x56: /* BDX-DE */
+ return (rapl_dram_energy_units = 15.3 / 1000000);
+ default:
+ return (rapl_energy_units);
+ }
+}
+
/*
* rapl_probe()
@@ -1994,6 +2014,8 @@ void rapl_probe(unsigned int family, unsigned int model)
else
rapl_energy_units = 1.0 / (1 << (msr >> 8 & 0x1F));
+ rapl_dram_energy_units = rapl_dram_energy_units_probe(model, rapl_energy_units);
+
time_unit = msr >> 16 & 0xF;
if (time_unit == 0)
time_unit = 0xA;
--
2.4.0.rc1
^ permalink raw reply related [flat|nested] 16+ messages in thread* [PATCH 11/12] tools/power turbostat: calculate TSC frequency from CPUID(0x15) on SKL
2015-04-19 0:40 ` [PATCH 01/12] tools/power turbostat: simplify default output Len Brown
` (8 preceding siblings ...)
2015-04-19 0:40 ` [PATCH 10/12] tools/power turbostat: correct DRAM RAPL units on recent Xeon processors Len Brown
@ 2015-04-19 0:40 ` Len Brown
2015-04-19 0:40 ` [PATCH 12/12] tools/power turbostat: correct dumped pkg-cstate-limit value Len Brown
2015-04-19 14:26 ` [PATCH 01/12] tools/power turbostat: simplify default output Doug Smythies
11 siblings, 0 replies; 16+ messages in thread
From: Len Brown @ 2015-04-19 0:40 UTC (permalink / raw)
To: linux-pm; +Cc: Len Brown
From: Len Brown <len.brown@intel.com>
turbostat --debug
...
CPUID(0x15): eax_crystal: 2 ebx_tsc: 100 ecx_crystal_hz: 0
TSC: 1200 MHz (24000000 Hz * 100 / 2 / 1000000)
Signed-off-by: Len Brown <len.brown@intel.com>
---
tools/power/x86/turbostat/turbostat.c | 39 ++++++++++++++++++++++++++++++++++-
1 file changed, 38 insertions(+), 1 deletion(-)
diff --git a/tools/power/x86/turbostat/turbostat.c b/tools/power/x86/turbostat/turbostat.c
index 064749d..203d45d 100644
--- a/tools/power/x86/turbostat/turbostat.c
+++ b/tools/power/x86/turbostat/turbostat.c
@@ -89,6 +89,8 @@ double rapl_joule_counter_range;
unsigned int do_core_perf_limit_reasons;
unsigned int do_gfx_perf_limit_reasons;
unsigned int do_ring_perf_limit_reasons;
+unsigned int crystal_hz;
+unsigned long long tsc_hz;
#define RAPL_PKG (1 << 0)
/* 0x610 MSR_PKG_POWER_LIMIT */
@@ -2496,6 +2498,41 @@ void process_cpuid()
do_ptm ? "" : "No ",
has_epb ? "" : "No ");
+ if (max_level > 0x15) {
+ unsigned int eax_crystal;
+ unsigned int ebx_tsc;
+
+ /*
+ * CPUID 15H TSC/Crystal ratio, possibly Crystal Hz
+ */
+ eax_crystal = ebx_tsc = crystal_hz = edx = 0;
+ __get_cpuid(0x15, &eax_crystal, &ebx_tsc, &crystal_hz, &edx);
+
+ if (ebx_tsc != 0) {
+
+ if (debug && (ebx != 0))
+ fprintf(stderr, "CPUID(0x15): eax_crystal: %d ebx_tsc: %d ecx_crystal_hz: %d\n",
+ eax_crystal, ebx_tsc, crystal_hz);
+
+ if (crystal_hz == 0)
+ switch(model) {
+ case 0x4E: /* SKL */
+ case 0x5E: /* SKL */
+ crystal_hz = 24000000; /* 24 MHz */
+ break;
+ default:
+ crystal_hz = 0;
+ }
+
+ if (crystal_hz) {
+ tsc_hz = (unsigned long long) crystal_hz * ebx_tsc / eax_crystal;
+ if (debug)
+ fprintf(stderr, "TSC: %lld MHz (%d Hz * %d / %d / 1000000)\n",
+ tsc_hz / 1000000, crystal_hz, ebx_tsc, eax_crystal);
+ }
+ }
+ }
+
do_nhm_platform_info = do_nhm_cstates = do_smi = probe_nhm_msrs(family, model);
do_snb_cstates = has_snb_msrs(family, model);
do_pc2 = do_snb_cstates && (pkg_cstate_limit >= PCL__2);
@@ -2834,7 +2871,7 @@ int get_and_dump_counters(void)
}
void print_version() {
- fprintf(stderr, "turbostat version 4.3 24 Mar, 2015"
+ fprintf(stderr, "turbostat version 4.4 2 Apr, 2015"
" - Len Brown <lenb@kernel.org>\n");
}
--
2.4.0.rc1
^ permalink raw reply related [flat|nested] 16+ messages in thread* [PATCH 12/12] tools/power turbostat: correct dumped pkg-cstate-limit value
2015-04-19 0:40 ` [PATCH 01/12] tools/power turbostat: simplify default output Len Brown
` (9 preceding siblings ...)
