Interbench benchmarks for MuQSS 116

As mentioned in my previous post, I recently upgraded interbench which is a benchmark application I invented/wrote to assess perceptible latency in the setting of various loads. The updates were to make the results meaningful on today's larger ram/multicore machines where the load scales accordingly.

The results for mainline 4.8.4 and 4.8.4-ck4 on a multithreaded hexcore (init 1) can be found here:
 http://ck.kolivas.org/patches/muqss/Benchmarks/20161024/
and are copied below. I do not have swap on this machine so the "memload" was not performed. This is a 3.6GHz hexcore with 64GB ram and fast Intel SSDs so to show any difference on this is nice. To make it easier, I've highlighted it in colours similar to the throughput benchmarks I posted previously. Blue means within 1% of each other, red means significantly worse and green significantly better.

Load set to 12 processors

Using 4008580 loops per ms, running every load for 30 seconds
Benchmarking kernel 4.8.4 at datestamp 201610242116
Comment: cfs

--- Benchmarking simulated cpu of Audio in the presence of simulated ---
Load Latency +/- SD (ms)  Max Latency   % Desired CPU  % Deadlines Met
None       0.1 +/- 0.1        0.1           100         100
Video      0.0 +/- 0.0        0.1           100         100
X          0.1 +/- 0.1        0.1           100         100
Burn       0.0 +/- 0.0        0.0           100         100
Write      0.1 +/- 0.1        0.1           100         100
Read       0.1 +/- 0.1        0.1           100         100
Ring       0.0 +/- 0.0        0.1           100         100
Compile    0.0 +/- 0.0        0.0           100         100

--- Benchmarking simulated cpu of Video in the presence of simulated ---
Load Latency +/- SD (ms)  Max Latency   % Desired CPU  % Deadlines Met
None       0.1 +/- 0.1        0.1           100         100
X          0.1 +/- 0.1        0.1           100         100
Burn      17.4 +/- 19.5      46.3            87        7.62
Write      0.1 +/- 0.1        0.1           100         100
Read       0.1 +/- 0.1        0.1           100         100
Ring       0.0 +/- 0.0        0.0           100         100
Compile   17.4 +/- 19.1      45.9          89.5        6.07

--- Benchmarking simulated cpu of X in the presence of simulated ---
Load Latency +/- SD (ms)  Max Latency   % Desired CPU  % Deadlines Met
None       0.0 +/- 0.1        1.0           100        99.3
Video     13.4 +/- 25.8      68.0          36.2        27.3
Burn      94.4 +/- 127.0    334.0          12.9        4.37
Write      0.1 +/- 0.4        4.0          97.4        96.4
Read       0.1 +/- 0.7        4.0          96.2        93.8
Ring       0.5 +/- 1.9        9.0          89.3        84.9
Compile   93.3 +/- 127.7    333.0          12.2         4.2

--- Benchmarking simulated cpu of Gaming in the presence of simulated ---
Load Latency +/- SD (ms)  Max Latency   % Desired CPU
None       0.0 +/- 0.2        2.2           100
Video      7.9 +/- 21.4      69.3          92.7
X          1.4 +/- 1.6        2.7          98.7
Burn     136.5 +/- 145.3    360.8          42.3
Write      1.8 +/- 2.0        4.4          98.2
Read      11.2 +/- 20.3      47.8          89.9
Ring       8.1 +/- 8.1        8.2          92.5
Compile  152.3 +/- 166.8    346.1          39.6

Load set to 12 processors

Using 4008580 loops per ms, running every load for 30 seconds
Benchmarking kernel 4.8.4-ck4+ at datestamp 201610242047
Comment: muqss116-int1

--- Benchmarking simulated cpu of Audio in the presence of simulated ---
Load Latency +/- SD (ms)  Max Latency   % Desired CPU  % Deadlines Met
None       0.0 +/- 0.0        0.0           100         100
Video      0.0 +/- 0.0        0.0           100         100
X          0.0 +/- 0.0        0.0           100         100
Burn       0.0 +/- 0.0        0.0           100         100
Write      0.0 +/- 0.0        0.1           100         100
Read       0.0 +/- 0.0        0.0           100         100
Ring       0.0 +/- 0.0        0.0           100         100
Compile    0.0 +/- 0.1        0.8           100         100

--- Benchmarking simulated cpu of Video in the presence of simulated ---
Load Latency +/- SD (ms)  Max Latency   % Desired CPU  % Deadlines Met
None       0.0 +/- 0.0        0.0           100         100
X          0.0 +/- 0.0        0.0           100         100
Burn       3.1 +/- 7.2       17.7           100        81.6
Write      0.0 +/- 0.0        0.5           100         100
Read       0.0 +/- 0.0        0.0           100         100
Ring       0.0 +/- 0.0        0.0           100         100
Compile   10.5 +/- 13.3      19.7           100        37.3

