All of lore.kernel.org
 help / color / mirror / Atom feed
* Re: intel-pstate driver not clocking down on Atom Z8750
       [not found] <CAAYJcCWViZ0DTj6e9XxN1xkiBR5CAbFA6R73jg=o8d2J6Z2raA@mail.gmail.com>
@ 2017-06-25 21:24 ` Rafael J. Wysocki
       [not found] ` <CAAYJcCXPUE78PBN_njrnHseBUOPXsek48ac4RGh5Jx_mgvYpKg@mail.gmail.com>
  1 sibling, 0 replies; 3+ messages in thread
From: Rafael J. Wysocki @ 2017-06-25 21:24 UTC (permalink / raw)
  To: Christopher Williamson
  Cc: srinivas.pandruvada, lenb, viresh.kumar, linux-pm, Hans de Goede

On Sunday, June 25, 2017 03:39:11 PM Christopher Williamson wrote:
> Hi there,
> 
> I've just purchased a device called a "GPD Pocket" (more info at
> indiegogo.com/projects/gpd-pocket-7-0-umpc-laptop-ubuntu-or-win-10-os-laptop--2/x/7959405)
> but in short - one of the major problems I have with the device is that the
> included Intel Atom Z8750 never clocks below the turbo boost speed meaning
> it overheats very quickly and crashes.
> 
> I have managed to avoid this behaviour in two ways:
> 
> 1.) Disable turbo boost - the device then sits firmly at 1.6GHz and refuses
> to clock below that.
> 
> 2.) Disable the intel_pstate driver and the CPU then clocks properly but
> cannot turbo boost at all so I lose that capability.
> 
> I'm really just reaching out to you fine folks to find out where best to
> discuss this issue. I'm happy to raise a kernel bug report if that helps
> and also happy to provide command outputs as needed.
> 
> I have included Hans de Goede in the CC list on this email since he has
> been a great help getting to this point - this device has a few strange
> quirks to work through but this is a fairly serious one.

What kernel versions have you tried?

Thanks,
Rafael

^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: intel-pstate driver not clocking down on Atom Z8750
       [not found] ` <CAAYJcCXPUE78PBN_njrnHseBUOPXsek48ac4RGh5Jx_mgvYpKg@mail.gmail.com>
@ 2017-06-26  0:27   ` Srinivas Pandruvada
  2017-06-26 21:22     ` Rafael J. Wysocki
  0 siblings, 1 reply; 3+ messages in thread
From: Srinivas Pandruvada @ 2017-06-26  0:27 UTC (permalink / raw)
  To: Christopher Williamson, Rafael J. Wysocki
  Cc: lenb, viresh.kumar, linux-pm, Hans de Goede

[-- Attachment #1: Type: text/plain, Size: 1892 bytes --]

On Sun, 2017-06-25 at 18:37 -0400, Christopher Williamson wrote:
> 4.8 and 4.12-rc6
> 
I just gave a quick test on a HP Cherrytrail laptop with 4.12 tip.
Attached is the output, which show that this does scale.
It is possible that you don't have any idle (C-states) or some task is
keeping system busy.

I suggest:
1. Run turbostat and attach output.
2. Attach output of tools/power/x86/intel_pstate_tracer for a minute
run.

Thanks,
Srinivas

> On Jun 25, 2017 at 10:24 PM, "Rafael J. Wysocki" <rjw@rjwysocki.net>
> wrote: 
> > On Sunday, June 25, 2017 03:39:11 PM Christopher Williamson wrote:
> > > Hi there,
> > >
> > > I've just purchased a device called a "GPD Pocket" (more info at
> > > indiegogo.com/projects/gpd-pocket-7-0-umpc-laptop-ubuntu-or-win-
> > 10-os-laptop--2/x/7959405)
> > > but in short - one of the major problems I have with the device
> > is that the
> > > included Intel Atom Z8750 never clocks below the turbo boost
> > speed meaning
> > > it overheats very quickly and crashes.
> > >
> > > I have managed to avoid this behaviour in two ways:
> > >
> > > 1.) Disable turbo boost - the device then sits firmly at 1.6GHz
> > and refuses
> > > to clock below that.
> > >
> > > 2.) Disable the intel_pstate driver and the CPU then clocks
> > properly but
> > > cannot turbo boost at all so I lose that capability.
> > >
> > > I'm really just reaching out to you fine folks to find out where
> > best to
> > > discuss this issue. I'm happy to raise a kernel bug report if
> > that helps
> > > and also happy to provide command outputs as needed.
> > >
> > > I have included Hans de Goede in the CC list on this email since
> > he has
> > > been a great help getting to this point - this device has a few
> > strange
> > > quirks to work through but this is a fairly serious one.
> > 
> > What kernel versions have you tried?
> > 
> > Thanks,
> > Rafael
> > 
> > 
> > 