2015-04-19 0:40 ` [PATCH 11/12] tools/power turbostat: calculate TSC frequency from CPUID(0x15) on SKL Len Brown
@ 2015-04-19 0:40 ` Len Brown
2015-04-19 14:26 ` [PATCH 01/12] tools/power turbostat: simplify default output Doug Smythies
11 siblings, 0 replies; 16+ messages in thread
From: Len Brown @ 2015-04-19 0:40 UTC (permalink / raw)
To: linux-pm; +Cc: Len Brown
From: Len Brown <len.brown@intel.com>
HSW expanded MSR_PKG_CST_CONFIG_CONTROL.Package-C-State-Limit,
from bits[2:0] used by previous implementations, to [3:0].
The value 1000b is unlimited, and is used by BDW and SKL too.
Signed-off-by: Len Brown <len.brown@intel.com>
---
tools/power/x86/turbostat/turbostat.c | 17 ++++++++---------
1 file changed, 8 insertions(+), 9 deletions(-)
diff --git a/tools/power/x86/turbostat/turbostat.c b/tools/power/x86/turbostat/turbostat.c
index 203d45d..bac98ca 100644
--- a/tools/power/x86/turbostat/turbostat.c
+++ b/tools/power/x86/turbostat/turbostat.c
@@ -1137,13 +1137,12 @@ int pkg_cstate_limit = PCLUKN;
char *pkg_cstate_limit_strings[] = { "reserved", "unknown", "pc0", "pc1", "pc2",
"pc3", "pc4", "pc6", "pc6n", "pc6r", "pc7", "pc7s", "pc8", "pc9", "unlimited"};
-int nhm_pkg_cstate_limits[8] = {PCL__0, PCL__1, PCL__3, PCL__6, PCL__7, PCLRSV, PCLRSV, PCLUNL};
-int snb_pkg_cstate_limits[8] = {PCL__0, PCL__2, PCL_6N, PCL_6R, PCL__7, PCL_7S, PCLRSV, PCLUNL};
-int hsw_pkg_cstate_limits[8] = {PCL__0, PCL__2, PCL__3, PCL__6, PCL__7, PCL_7S, PCLRSV, PCLUNL};
-int skl_pkg_cstate_limits[8] = {PCL__0, PCL__2, PCL__3, PCL__6, PCL__7, PCL_7S, PCL__8, PCL__9};
-int slv_pkg_cstate_limits[8] = {PCL__0, PCL__1, PCLRSV, PCLRSV, PCL__4, PCLRSV, PCL__6, PCL__7};
-int amt_pkg_cstate_limits[8] = {PCL__0, PCL__1, PCL__2, PCLRSV, PCLRSV, PCLRSV, PCL__6, PCL__7};
-int phi_pkg_cstate_limits[8] = {PCL__0, PCL__2, PCL_6N, PCL_6R, PCLRSV, PCLRSV, PCLRSV, PCLUNL};
+int nhm_pkg_cstate_limits[16] = {PCL__0, PCL__1, PCL__3, PCL__6, PCL__7, PCLRSV, PCLRSV, PCLUNL, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV};
+int snb_pkg_cstate_limits[16] = {PCL__0, PCL__2, PCL_6N, PCL_6R, PCL__7, PCL_7S, PCLRSV, PCLUNL, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV};
+int hsw_pkg_cstate_limits[16] = {PCL__0, PCL__2, PCL__3, PCL__6, PCL__7, PCL_7S, PCL__8, PCL__9, PCLUNL, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV};
+int slv_pkg_cstate_limits[16] = {PCL__0, PCL__1, PCLRSV, PCLRSV, PCL__4, PCLRSV, PCL__6, PCL__7, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV};
+int amt_pkg_cstate_limits[16] = {PCL__0, PCL__1, PCL__2, PCLRSV, PCLRSV, PCLRSV, PCL__6, PCL__7, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV};
+int phi_pkg_cstate_limits[16] = {PCL__0, PCL__2, PCL_6N, PCL_6R, PCLRSV, PCLRSV, PCLRSV, PCLUNL, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV, PCLRSV};
static void
dump_nhm_platform_info(void)
@@ -1707,7 +1706,7 @@ int probe_nhm_msrs(unsigned int family, unsigned int model)
}
get_msr(0, MSR_NHM_SNB_PKG_CST_CFG_CTL, &msr);
- pkg_cstate_limit = pkg_cstate_limits[msr & 0x7];
+ pkg_cstate_limit = pkg_cstate_limits[msr & 0xF];
return 1;
}
@@ -2871,7 +2870,7 @@ int get_and_dump_counters(void)
}
void print_version() {
- fprintf(stderr, "turbostat version 4.4 2 Apr, 2015"
+ fprintf(stderr, "turbostat version 4.5 2 Apr, 2015"
" - Len Brown <lenb@kernel.org>\n");
}
--
2.4.0.rc1
^ permalink raw reply related [flat|nested] 16+ messages in thread* RE: [PATCH 01/12] tools/power turbostat: simplify default output
2015-04-19 0:40 ` [PATCH 01/12] tools/power turbostat: simplify default output Len Brown
` (10 preceding siblings ...)