--- Benchmarking simulated cpu of X in the presence of simulated ---
Load Latency +/- SD (ms)  Max Latency   % Desired CPU  % Deadlines Met
None       0.0 +/- 0.1        1.0           100        99.3
Video      3.7 +/- 12.1      56.0            89        82.6
Burn      47.2 +/- 66.5     142.0          16.7        7.58
Write      0.1 +/- 0.5        5.0          97.7        95.7
Read       0.1 +/- 0.7        4.0          95.6        93.5
Ring       0.5 +/- 1.9       12.0          89.8          86
Compile   55.9 +/- 77.6     196.0          18.6        8.12

--- Benchmarking simulated cpu of Gaming in the presence of simulated ---
Load Latency +/- SD (ms)  Max Latency   % Desired CPU
None       0.0 +/- 0.1        0.5           100
Video      1.2 +/- 1.2        1.8          98.8
X          1.4 +/- 1.6        2.9          98.7
Burn     130.9 +/- 132.1    160.3          43.3
Write      2.4 +/- 2.5        7.0          97.7
Read       3.2 +/- 3.2        3.6          96.9
Ring       5.9 +/- 6.2       10.3          94.4
Compile  146.5 +/- 149.3    209.2          40.6

As you can see, the only times mainline is better, there is less than 1% difference between them which is within the margins for noise. MuQSS meets more deadlines, gives the benchmarked task more of its desired CPU and has substantially lower max latencies.

I'm reasonably confident that I've been able to maintain the interactivity people have come to expect from BFS in the transition to MuQSS now and have the data to support it above.

Enjoy!
お楽しみ下さい
-ck

linux-4.8-ck4, MuQSS CPU scheduler v0.116

Yet another bugfix release for MuQSS and the -ck patchset with one of the most substantial latency fixes yet. Everyone should upgrade if they're on a previous 4.8 patchset of mine. Sorry about the frequency of these releases but I just can't allow a known buggy release be the latest version.

4.8-ck4 patchset:
http://ck.kolivas.org/patches/4.0/4.8/4.8-ck4/

MuQSS by itself for 4.8:
4.8-sched-MuQSS_116.patch

MuQSS by itself for 4.7:
4.7-sched-MuQSS_116.patch

I'm hoping this is the release that allows me to not push any more -ck versions out till 4.9 is released since it addresses all remaining issues that I know about.

A lingering bug that has been troubling me for some time was leading to occasional massive latencies and thanks to some detective work by Serge Belyshev I was able to narrow it down to a single line fix which dramatically improves worst case latency when measured. Throughput is virtually unchanged. The flow-on effect to other areas was also apparent with sometimes unused CPU cycles and weird stalls on some workloads.

Sched_yield was reverted to the old BFS mechanism again which GPU drivers prefer but it wasn't working previously on MuQSS because of the first bug. The difference is substantial now and drivers (such as nvidia proprietary) and apps that use it a lot (such as the folding @ home client) behave much better now.

The late introduced bugs that got into ck3/muqss115 were reverted.

The results come up quite well now with interbench (my latency under load benchmark) which I have recently updated and should now give sensible values:

https://github.com/ckolivas/interbench

If you're baffled by interbench results, the most important number is %deadlines met which should be as close to 100% as possible followed by max latency which should be as low as possible for each section. In the near future I'll announce an official new release version.

Pedro in the comments section previously was using runqlat from bcc tools to test latencies as well, but after some investigation it became clear to me that the tool was buggy and did not work properly with bfs/muqss either so I've provided a slightly updated version here which should work properly:

runqlat.py

Enjoy!
お楽しみ下さい
-ck

linux-4.8-ck2, MuQSS version 0.114

Announcing an updated version, and the first -ck release with MuQSS as the scheduler, officially retiring BFS from further development, in line with the diminished rate of bug reports with MuQSS. It is clear that the little attention BFS had received over the years apart from rushed synchronisation with mainline had cause a number of bugs to creep in and MuQSS is basically a rewritten evolution of the same code so it makes no sense to maintain both.

http://ck.kolivas.org/patches/4.0/4.8/4.8-ck2/

MuQSS version 0.114 by itself:

4.8-sched-MuQSS_114.patch

Git tree includes branches for MuQSS and -ck:

https://github.com/ckolivas/linux

In addition to the most up to date version of MuQSS replacing BFS, this is the first release with BFQ included. It is configurable and is set by default in -ck though it is entirely optional.

The MuQSS changes since 112 are as follows:
- Added cacheline alignment to atomic variables courtesy of Holger Hoffstätte
- Fixed PPC build courtesy of Serge Belyshev.
- Implemented wake lists for separate CPU packages.
- Send hotplug threads to CPUs even if they're not alive yet since they'll be enabling them.
- Build fixes for uniprocessor.
- A substantial revamp of the sub-tick process accounting, decreasing the number of variables used, simplifying the code, and increasing the resolution to nanosecond accounting. Now even tasks that run for less than 100us will not escape visible accounting.

