| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <3c36b52e1dab2194e1a5babbfa333e4938cdfc85.camel@linux.ibm.com>
Date: Mon, 24 Feb 2025 12:08:58 +0530
From: Aboorva Devarajan <aboorvad@...ux.ibm.com>
To: "Rafael J. Wysocki" <rjw@...ysocki.net>,
Linux PM
<linux-pm@...r.kernel.org>
Cc: LKML <linux-kernel@...r.kernel.org>,
Daniel Lezcano
<daniel.lezcano@...aro.org>,
Christian Loehle <christian.loehle@....com>,
Artem Bityutskiy <artem.bityutskiy@...ux.intel.com>,
aboorvad@...ux.ibm.com
Subject: Re: [RFT][PATCH v1 0/5] cpuidle: menu: Avoid discarding useful
information when processing recent idle intervals
On Mon, 2025-02-24 at 11:57 +0530, Aboorva Devarajan wrote:
> On Thu, 2025-02-06 at 15:21 +0100, Rafael J. Wysocki wrote:
>
> > Hi Everyone,
> >
> > ...
> >
> >
> >
>
> Hi Rafael,
>
> I ran some tests using a cpuidle microbenchmark that I have [1]:
I missed including the system configuration in my previous email, so here it is,
I carried out these experiments on a Power10 - pseries logical partition with
the following configuration:
$ lscpu
Architecture: ppc64le
Byte Order: Little Endian
CPU(s): 80
On-line CPU(s) list: 0-79
Model name: POWER10 (architected), altivec supported
Model: 2.0 (pvr 0080 0200)
Thread(s) per core: 8
Core(s) per socket: 10
Socket(s): 1
Physical sockets: 8
Physical chips: 1
Physical cores/chip: 10
Virtualization features:
Hypervisor vendor: pHyp
Virtualization type: para
Caches (sum of all):
L1d: 640 KiB (20 instances)
L1i: 960 KiB (20 instances)
L2: 20 MiB (20 instances)
L3: 80 MiB (20 instances)
NUMA:
NUMA node(s): 1
NUMA node4 CPU(s): 0-79
$ cpupower idle-info
CPUidle driver: pseries_idle
CPUidle governor: menu
analyzing CPU 47:
Number of idle states: 2
Available idle states: snooze CEDE
snooze:
Flags/Description: snooze
Latency: 0
Residency: 0
Usage: 5899
Duration: 360940
CEDE:
Flags/Description: CEDE
Latency: 12
Residency: 120
Usage: 31037
Duration: 16664409173
>
> The test does a uniform distribution of idle durations, which means the past eight idle intervals approximately match
> the sleep duration. So as anticipated, this change-set does not impact this case, and the behavior remains mostly
> consistent even after applying the patch.
>
> Pipe based wakeup - Above %
> ----------------------------------------------------------------------
> > Sleep time (s) | Menu No Patch(%) | Menu Patch(%)| Teo No Patch (%) |
> > ---------------- |------------------|--------------|------------------|
> > 5 | 0.00% | 0.00% | 0.00% |
> > 10 | 0.00% | 0.00% | 0.00% |
> ------------------------------------------------------------------------
> > 30 | 99.97% | 99.96% | 0.00% |
> > 100 | 98.61% | 98.67% | 0.00% |
> > 120 | 97.88% | 98.42% | 1.03% | -> (*)
> ------------------------------------------------------------------------
> > 130 | 2.82% | 1.26% | 0.22% |
> > 150 | 1.68% | 1.43% | 0.32% |
> > 200 | 2.09% | 1.91% | 0.15% |
> > 300 | 1.22% | 1.22% | 0.19% |
> > 400 | 1.20% | 1.21% | 0.19% |
> > 500 | 1.16% | 1.12% | 0.12% |
> ------------------------------------------------------------------------
>
> Pipe based wakeup - Below %
> -----------------------------------------------------------------------
> > Sleep time (s) | Menu No Patch(%) | Menu Patch(%)| Teo No Patch (%) |
> > ---------------- |------------------|--------------|------------------|
> > 5 | 0.00% | 0.00% | 0.00% |
> > 10 | 0.00% | 0.00% | 0.00% |
> > 30 | 0.00% | 0.00% | 0.00% |
> > 100 | 0.00% | 0.00% | 0.00% |
> > 120 | 2.76% | 1.14% | 0.93% |
> > 130 | 3.11% | 1.13% | 0.12% |
> > 150 | 1.34% | 1.16% | 0.18% |
> > 200 | 1.38% | 1.15% | 0.09% |
> > 300 | 1.33% | 1.21% | 0.11% |
> > 400 | 1.36% | 1.25% | 0.10% |
> > 500 | 1.25% | 1.22% | 0.10% |
> > ---------------- |------------------|--------------|------------------|
>
> Time based wakeup - Above %
> -----------------------------------------------------------------------
> > Sleep time (s) | Menu No Patch(%) | Menu Patch(%)| Teo No Patch (%) |
> > ---------------- |------------------|--------------|------------------|
> > 5 | 0.00% | 0.00% | 0.00% |