2015-04-19 0:40 ` [PATCH 12/12] tools/power turbostat: correct dumped pkg-cstate-limit value Len Brown
@ 2015-04-19 14:26 ` Doug Smythies
2015-04-19 21:02 ` Len Brown
11 siblings, 1 reply; 16+ messages in thread
From: Doug Smythies @ 2015-04-19 14:26 UTC (permalink / raw)
To: 'Len Brown'; +Cc: 'Len Brown', linux-pm
On 2015.04.18 17:40 Len Brown wrote:
> Casual turbostat users generally just want to know MHz.
> So by default, just print enough information to make sense of MHz.
>
> All the other configuration data and columns for C-states and temperature etc,
> are printed with the --debug option.
I almost never use turbostat to obtain MHz.
I only use turbostat to observe Energy or Power, and C-states.
I am not in favor of this change.
I do not want the extra clutter that comes with the --debug option
(although I do use it sometimes).
When doing experiments manually, without a script, it is convenient
to have many energy or power samples visible on the screen all at once.
Note: If I am using the acpi cpufreq driver (i.e. rarely), then yes
I use turbostat to obtain actual CPU frequencies, as I know of no
other way.
I reviewed some help type forums, and yes while many postings involving
turbostat are about CPU frequency, there are also some about power.
^ permalink raw reply [flat|nested] 16+ messages in thread* Re: [PATCH 01/12] tools/power turbostat: simplify default output
2015-04-19 14:26 ` [PATCH 01/12] tools/power turbostat: simplify default output Doug Smythies
@ 2015-04-19 21:02 ` Len Brown
0 siblings, 0 replies; 16+ messages in thread
From: Len Brown @ 2015-04-19 21:02 UTC (permalink / raw)
To: Doug Smythies; +Cc: Len Brown, Linux PM list
I hear you, Doug.
Personally, I *always* use --debug, and I *always* re-direct output to a file.
The fact of the matter is that on modern hardware, there are now more
columns than will fit on a screen, and most users will not even
understand them. Further, there are more columns on the way...
I started to go down the road of picking and choosing columns, but it
seemed that "minimum/everything" made more sense. And in that world,
minimum seems to be a more reasonable default.
thanks,
-Len
On Sun, Apr 19, 2015 at 10:26 AM, Doug Smythies <dsmythies@telus.net> wrote:
> On 2015.04.18 17:40 Len Brown wrote:
>
>> Casual turbostat users generally just want to know MHz.
>> So by default, just print enough information to make sense of MHz.
>>
>> All the other configuration data and columns for C-states and temperature etc,
>> are printed with the --debug option.
>
> I almost never use turbostat to obtain MHz.
> I only use turbostat to observe Energy or Power, and C-states.
>
> I am not in favor of this change.
>
> I do not want the extra clutter that comes with the --debug option
> (although I do use it sometimes).
> When doing experiments manually, without a script, it is convenient
> to have many energy or power samples visible on the screen all at once.
>
> Note: If I am using the acpi cpufreq driver (i.e. rarely), then yes
> I use turbostat to obtain actual CPU frequencies, as I know of no
> other way.
>
> I reviewed some help type forums, and yes while many postings involving
> turbostat are about CPU frequency, there are also some about power.
>
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-pm" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
--
Len Brown, Intel Open Source Technology Center
^ permalink raw reply [flat|nested] 16+ messages in thread