This release should bring slightly better performance, more so on multi-cpu machines, and fairer accounting/latency.

Enjoy!
お楽しみ下さい
-ck

First MuQSS Throughput Benchmarks

The short version graphical summary:

Red = MuQSS 112 interactive off
Purple = MuQSS 112 interactive on
Blue = CFS

The detail:
http://ck.kolivas.org/patches/muqss/Benchmarks/20161018/

I went on a journey looking for meaningful benchmarks to conduct to assess the scalability aspect as far as I could on my own 12x machine and was really quite depressed to see what the benchmark situation on linux is like. Only the old and completely invalid benchmarks seem to still be hanging around in public sites and promoted, like Reaim, aim7, dbench, volanomark, etc. and none of those are useful scalability benchmarks. Even more depressing was the only ones with any reputation are actually commercial benchmarks costing hundreds of dollars.

This made me wonder out loud just how the heck mainline is even doing scalability improvements if there are precious few valid benchmarks for linux and no one's using them. The most promising ones, like mosbench, need multiple machines and quite a bit of set up to get them going.

I spent a day wading through the phoronix test suite - a site and its suite not normally known for meaningful high performance computing discussion and benchmarks - looking for benchmarks that could be used for meaningful results for multicore scalability assessment and were not too difficult to deploy and came up with the following collection:

John The Ripper - a CPU bound application that is threaded to the number of CPUs and intermittently drops to one thread making for slightly more interesting behaviour than just a fully CPU bound workload.

7-Zip Compression - a valid real world CPU bound application that is threaded but rarely able to spread out to all CPUs making it an interesting light load benchmark.

ebizzy - This emulates a heavy content delivery server load which scales beyond the number of CPUs and emulates what goes on between a http server and database.

Timed Linux Kernel Compilation - A perennial favourite because it is a real world case and very easy to reproduce. Despite numerous complaints about its validity as a benchmark, it is surprisingly consistent in its results and tests many facets of scalability, though does not scale to use all CPUs at all time either.

C-Ray - A ray tracing benchmark that uses massive threading per CPU and is completely CPU bound but overloads all CPUs.

Primesieve - A prime number generator that is threaded to the number of CPUs exactly, is fully CPU bound and is cache intensive.

PostgreSQL pgbench - A meaningful database benchmark that is done at 3 different levels - single threaded, normal loaded and heavily contended, each testing different aspects of scalability.

And here is a set of results comparing 4.8.2 mainline (labelled CFS), MuQSS 112 in interactive mode (MuQSS-int1) and MuQSS 112 in non-interactive mode (MuQSS-int0):

http://ck.kolivas.org/patches/muqss/Benchmarks/20161018/

It's worth noting that there is quite a bit of variance in these benchmarks and some are bordering on the difference being just noise. However there is a clear pattern here - when the load is light, in terms of throughput, CFS outperforms MuQSS. When load is heavy, the heavier it gets, MuQSS outperforms CFS, especially in non-interactive mode. As a friend noted, for the workloads where you wouldn't be running MuQSS in interactive mode, such as a web server, database etc, non-interactive mode is of clear performance benefit. So at least on the hardware I had available to me, on a 12x machine, MuQSS is scaling better than mainline on these workloads as load increases.

The obvious question people will ask is why MuQSS doesn't perform better at light loads, and in fact I have an explanation. The reason is that mainline tends to cling to processes much more so that if it is hovering at low numbers of active processes, they'll all cluster on one CPU or fewer CPUs than being spread out everywhere. This means the CPU benefits more from the turbo modes virtually all newer CPUs have, but it comes at a cost. The latency to tasks is greater because they're competing for CPU time on fewer busy CPUs rather than spreading out to idle cores or threads. It is a design decision in MuQSS, as taken from BFS, to always spread out to any idle CPUs if they're available, to minimise latency, and that's one of the reasons for the interactivity and responsiveness of MuQSS. Of course I am still investigating ways of closing that gap further.

Hopefully I can get some more benchmarks from someone with even bigger hardware, and preferably with more than one physical package since that's when things really start getting interesting. All in all I'm very pleased with the performance of MuQSS in terms of scalability on these results, especially assuming I'm able to maintain the interactivity of BFS which were my dual goals.

There is MUCH more to benchmarking than pure throughput of CPU - which is almost the only thing these benchmarks is checking - but that's what I'm interested in here. I hope that providing my list of easy to use benchmarks and the reasoning behind them can generate interest in some kind of meaningful standard set of benchmarks. I did start out in kernel development originally after writing and being a benchmarker :P

To aid that, I'll give simple instructions here for how to ~imitate the benchmarks and get results like I've produced above.

