Re: Re: [PATCH] cpufreq: CPPC: Resolve the large frequency discrepancy from cpuinfo_cur_freq
From: Vanshidhar Konda
Date: Tue Feb 20 2024 - 11:11:39 EST
On Mon, Feb 19, 2024 at 08:15:50PM +0800, lihuisong (C) wrote:
在 2024/2/9 18:55, Beata Michalska 写道:
On Tue, Feb 06, 2024 at 04:02:15PM +0800, lihuisong (C) wrote:
在 2024/2/2 16:08, Beata Michalska 写道:
On Wed, Jan 17, 2024 at 05:18:40PM +0800, lihuisong (C) wrote:
Hi ,
Again, apologies for delay,
Hi,
在 2024/1/16 22:10, Beata Michalska 写道:
Hi,
Apologies for jumping in so late....
On Wed, Jan 10, 2024 at 03:09:48PM +0800, lihuisong (C) wrote:
Hi Ionela,
在 2024/1/8 22:03, Ionela Voinescu 写道:
Hi,
On Friday 05 Jan 2024 at 15:04:47 (+0800), lihuisong (C) wrote:
Hi Vanshi,
在 2024/1/5 8:48, Vanshidhar Konda 写道:
On Thu, Jan 04, 2024 at 05:36:51PM +0800, lihuisong (C) wrote:
在 2024/1/4 1:53, Ionela Voinescu 写道:
Hi,
On Tuesday 12 Dec 2023 at 15:26:17 (+0800), Huisong Li wrote:
Many developers found that the cpu current frequency is greater than
the maximum frequency of the platform, please see [1], [2] and [3].
In the scenarios with high memory access pressure, the patch [1] has
proved the significant latency of cpc_read() which is used to obtain
delivered and reference performance counter cause an absurd frequency.
The sampling interval for this counters is very critical and
is expected
to be equal. However, the different latency of cpc_read() has a direct
impact on their sampling interval.
Would this [1] alternative solution work for you?
It would work for me AFAICS.
Because the "arch_freq_scale" is also from AMU core and constant
counter, and read together.
But, from their discuss line, it seems that there are some tricky
points to clarify or consider.
I think the changes in [1] would work better when CPUs may be idle. With
this
patch we would have to wake any core that is in idle state to read the
AMU
counters. Worst case, if core 0 is trying to read the CPU frequency of
all
cores, it may need to wake up all the other cores to read the AMU
counters.
From the approach in [1], if all CPUs (one or more cores) under one policy
are idle, they still cannot be obtained the CPU frequency, right?
In this case, the [1] API will return 0 and have to back to call
cpufreq_driver->get() for cpuinfo_cur_freq.
Then we still need to face the issue this patch mentioned.
With the implementation at [1], arch_freq_get_on_cpu() will not return 0
for idle CPUs and the get() callback will not be called to wake up the
CPUs.
Right, arch_freq_get_on_cpu() will not return 0 for idle CPUs.
However, for no-housekeeping CPUs, it will return 0 and have to call get()
callback, right?
Worst case, arch_freq_get_on_cpu() will return a frequency based on the
AMU counter values obtained on the last tick on that CPU. But if that CPU
is not a housekeeping CPU, a housekeeping CPU in the same policy will be
selected, as it would have had a more recent tick, and therefore a more
recent frequency value for the domain.
But this frequency is from the last tick,
this last tick is probably a long time ago and it doesn't update
'arch_freq_scale' for some reasons like CPU dile.
In addition, I'm not sure if there is possible that amu_scale_freq_tick() is
executed delayed under high stress case.
It also have an impact on the accuracy of the cpu frequency we query.
I understand that the frequency returned here will not be up to date,
but there's no proper frequency feedback for an idle CPU. If one only
wakes up a CPU to sample counters, before the CPU goes back to sleep,
the obtained frequency feedback is meaningless.
For systems with 128 cores or more, this could be very expensive and
happen
very frequently.
AFAICS, the approach in [1] would avoid this cost.
But the CPU frequency is just an average value for the last tick period
instead of the current one the CPU actually runs at.