[-- Attachment #2: cht_turbostat.txt --]
[-- Type: text/plain, Size: 3088 bytes --]

[testuser@otcpl-hp-cht-fedora turbostat]$ sudo turbostat --show Core,CPU,Avg_MHz,Busy%,Bzy_MHz,TSC_MHz
turbostat version 17.04.12 - Len Brown <lenb@kernel.org>
CPUID(0): GenuineIntel 11 CPUID levels; family:model:stepping 0x6:4c:3 (6:76:3)
CPUID(1): SSE3 MONITOR - EIST TM2 TSC MSR ACPI-TM TM
CPUID(6): APERF, TURBO, DTS, No-PTM, No-HWP, No-HWPnotify, No-HWPwindow, No-HWPepp, No-HWPpkg, EPB
cpu3: MSR_IA32_MISC_ENABLE: 0x00850089 (TCC EIST No-MWAIT PREFETCH TURBO)
CPUID(7): No-SGX
cpu3: MSR_PLATFORM_INFO: 0x60006001400
6 * 133.3 = 800.0 MHz max efficiency frequency
20 * 133.3 = 2666.6 MHz base frequency
cpu3: MSR_IA32_POWER_CTL: 0x00000000 (C1E auto-promotion: DISabled)
cpu3: MSR_TURBO_RATIO_LIMIT: 0x00000000
cpu3: MSR_PKG_CST_CONFIG_CONTROL: 0x0006000f (UNlocked: pkg-cstate-limit=15: unknown)
cpu3: cpufreq driver: intel_pstate
cpu3: cpufreq governor: powersave
cpufreq intel_pstate no_turbo: 0
cpu0: MSR_IA32_ENERGY_PERF_BIAS: 0x00000006 (balanced)
cpu0: MSR_IA32_TEMPERATURE_TARGET: 0x035a0000 (90 C)
Core	CPU	Avg_MHz	Busy%	Bzy_MHz	TSC_MHz
-	-	6	1.04	925	1600
0	0	8	1.47	913	1600
1	1	8	1.39	951	1600
2	2	2	0.34	856	1600
3	3	5	0.97	928	1600
Core	CPU	Avg_MHz	Busy%	Bzy_MHz	TSC_MHz
-	-	7	1.11	1109	1600
0	0	15	2.12	1217	1600
1	1	12	1.91	1055	1600
2	2	1	0.12	800	1600
3	3	1	0.30	800	1600
Core	CPU	Avg_MHz	Busy%	Bzy_MHz	TSC_MHz
-	-	3	0.53	964	1600
0	0	7	1.13	1006	1600
1	1	3	0.46	1049	1600
2	2	0	0.10	800	1600
3	3	2	0.43	800	1600
Core	CPU	Avg_MHz	Busy%	Bzy_MHz	TSC_MHz
-	-	6	1.06	964	1600
0	0	11	1.96	954	1600
1	1	9	1.38	1035	1600
2	2	1	0.27	884	1600
3	3	3	0.61	873	1600
Core	CPU	Avg_MHz	Busy%	Bzy_MHz	TSC_MHz
-	-	21	2.03	1717	1600
0	0	22	2.88	1295	1600
1	1	34	2.81	1995	1600
2	2	13	1.08	2025	1600
3	3	14	1.34	1798	1600
Core	CPU	Avg_MHz	Busy%	Bzy_MHz	TSC_MHz
-	-	88	8.16	1800	1600
0	0	75	9.64	1296	1600
1	1	76	7.86	1617	1600
2	2	100	7.53	2217	1600
3	3	101	7.61	2216	1600
Core	CPU	Avg_MHz	Busy%	Bzy_MHz	TSC_MHz
-	-	105	8.44	2068	1600
0	0	228	15.50	2451	1600
1	1	151	12.21	2067	1600
2	2	13	2.29	955	1600
3	3	26	3.74	1158	1600
Core	CPU	Avg_MHz	Busy%	Bzy_MHz	TSC_MHz
-	-	415	25.34	2730	1600
0	0	388	22.84	2832	1600
1	1	575	33.93	2823	1600
2	2	438	26.27	2781	1600
3	3	259	18.33	2359	1600
Core	CPU	Avg_MHz	Busy%	Bzy_MHz	TSC_MHz
-	-	38	4.38	1441	1600
0	0	69	7.07	1631	1600
1	1	41	4.58	1479	1600
2	2	36	4.85	1242	1600
3	3	5	1.01	896	1600
Core	CPU	Avg_MHz	Busy%	Bzy_MHz	TSC_MHz
-	-	14	1.21	1854	1600
0	0	8	1.12	1246	1600
1	1	14	1.16	2052	1600
2	2	20	1.52	2179	1600
3	3	12	1.07	1816	1600
Core	CPU	Avg_MHz	Busy%	Bzy_MHz	TSC_MHz
-	-	21	3.13	1116	1600
0	0	18	2.51	1221	1600
1	1	4	0.65	1105	1600
2	2	20	3.38	1007	1600
3	3	41	5.99	1134	1600
Core	CPU	Avg_MHz	Busy%	Bzy_MHz	TSC_MHz
-	-	232	17.48	2212	1600
0	0	245	18.85	2166	1600
1	1	163	13.29	2040	1600
2	2	270	19.51	2305	1600
3	3	250	18.26	2284	1600
Core	CPU	Avg_MHz	Busy%	Bzy_MHz	TSC_MHz
-	-	67	7.05	1582	1600
0	0	71	8.93	1333	1600
1	1	49	4.01	2024	1600
2	2	93	8.57	1807	1600
3	3	55	6.69	1364	1600
Core	CPU	Avg_MHz	Busy%	Bzy_MHz	TSC_MHz
-	-	24	3.70	1072	1600
0	0	33	5.72	960	1600
1	1	15	2.75	891	1600
2	2	24	3.46	1152	1600
3	3	24	2.87	1374	1600