Download the phoronix test suite from here:
http://www.phoronix-test-suite.com/

The generic tar.gz is perfectly fine. Then extract it and install the relevant benchmarks like so:

tar xf phoronix-test-suite-6.6.1.tar.gz
cd phoronix-test-suite
./phoronix-test-suite install build-linux-kernel c-ray compress-7zip ebizzy john-the-ripper pgbench primesieve
./phoronix-test-suite default-run build-linux-kernel c-ray compress-7zip ebizzy john-the-ripper pgbench primesieve


Now obviously this is not ideal since you shouldn't run benchmarks on a multiuser login with Xorg and all sorts of other crap running so I actually always run benchmarks at init level 1.

Enjoy!
お楽しみ下さい
-ck

MuQSS - The Multiple Queue Skiplist Scheduler v0.108

A new version of the MuQSS CPU scheduler

Incrementals and full patches available for 4.8 and 4.7 respectively here:
http://ck.kolivas.org/patches/muqss/4.0/4.8/


http://ck.kolivas.org/patches/muqss/4.0/4.7/

Yet more minor bugfixes and some important performance enhancements.

This version brings to the table the same locking scheme for trying to wake tasks up as mainline which is advantageous on process busy workloads and many CPUs. This is important because the main reason for moving to multiple runqueues was to minimise lock contention for the global runqueue lock that is in BFS (as mentioned here numerous times before) and this wake up scheme helps make the most of the multiple discrete runqueue locks.

Note this change is much more significant than the last releases so new instability is a possibility. Please report any problems or stacktraces!

There was a workload when I started out that I used lockstat to debug to get an idea of how much lock contention was going on and how long it lasted. Originally with the first incarnations of MuQSS on a 14 second benchmark with thousands of tasks on a 12x CPU it obtained 3 million locks and had almost 300k contentions with the longest contention lasting 80us. Now the same workload grabs the lock just 5k times with only 18 contentions in total and the longest lasted 1us.

This clearly demonstrates that the target endpoint for avoiding lock contention has been achieved. It does not translate into performance improvements on ordinary hardware today because you need ridiculous workloads on many CPUs to even begin deriving advantage from it. However as even our phones now have reached 8 logical CPUs, it will only be a matter of time before 16 threads appears on commodity hardware - a complaint that was directed at BFS when it came out 7 years ago but they still haven't appeared just yet. BFS was shown to be scalable for all workloads up to 16 CPUs, and beyond for certain workloads, but suffered dramatically for others. MuQSS now makes it possible for what was BFS to be useful much further into the future.

Again - MuQSS is aimed primarily at desktop/laptop/mobile device users for the best possible interactivity and responsiveness, and is still very simple in its approach to balancing workloads to CPUs so there are likely to be throughput workloads on mainline that outperform it, though there are almost certainly workloads where the opposite is true.

I've now addressed all planned changes to MuQSS and plan to hopefully only look at bug reports instead of further development from here on for a little while. In my eyes it is now stable enough to replace BFS in the next -ck release barring some unexpected showstopper bug appearing.

EDIT: If you blinked you missed the 107 announcement which was shortly superseded by 108.

EDIT2: Always watch the pending directory for updated pending patches to add.
http://ck.kolivas.org/patches/muqss/4.0/4.8/Pending/

Enjoy!
お楽しみ下さい
-ck

MuQSS - The Multiple Queue Skiplist Scheduler v0.106

Another day and time for yet another release.

There are 0.106 versions and incrementals available for linux-4.7:
 http://ck.kolivas.org/patches/muqss/4.0/4.7/
and linux-4.8:
http://ck.kolivas.org/patches/muqss/4.0/4.8


Two large remaining races that could lead to warnings, stalls, or in the worst case, crashes, have been fixed in this version.


Additionally the multiple-runqueue locking has been significantly optimised to take only the runqueues needed for as long as they're needed only and dropped as soon as possible which should bring the lock contention levels down even further. This is a performance enhancement, more so in non-interactive mode, though it will only start being demonstrable if you're lucky enough to have many CPUs.


This version addresses all the known bugs and warnings I've received to date so hopefully I can have a little rest and let people out there actually give it a go. What will you expect if you use this instead of BFS? If I've done this correctly, you will notice absolutely no difference since the idea was to preserve the interactivity and responsiveness of BFS and make it scalable to more CPUs than most people can afford.


Keep the feedback coming, thanks.

Enjoy!
 お楽しみ下さい
-ck

MuQSS - The Multiple Queue Skiplist Scheduler v0.105

I spent the last few days fighting with various lock debugging techniques and the numerous bug reports and am pleased to announce a new version of MuQSS, version 0.105

There are versions and incrementals available for linux-4.7:
 http://ck.kolivas.org/patches/muqss/4.0/4.7/
and linux-4.8:
http://ck.kolivas.org/patches/muqss/4.0/4.8

If you've been waiting for me to say it's stable enough to try, then now's your chance for I've addressed all known bugs at this time and it's working well for me.