In addition, there are some conditions to use 'arch_freq_scale' in this
approach.
What are the conditions you are referring to?
It depends on the housekeeping CPUs.
So I'm not sure if this approach can entirely cover the frequency
discrepancy issue.
Unfortunately there is no perfect frequency feedback. By the time you
observe/use the value of scaling_cur_freq/cpuinfo_cur_freq, the frequency
of the CPU might have already changed. Therefore, an average value might
be a better indication of the recent performance level of a CPU.
An average value for CPU frequency is ok. It may be better if it has not any
delaying.
The original implementation for cpuinfo_cur_freq can more reflect their
meaning in the user-guide [1]. The user-guide said:
"cpuinfo_cur_freq : Current frequency of the CPU as obtained from the
hardware, in KHz.
This is the frequency the CPU actually runs at."
[1]https://www.kernel.org/doc/Documentation/cpu-freq/user-guide.txt
Would you be able to test [1] on your platform and usecase?
I has tested it on my platform (CPU number: 64, SMT: off and CPU base
frequency: 2.7GHz).
Accoding to the testing result,
1> I found that patch [1] and [2] cannot cover the no housekeeping CPUs.
They still have to face the large frequency discrepancy issue my patch
mentioned.
2> Additionally, the frequency value of all CPUs are almost the same by
using the 'arch_freq_scale' factor way. I'm not sure if it is ok.
The patch [1] has been modified silightly as below:
-->
@@ -1756,7 +1756,10 @@ static unsigned int
cpufreq_verify_current_freq(struct cpufreq_policy *policy, b
{
unsigned int new_freq;
- new_freq = cpufreq_driver->get(policy->cpu);
+ new_freq = arch_freq_get_on_cpu(policy->cpu);
+ if (!new_freq)
+ new_freq = cpufreq_driver->get(policy->cpu);
+
As pointed out this change will not make it to the next version of the patch.
So I'd say you can safely ignore it and assume that arch_freq_get_on_cpu will
only be wired for sysfs nodes for scaling_cur_freq/cpuinfo_cur_freq
if (!new_freq)
return 0;
And the result is as follows:
*case 1:**No setting the nohz_full and cpufreq use performance governor*
*--> Step1: *read 'cpuinfo_cur_freq' in no pressure
0: 2699264 2: 2699264 4: 2699264 6: 2699264
8: 2696628 10: 2696628 12: 2696628 14: 2699264
16: 2699264 18: 2696628 20: 2699264 22: 2696628
24: 2699264 26: 2696628 28: 2699264 30: 2696628
32: 2696628 34: 2696628 36: 2696628 38: 2696628
40: 2699264 42: 2699264 44: 2696628 46: 2696628
48: 2696628 50: 2699264 52: 2699264 54: 2696628
56: 2696628 58: 2696628 60: 2696628 62: 2696628
64: 2696628 66: 2699264 68: 2696628 70: 2696628
72: 2699264 74: 2696628 76: 2696628 78: 2699264
80: 2696628 82: 2696628 84: 2699264 86: 2696628
88: 2696628 90: 2696628 92: 2696628 94: 2699264
96: 2696628 98: 2699264 100: 2699264 102: 2696628
104: 2699264 106: 2699264 108: 2699264 110: 2696628
112: 2699264 114: 2699264 116: 2699264 118: 2699264
120: 2696628 122: 2699264 124: 2696628 126: 2699264
Note: the frequency of all CPUs are almost the same.
Were you expecting smth else ?
The frequency of each CPU might have a different value.
All value of all CPUs is the same under high pressure.
I don't know what the phenomenon is on other platform.
Do you know who else tested it?
So I might have rushed a bit with my previous comment/question: apologies for
that.
The numbers above: those are on a fairly idle/lightly loaded system right?
Yes.
Would you mind having another go with just the arch_freq_get_on_cpu
implementation beign added and dropping the changes in the cpufreq and
All my tests are done when cpufreq policy is "performance" and OS isn't on a
high load.
Reading "scaling_cur_freq" or "scaling_cur_freq" for each physical core on
platform
The testing result for "cpuinfo_cur_freq" with your changes on a fairly idle
and high loaded system can also be found in this thread.