^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: intel-pstate driver not clocking down on Atom Z8750
  2017-06-26  0:27   ` Srinivas Pandruvada
@ 2017-06-26 21:22     ` Rafael J. Wysocki
  0 siblings, 0 replies; 3+ messages in thread
From: Rafael J. Wysocki @ 2017-06-26 21:22 UTC (permalink / raw)
  To: Srinivas Pandruvada, Christopher Williamson
  Cc: lenb, viresh.kumar, linux-pm, Hans de Goede

On Sunday, June 25, 2017 05:27:59 PM Srinivas Pandruvada wrote:
> On Sun, 2017-06-25 at 18:37 -0400, Christopher Williamson wrote:
> > 4.8 and 4.12-rc6
> > 
> I just gave a quick test on a HP Cherrytrail laptop with 4.12 tip.
> Attached is the output, which show that this does scale.
> It is possible that you don't have any idle (C-states) or some task is
> keeping system busy.
> 
> I suggest:
> 1. Run turbostat and attach output.
> 2. Attach output of tools/power/x86/intel_pstate_tracer for a minute
> run.

The output of acpidump may be useful as well.

I'd suggest filing a bug at bugzilla.kernel.org against intel_pstate and
attaching all of the requested information there.

You don't have to provide a full problem description again, a link to the
initial message in this thread should suffice.

Thanks,
Rafael

^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2017-06-26 21:29 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
     [not found] <CAAYJcCWViZ0DTj6e9XxN1xkiBR5CAbFA6R73jg=o8d2J6Z2raA@mail.gmail.com>
2017-06-25 21:24 ` intel-pstate driver not clocking down on Atom Z8750 Rafael J. Wysocki
     [not found] ` <CAAYJcCXPUE78PBN_njrnHseBUOPXsek48ac4RGh5Jx_mgvYpKg@mail.gmail.com>
2017-06-26  0:27   ` Srinivas Pandruvada
2017-06-26 21:22     ` Rafael J. Wysocki

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.