Most of the issues were to do with races and unstable handling of cross-cpu task movement. No effort went into improving performance from 104 though this version should address many of the crashes and hangs that have been reported with earlier versions.

Additionally there is a pending patch being uploaded for BFS512 which, as per usual, had some last minute issues that only just showed up. If enough users complain loudly enough or more issues show up I might just release another bfs and -ck since it should be stable, especially being one of the last BFS releases.

http://ck.kolivas.org/patches/bfs/4.0/4.8/Pending/

Keep the feedback and bug reports coming. Next I need to put more care into the non-interactive mode of muqss for your enjoyment.

Enjoy!
お楽しみ下さい
-ck

BFS version 0.512, linux-4.8-ck1, MuQSS for linux-4.8

This is to announce an updated version of BFS for the new stable linux kernel 4.8.

BFS by itself:
4.8-sched-bfs-512.patch

-ck patches with BFS:
4.8-ck1

EDIT: Here's a bugfix post release for the above kernels that I highly recommend you include:
http://ck.kolivas.org/patches/bfs/4.0/4.8/Pending/bfs512-fixes.patch


Following on from the aggressive development towards a new scheduler, this BFS incorporates a number of fixes and performance improvements discovered while working on the Multiple Queue Skiplist Scheduler, MuQSS (pronounced mux) and should be the best performing BFS yet.

Note that this may be the last BFS based -ck release as MuQSS is designed to replace it, being the logical evolution of the same scheduler into a more scalable discrete runqueue design.

For those willing to try it in its current version, an incremental patch can be applied to BFS 512:

bfs512-muqss104.patch

or there is a full patch against 4.8:

4.8-sched-MuQSS_104.patch


Again, MuQSS is still immature code and while I have been running it stably for a few days now, and have spent a lot of time debugging locking issues and stability, it is not intended for production use just yet. Having said that, all testing is most welcome, especially benchmarks and stacktraces if you get any crashes.


I've been asked numerous times why I decided to change the name. There are two major reasons. The first is that it signifies just what a dramatic overhaul to the codebase it is, where it is virtually a new scheduler, even though it uses the same scheduling decision policy as BFS. The second is that I've had many people approach me saying they would like to use BFS for their own production environment but alas the offensive name is a showstopper for them. Additionally I had to choose a name that wasn't being used by anything else which both BFS and brainfuck had been used before.


Enjoy!
お楽しみ下さい
-ck

MuQSS - The Multiple Queue Skiplist Scheduler v0.105

Announcing a multiple runqueue variant of BFS, with the more mundane name of MuQSS (pronounced mux) for linux 4.7:

Full patch for linux-4.7
4.7-sched-MuQSS_105.patch

Keep watching this blog for newer versions!

Incremental to patch bfs502 to MuQSS 0.1:
bfs502-MuQSS_103.patch

It was inevitable that one day I would find myself tackling the 2 major scalability limitations in BFS and this is the result of it. These two issues were

1. The single runqueue which means all CPUs would fight for lock contention over the one runqueue, and
2. The O(n) look up which means linear increase in overhead for task lookups as number of processes increases.

As you're all aware by now, skiplists were recently introduced into BFS to tackle number 2 with a modest improvement in throughput at high loads.

Till now I did not have the energy nor time to try and find a solution for number 1. that maintained BFS' scheduling decision algorithm as the single runqueue was actually the reason latency remains bound and deterministic on BFS, capitalising with more CPUs instead of fighting against them for scalability.

This scheduler variant is an evolution of BFS, which hopefully will be mature enough to replace BFS one day when stability is assured. It is able to still use the same scheduling algorithm as BFS meaning latency and responsiveness remains as good as always, but with the per-CPU runqueue and discrete locking, it also means it will scale to any number of CPUs, as the mainline scheduler does.

It does NOT guarantee the best possible throughput as there still is virtually no complex balancing mechanism whatsoever, selecting tasks according to deadline primarily with only CPU cache distances being used to determine which idle CPU to go to, or in non-interactive mode, which overloaded CPU to pull from to fill an idle CPU.

It would be possible, with a lot of effort, to wedge the entire balancing algorithm for scalability from mainline into this, though it will probably offset the deterministic latency that makes it special.

This is a massive rewrite and consequently there are bound to still be race conditions and hidden bugs though I have been running it for a while now with reasonable stability. I'm putting this out there for the braver people to test. There's a lot more to document about it but for now let's just say, give it a try.

Please don't use any lock debugging as it will light up every possible complaint for the time being!

Regarding 4.8, for the time being I will still be releasing BFS for it and incorporate it into -ck

EDIT: Updated to version 0.105 with significant bugfixes.