*A: the result with your changes*
--> Reading "scaling_cur_freq"
0: 2688720 2: 2696628 4: 2699264 6: 2696628
8: 2699264 10: 2696628 12: 2699264 14: 2699264
16: 2699264 18: 2696628 20: 2696628 22: 2696628
24: 2699264 26: 2696628 28: 2696628 30: 2696628
32: 2699264 34: 2691356 36: 2696628 38: 2699264
40: 2699264 42: 2696628 44: 2696628 46: 2699264
48: 2699264 50: 2696628 52: 2696628 54: 2696628
56: 2696628 58: 2699264 60: 2691356 62: 2696628
64: 2696628 66: 2696628 68: 2696628 70: 2696628
72: 2696628 74: 2696628 76: 2699264 78: 2696628
80: 2696628 82: 2696628 84: 2699264 86: 2696628
88: 2625456 90: 2696628 92: 2699264 94: 2696628
96: 2696628 98: 2696628 100: 2699264 102: 2699264
104: 2699264 106: 2696628 108: 2699264 110: 2696628
112: 2699264 114: 2699264 116: 2696628 118: 2696628
120: 2696628 122: 2699264 124: 2696628 126: 2696628
-->Reading "cpuinfo_cur_freq"
0: 2696628 2: 2696628 4: 2699264 6: 2688720
8: 2699264 10: 2700000 12: 2696628 14: 2698322
16: 2699264 18: 2699264 20: 2696628 22: 2699264
24: 2699264 26: 2699264 28: 2699264 30: 2699264
32: 2699264 34: 2693992 36: 2696628 38: 2696628
40: 2699264 42: 2699264 44: 2699264 46: 2696628
48: 2696628 50: 2699264 52: 2696628 54: 2696628
56: 2699264 58: 2699264 60: 2696628 62: 2699264
64: 2696628 66: 2699264 68: 2696628 70: 2699264
72: 2696628 74: 2696628 76: 2696628 78: 2693992
80: 2696628 82: 2696628 84: 2696628 86: 2696628
88: 2696628 90: 2699264 92: 2696628 94: 2699264
96: 2699264 98: 2696628 100: 2699264 102: 2699264
104: 2691356 106: 2699264 108: 2699264 110: 2699264
112: 2699264 114: 2696628 116: 2699264 118: 2699264
120: 2696628 122: 2696628 124: 2696628 126: 2696628
*B: the result without your changes*
-->Reading "scaling_cur_freq"
0: 2698245 2: 2706690 4: 2699649 6: 2702105
8: 2704362 10: 2697993 12: 2701672 14: 2704362
16: 2701052 18: 2701052 20: 2694385 22: 2699650
24: 2706802 26: 2702389 28: 2698299 30: 2698299
32: 2697333 34: 2697993 36: 2701337 38: 2699328
40: 2700330 42: 2700330 44: 2698019 46: 2697697
48: 2699659 50: 2701700 52: 2703401 54: 2701700
56: 2704013 58: 2697658 60: 2695000 62: 2697666
64: 2697902 66: 2701052 68: 2698245 70: 2695789
72: 2701315 74: 2696655 76: 2693666 78: 2695317
80: 2704912 82: 2699649 84: 2698245 86: 2695454
88: 2697966 90: 2697959 92: 2699319 94: 2700680
96: 2695317 98: 2698996 100: 2700000 102: 2700334
104: 2701320 106: 2695065 108: 2700986 110: 2703960
112: 2697635 114: 2704421 116: 2700680 118: 2702040
120: 2700334 122: 2697993 124: 2700334 126: 2705351
-->Reading "cpuinfo_cur_freq"
0: 2696853 2: 2695454 4: 2699649 6: 2706993
8: 2706060 10: 2704362 12: 2704362 14: 2697658
16: 2707719 18: 2697192 20: 2702456 22: 2699650
24: 2705782 26: 2698299 28: 2703061 30: 2705802
32: 2700000 34: 2700671 36: 2701337 38: 2697658
40: 2700330 42: 2700330 44: 2699672 46: 2697697
48: 2703061 50: 2696610 52: 2692542 54: 2704406
56: 2695317 58: 2699331 60: 2698996 62: 2702675
64: 2704912 66: 2703859 68: 2699649 70: 2698596
72: 2703908 74: 2703355 76: 2697658 78: 2695317
80: 2702105 82: 2707719 84: 2702105 86: 2699649
88: 2697966 90: 2691525 92: 2701700 94: 2700680
96: 2695317 98: 2698996 100: 2698666 102: 2700334
104: 2690429 106: 2707590 108: 2700986 110: 2701320
112: 2696283 114: 2692881 116: 2697627 118: 2704421
120: 2698996 122: 2696321 124: 2696655 126: 2695000
So in both cases : whether you use arch_freq_get_on_cpu or not
(so with and without the patch) you get roughly the same frequencies
on all cores - or am I missing smth from the dump above ?