> > 10 | 0.00% | 0.00% | 0.00% |
> > 30 | 0.00% | 0.00% | 0.00% |
> > 100 | 0.01% | 0.00% | 0.00% |
> > 120 | 0.00% | 0.00% | 0.15% |
> > 130 | 15.84% | 0.14% | 0.23% |
> > 150 | 0.39% | 0.23% | 0.48% |
> > 200 | 0.95% | 0.87% | 0.10% |
> > 300 | 0.20% | 0.17% | 0.15% |
> > 400 | 0.14% | 0.12% | 0.17% |
> > 500 | 0.10% | 0.20% | 0.11% |
> > ---------------- |------------------|--------------|------------------|
>
> Time based wakeup - Below %
> -----------------------------------------------------------------------
> > Sleep time (s) | Menu No Patch(%) | Menu Patch(%)| Teo No Patch (%) |
> > ---------------- |------------------|--------------|------------------|
> > 5 | 0.00% | 0.00% | 0.00% |
> > 10 | 0.00% | 0.00% | 0.00% |
> > 30 | 0.00% | 0.00% | 0.00% |
> > 100 | 0.00% | 0.00% | 0.00% |
> > 120 | 1.85% | 1.66% | 2.67% |
> > 130 | 16.71% | 1.13% | 1.11% |
> > 150 | 1.36% | 1.16% | 1.13% |
> > 200 | 1.33% | 1.14% | 1.19% |
> > 300 | 1.44% | 1.20% | 1.17% |
> > 400 | 1.51% | 1.21% | 1.21% |
> > 500 | 1.42% | 1.24% | 1.25% |
> > ---------------- |------------------|--------------|------------------|
>
>
> (*) - Above and below values are higher even without the patch with menu governor,
> this issue still persists, as previously reported in [2]. I will investigate
> further and submit a revision to get additional feedback.
>
> I also carried out some benchmarks using pgbench:
>
> pgbench Results
>
> Without Patch:
> -----------------------------------------------------------------------------
> > Run | Transactions Processed | Latency Avg (ms) | TPS (without init time) |
> -----------------------------------------------------------------------------
> > 1 | 11,936,327 | 0.050 | 198,946.141025 |
> > 2 | 11,899,540 | 0.050 | 198,333.097547 |
> > 3 | 11,870,792 | 0.051 | 197,853.728614 |
> > 4 | 11,901,670 | 0.050 | 198,368.526139 |
> > 5 | 11,922,046 | 0.050 | 198,708.112243 |
> -----------------------------------------------------------------------------
> > Avg | 11,906,075 | 0.0502 | 198,441.921114 |
> -----------------------------------------------------------------------------
>
> With Patch:
> -----------------------------------------------------------------------------
> > Run | Transactions Processed | Latency Avg (ms) | TPS (without init time) |
> -----------------------------------------------------------------------------
> > 1 | 12,052,865 | 0.050 | 200,888.492771 |
> > 2 | 12,058,359 | 0.050 | 200,979.895325 |
> > 3 | 12,071,012 | 0.050 | 201,190.809734 |
> > 4 | 12,054,646 | 0.050 | 200,918.076736 |
> > 5 | 12,053,087 | 0.050 | 200,892.045581 |
> -----------------------------------------------------------------------------
> > Avg | 12,058,394 | 0.0500 | 200,973.464029 |
> ------------------------------------------------------------------------------
>
> Performance Improvement After Patch:
> --------------------------------------------------------------------------------------
> > Metric | Without Patch (Avg.)| With Patch (Avg.) | % Improvement |
> --------------------------------------------------------------------------------------
> > Transactions Processed | 11,906,075 | 12,058,394 | +1.28% |
> > Latency Avg (ms) | 0.0502 | 0.0500 | -0.40% |
> > TPS (without init time) | 198,441.921114 | 200,973.464029 | +1.28% |
> --------------------------------------------------------------------------------------
>
> So with the patch pgbench's "Transactions Processed" improves by ~1.28%, and I did not observe
> any major variations on other benchmarks with the patch.
>
> So for the entire series:
>
> Tested-by: Aboorva Devarajan <aboorvad@...ux.ibm.com>
>
> I'm also trying a minimal unit fuzz-test with the pre- and post- patched version of the get_typical_interval
> function to understand this better, will post the results soon.
>
> [1] - https://github.com/AboorvaDevarajan/linux-utils/tree/main/cpuidle/cpuidle_wakeup
> [2] - https://lore.kernel.org/all/20240809073120.250974-1-aboorvad@linux.ibm.com/
>
>
> Thanks,
> Aboorva
Powered by blists - more mailing lists