Enjoy!
お楽しみ下さい
-ck

BFS 502, linux-4.7-ck5

With the fix for the last of the freezes with BFS497 becoming clearer and a number of other minor issues being attended to, such as build failures and minor improvements accumulating, I'm releasing a new BFS that combines all into yet another release, which should be the last of the releases for the 4.7 kernel.

BFS by itself:
4.7-sched-bfs-502.patch

-ck patches with BFS:
4.7-ck5

In addition to the update to BFS, this -ck release is the first in a very long time to include a patch from another developer - the Throttled background buffered writeback v7 patch by Jens Axboe. This makes a massive difference to a system's ability to read files, open new applications etc. under heavy write loads in my testing and is a change which I believe is essential and will eventually make its way into the mainline kernel.

The changes to BFS 502 are as follows:

bfs497-build_other_arches.patch
bfs497-no_smtload_avg.patch
bfs497-recognise_nodes2.patch
bfs497-revert-othercpufreq.patch
bfs497-fix_smt_nonice.patch  

• A build fix for building on other architectures (notably ARM).
• Simplifying the load measurement on SMT machines reported to cpufreq - trying to account for load on the SMT sibling is unnecessary as each core will run at the speed of the most loaded sibling anyway on any existing hardware.
• A fix for detecting CPUs on other NUMA nodes and setting their locality correctly.
• Not trying to signal CPU load to cpufreq on other CPUs when tasks migrate - this was leading to the hangs and there is enough rescheduling for cpufreq to get the load later on.
• A build fix for when SMT_NICE is not configured.

Enjoy!
お楽しみ下さい
-ck

BFS 497, linux-4.7-ck4

For the first time in a very long time, I'm announcing yet another -ck release up to ck4 along with yet more substantial updates for BFS for linux-4.7 based kernels.

BFS by itself:
4.7-sched-bfs-497.patch

-ck branded linux-4.7-ck4 patches:
linux-4.7-ck4

Thanks(?) to the massive changes to the mainline kernel I'd been forced to rewrite significant components of BFS to work properly with them, specifically the cpu frequency governors. At the same time I've had quite a bit of energy and enthusiasm for working on BFS in a way I haven't had in a long time. As a result, this updated version not only addresses the remaining cgroup stub patch bug (mentioned on the previous announcement) but implements further improvements and clean ups to go with those improvements.

Alas I still have no explanation for the random lockups some people are seeing, but I have seen reports of it happening on mainline kernels as well now, so while I'm always suspicious of my own code, there is also the chance that BFS exacerbates an issue in mainline. Something that appears common is onboard Intel graphics with the Haswell chipset.

Additionally I had reports of people being unable to suspend with BFS from 4.7 but I haven't heard back from them on later versions.

The short summary of improvements in this version are less overhead, higher throughput and less latencies.

I've rewritten the skiplist implementation to not require a malloc/free on insertion/removal of a new node which seemed to noticeably improve throughput at high loads.
Now that CPU frequency governors know what the scheduler is doing, the approach of BFS of old of knowing what the governor was doing and working around it is no longer helpful and I've removed the whole sticky task and offset for throttled CPUs and throughput has actually improved instead.
I've also added some micro-optimisations and cleanups.
I've added a minor change for offlining CPUs to prevent tasks trying to schedule to them.

The set of patches in ck4 is the largest in the ck patchset since the early 2.6 patchset days. I've also included the patch from Alfred (thanks!) to fix the warning that happens with suspend which is mostly harmless.

Each patch included has a mini changelog at the top.

I'm also keen to get feedback from people on if they see any noticeable interactive/responsiveness regressions by disabling the interactive flag as follows:

echo 0 > /proc/sys/kernel/interactive

Enjoy!
お楽しみ下さい
-ck

BFS 490, linux-4.7-ck3

Announcing yet another substantial update for BFS for linux-4.7 based kernels.

BFS by itself:
4.7-sched-bfs-490.patch

-ck branded linux-4.7-ck3 patches:
linux-4.7-ck3

Following on from the large update to BFS in 480 to skip lists, numerous regressions became apparent, the bulk of which were related to doing a poor job of signalling cpu load to the various cpufrequency governors. Some were affected badly, others not so, but there were plenty of helpful people giving feedback about those regressions which encouraged me to slowly but surely chip away at the problems. Additionally, there were some minor behavioural regressions which were oversights during the updates to BFS 480. Finally the rudimentary cgroup stub patch would crash the system.

As the number of patches required to address these issues got larger and larger, it became hard for people on this blog to keep up with the changes so I've released 490 which hopefully should address the bulk of these issues - there are patches in there that haven't been posted on this blog, but I've included all of them with a brief description in the incremental/ directory for your perusal.

Anyway it is much easier for people to grab the latest version which includes all of those changes, including the updated cgroups stub patch.