The changes in "with/without your changes" I said is your patch
intruduced arch_freq_get_on_cpu.
I just test them according to your requesting.
And those are reflecting max freq you have provided earlier (?)
I know it is an average frequency for the last tickfor using
arch_freq_get_on_cpu.
I have no any doubt that the freq is maximum value on performance governor.
I just want to say the difference between having or not having your patch.
The frequency values of all cores from cpuinfo_cur_freq and
scaling_cur_freq are almost the same if use this arch_freq_get_on_cpu
on my platform.
However, the frequency values of all cores are different if doesn't
use this arch_freq_get_on_cpu and just use .get().
Note that the arch_freq_get_on_cpu will return an average frequency for
the last tick, so even if your system is roughly idle with your performance
governor those numbers make sense (some/most of the cores might be idle
but you will see the last freq the core was running at before going to idle).
I do not think there is an agreement what should be shown for idle core when
querying their freq through sysfs. Showing last known freq makes sense, even
more than waking up core just to try to get one.
I'm not opposed to using frequency scale factor to get CPU frequency.
But it better be okay.
@Ionela: Please jump in if I got things wrong.
then read 'scaling_cur_freq', doing several reads in some intervals ?
It seems that above phenomenon has not a lot to do with reading intervals.
The change has been tested on RD-N2 model (Neoverse N2 ref platform),
it has also been discussed here [1]
I doesn't get the testing result on this platform in its thread.
It might be missing exact numbers but the conclusions should be here [1]
*--> Step 2: *read 'cpuinfo_cur_freq' in the high memory access pressure.
0: 2696628 2: 2696628 4: 2696628 6: 2696628
8: 2696628 10: 2696628 12: 2696628 14: 2696628
16: 2696628 18: 2696628 20: 2696628 22: 2696628
24: 2696628 26: 2696628 28: 2696628 30: 2696628
32: 2696628 34: 2696628 36: 2696628 38: 2696628
40: 2696628 42: 2696628 44: 2696628 46: 2696628
48: 2696628 50: 2696628 52: 2696628 54: 2696628
56: 2696628 58: 2696628 60: 2696628 62: 2696628
64: 2696628 66: 2696628 68: 2696628 70: 2696628
72: 2696628 74: 2696628 76: 2696628 78: 2696628
80: 2696628 82: 2696628 84: 2696628 86: 2696628
88: 2696628 90: 2696628 92: 2696628 94: 2696628
96: 2696628 98: 2696628 100: 2696628 102: 2696628
104: 2696628 106: 2696628 108: 2696628 110: 2696628
112: 2696628 114: 2696628 116: 2696628 118: 2696628
120: 2696628 122: 2696628 124: 2696628 126: 2696628
*Case 2: setting nohz_full and cpufreq use ondemand governor*
There is "isolcpus=1-10,41-50 nohz_full=1-10,41-50 rcu_nocbs=1-10,41-50" in
/proc/cmdline.
Right, so if I remember correctly nohz_full implies rcu_nocbs, so no need to
set that one.
Now, afair, isolcpus will make the selected CPUs to disappear from the
schedulers view (no balancing, no migrating), so unless you affine smth
explicitly to those CPUs, you will not see much of an activity there.