EDIT: Here's a patch to make cgroup stubs safer cgroup-stubs-safe2.patch

Enjoy!
お楽しみ下さい
-ck

Cgroup stubs for BFS

In addition to some minor pending changes for BFS 480 here:

Pending

I've implemented basic stubs for the CPU controller cgroups feature. This patch is experimental and does NOT implement the actual controller groups features, it simply creates a compatibility layer for the cgroup filesystem. The point of this is to allow environments and applications that refuse to work (such as docker), or work improperly without them, to work. While the actual control of CPU resources won't happen, there's a good chance that it won't make any difference whatsoever since their actual use on a desktop filesystem is serious overkill and worsens throughput. I don't have any plans to implement the actual CPU groups features. This is only lightly tested but already I've noticed that a laptop that would always take ages to shutdown with BFS is now much happier, so who knows it might help machines that refuse to suspend for the same reason:

bfs480-cgroup-stubs.patch

As an aside I did an experiment today with BFS480 on a machine with 12 logical cores and 64GB ram and ran a make -j allyesconfig at sched idleprio to see how it would scale and control the many idleprio tasks along with SMT nice. The machine was fine right up until it ran out of ram and then stalled while it evicted whatever it could and then continued. The load peaked and stayed around 7200 for 10 minutes. While the load was 7200, existing applications continued to work fine and browsing on firefox was virtually normal, but starting new applications took forever thanks to seriously delayed I/O.

EDIT: More testing shows this patch is UNSTABLE and can crash so just see it as proof-of-concept for now

Enjoy!
お楽しみ下さい
-ck

BFS 480 with skip lists, linux-4.7-ck2

Announcing a major update for BFS for linux-4.7 based kernels.

BFS by itself:
4.7-sched-bfs-480.patch

-ck branded linux-4.7-ck2 patches:
linux-4.7-ck2

This is the largest BFS update in a long time. The various problems that had been accumulating forced me to spend a more extended period fixing BFS to work with the latest mainline changes and encouraged me to overhaul some areas that had long been needing it.

The changes are:

• Fixed the crash when SMT NICE is configured in on a CPU without SMT.
• Added my skiplist implementation.
• Converted BFS from its long-standing O(n) lookup to use skiplists.
• Fix crash when SMT NICE is enabled on some hardware
• Fix try_preempt missing the locality diff effect in non-interactive mode
• Ignore busy threads/caches when still on the same core
• Reworked the testing of idle threads and cores for less overhead and to correctly identify idle siblings
• Fix the CPU load that's passed to the cpu frequency governor, fixing a crash and non-working schedutil governor.

The short summary is I've fixed a number of showstopper bugs on the last version, and improved throughput .

Actually incorporating the skiplists that I had experimented with a long time ago was decided on by the fact that I was able to trim the skiplist overhead further and maintain identical semantics for process selection (maintaining interactivity) whereas on the previous experiment I had never completed the work. Throughput testing shows virtually identical performance on normal workloads and theoretically would be helpful in extreme overload cases.

The original post regarding skip lists was here:
bfs-and-skip-lists.html


This now means that BFS is no longer O(n) lookup after O(1) insertion. It is now O(log(n)) insertion, O(1) lookup and O(k) removal where k <= 16, thereby tackling a long-standing criticism of the overall design.

I did not find a specific cause for peoples' inability to suspend to ram so I doubt this has been fixed despite the large code update.


The list of patches making up bfs480 is as follows:

bfs472-fix_set_task_cpu.patch
skiplists.patch
bfs472-skiplist.patch
bfs-delay-smt-siblings.patch
bfs-fix-noninteractive-try-preempt.patch
bfs-ignore_local_busy.patch
bfs-rework-idles.patch
bfs-fix-schedutil.patch
bfs-v480.patch

As always I'm giving this to you not long after I've finished coding it so all the usual warnings apply, especially with an update of this size.

EDIT: Uniprocessor build fix: bfs480-fix-upbuild.patch
EDIT2: Here is a test patch to try and improve cpufreq behaviour: bfs480-rework_cpufreq.patch

Enjoy!
お楽しみ下さい
-ck

BFS 472, linux-4.7-ck1

Announcing an updated BFS for linux-4.7 based kernels.

BFS by itself:
4.7-sched-bfs-472.patch

-ck branded linux-4.7-ck1 patches:
linux-4.7-ck1

This was quite a substantial merge effort this time around with a fair amount of changes in mainline kernel that affected the patch. Nonetheless everything appears to be working as planned in my limited testing. I'm unsure if the changes will fix the problems people had with suspend during the 4.6-bfs patches but the new code does touch that area. I was never affected on any of my machines so was unable to reproduce the problem in the first place.

In addition to the resync, a few minor changes have made their way into this release with respect to the way tasks preempt other tasks. See bfs470-updates.patch for details.