Correct.
Need to double check though as it has been a while ...
*--> Step 1: *setting ondemand governor to all policy and query
'cpuinfo_cur_freq' in no pressure case.
And the frequency of CPUs all are about 400MHz.
*--> Step 2:* read 'cpuinfo_cur_freq' in the high memory access pressure.
The high memory access pressure is from the command: "stress-ng -c 64
--cpu-load 100% --taskset 0-63"
I'm not entirely convinced that this will affine to isolated cpus, especially
that the affinity mask spans all available cpus. If that is the case, no wonder
your isolated cpus are getting wasted being idle. But I would have to double
check how this is being handled.
The result:
0: 2696628 1: 400000 2: 400000 3: 400909
4: 400000 5: 400000 6: 400000 7: 400000
8: 400000 9: 400000 10: 400600 11: 2696628
12: 2696628 13: 2696628 14: 2696628 15: 2696628
16: 2696628 17: 2696628 18: 2696628 19: 2696628
20: 2696628 21: 2696628 22: 2696628 23: 2696628
24: 2696628 25: 2696628 26: 2696628 27: 2696628
28: 2696628 29: 2696628 30: 2696628 31: 2696628
32: 2696628 33: 2696628 34: 2696628 35: 2696628
36: 2696628 37: 2696628 38: 2696628 39: 2696628
40: 2696628 41: 400000 42: 400000 43: 400000
44: 400000 45: 398847 46: 400000 47: 400000
48: 400000 49: 400000 50: 400000 51: 2696628
52: 2696628 53: 2696628 54: 2696628 55: 2696628
56: 2696628 57: 2696628 58: 2696628 59: 2696628
60: 2696628 61: 2696628 62: 2696628 63: 2699264
Note:
(1) The frequency of 1-10 and 41-50 CPUs work on the lowest frequency.
It turned out that nohz full was already work.
I guess that stress-ng cannot use the CPU in the range of nohz full.
Because the CPU frequency will be increased to 2.7G by binding CPU to
other application.
(2) The frequency of the nohz full core is calculated by get() callback
according to ftrace.
It is as there is no sched tick on those, and apparently there is nothing
running on them either.
Yes.
If we select your approach and the above phenomenon is normal,
the large frequency discrepancy issue can be resolved for CPUs with sched
tick by the way.
But the nohz full cores still have to face this issue. So this patch is also
needed.
Yes, nohz cores full have to be handled by the cpufreq driver.
Correct. So we still have to face the issue in this patch and push this
patch.
Beata, would you please review this patch?
Just to clarify for my benefit (apologies but I do have to contex switch
pretty often these days): by reviewing this patch do you mean:
1) review your changes (if so I think there are few comments already to be
addressed, but I can try to have another look)
Currently, the main comments is that my patch will wake up CPU to get
frequency.
BTW, the core's always been wakened up to get the frequency for FFH
way in cppc_acpi. please see cpc_read_ffh().
So it may be acceptable. After all, we don't query CPU frequency very often.
Today's implementation of cpc_read_ffh() wakes up the core to read AMU
counters - this is far from ideal. According to the architecture
specification the CPU_CYCLES and CNT_CYCLES counters in AMU do not
increment when the core is in WFI or WFE. If we cache the value of the
AMU counter before a PE goes idle, we may be able to avoid waking up a
PE just to read the AMU counters. I'm wondering if it makes sense to
cache the value in cpu_do_idle() and return this cached value if
idle_cpu() returns true.
But your patch doesn't meet the non-housekeeping cpus.
For non-housekeeping CPUs may be it is better to just invoke
cpufreq->get() call?
Thanks,
Vanshi
2) review changes for AMU-based arch_freq_get_on_cpu ?
*note: I will still try to have a look at the non-housekeeping cpus case
I am very much hope that this issue my patch mentioned can be resolved ASAP.
So what's your plan about non-housekeeping cpus case?
---
[1] https://lore.kernel.org/lkml/691d3eb2-cd93-f0fc-a7a4-2a8c0d44262c@xxxxxxxxxx/
---
BR
Beata
/Huisong
[...]
.