One other fairly significant change was properly hooking into the new schedutil parameters that drive cpufreq scaling governors. What I committed into bfs470 would not have been working properly in choosing the correct CPU frequency to run at and may have led to slowdowns and/or more power usage. This should be fixed in 472.

I should also mention that if, like me, you use the evil proprietary nvidia driver, the latest will not build with the current kernel and you'll need a couple of patches to get it working.

Enjoy!
お楽しみ下さい
-ck

EDIT: This patch will fix crashes when configured without SMT_NICE enabled:
bfs472-fix_set_task_cpu.patch
And will be applied to the next BFS release.

BFS 470, linux-4.6-ck1

Announcing an updated BFS for linux-4.6 based kernels.

BFS by itself:
4.6-sched-bfs-470.patch

-ck branded linux-4.6-ck1 patches:
linux-4.6-ck1

Resync to 4.6. You know the drill.


Enjoy!
お楽しみ下さい
-ck

BFS 469, linux-4.4-ck1, linux-4.5-ck1

Announcing an updated BFS for linux-4.4 and 4.5 based kernels.

BFS by itself:
4.5-sched-bfs-469.patch

-ck branded linux-4.5-ck1 patches:
linux-4.5-ck1



This is purely a resync of BFS 467 from 4.3-ck3 to the current kernels. The only change is extra documentation of the interactive tunable in the scheduler documentation, and a build warning fix for uniprocessor builds.


While linux-4.5 is the latest kernel, as I had been slow in syncing up and missed 4.4, and given that 4.4 is deemed a Long Term Stable release, I've provided resyncs with both. Version number differences of 467/469 are only due to syncing with different kernels and otherwise they are only trivially different.


The patches are fairly new without a great deal of testing, so the usual warnings apply, but given how long it took me to getting around to catching up, I didn't want to delay releasing them.


Enjoy!
お楽しみ下さい
-ck

BFS 467, linux-4.3-ck3

Announcing an updated BFS for linux-4.3 based kernels.

BFS by itself:
4.3-sched-bfs-467.patch

-ck branded linux-4.3-ck3 patches:
4.3-ck3

After my initial enthusiasm regarding the improved throughput on the previous BFS release, I unfortunately had some reports of regressions in interactive behaviour for the first time in a while, both on this forum and through other channels. So for the first time in a while, I've released a -ck3 with yet another updated BFS to address this since I can't stand having a dodgy release out for any extended period.

With this release what I've done is reinstate an old tunable that used to be on my scheduler patches many years ago

/proc/sys/kernel/interactive

This has two settings, 1 for on and 0 for off. By default - of course since this is BFS - it is set to 1. What it does is prioritise latency over throughput in mode 1 and vice versa in mode 0. In addition to addressing the latency issues in the previous kernel, mode 1 actually completely turns off all soft affinity scheduling in the kernel, for the lowest possible latencies all round, so this may be the first kernel with an improvement in latency in a while too.

Bear in mind none of these changes make any difference on uniprocessor kernels so there is no need for UP users to update unless they need the build fixes that came with BFS466+.

Amusingly enough, linux-4.3.2 wouldn't boot for me for unrelated reasons so I'm using 4.3.0-ck3 myself. So if you can't get 4.3.2 to boot, roll back.

Enjoy!
お楽しみください

BFS 466, linux-4.3-ck2

Announcing an updated BFS for linux-4.3 based kernels.

BFS by itself:
4.3-sched-bfs-466.patch

-ck branded linux-4.3-ck2 patches:
4.3-ck2

In addition to a build fix for the nohz compile issue with BFS 465, this is the first BFS in a very long time to have performance improvements. For some time now it's bugged me that tasks would have very poor affinity with CPUs on BFS, even if the performance was good. By this I mean if you fired up a fully CPU bound task and watched a CPU monitor/graph on a multicore machine, you'd see the one task would bounce around from CPU to CPU very frequently instead of occasionally. While this was great for latency purposes and interactivity, single threaded workloads would suffer as a result, and additionally it would represent a small amount of performance loss in multithreaded workloads too since CPU cache effects improving throughput would be diminished. Every time I'd previously tackled this issue, I found myself making some other workload worse.

After approximately 100 rebuilds of the kernel and benchmarking, I finally found where the problem lay, and it wasn't just trying to maintain bias against moving tasks from CPU to CPU, it was also that the code responsible is in the most frequently traversed code path in the schedule() call. Simplifying the code that biased against moving tasks in earliest_deadline_task, as well as calling on the bias for all tasks, not just fully CPU bound tasks, improved performance statistically significantly without detriment to latency or other workloads in my testing.

Apart from improving measurable throughput benchmarks, users may notice that some workloads that are single threaded (such as some video playback software, or even virtualisation with kvm etc.) may actually improve because of their ability to bind to one CPU better and not incur the wrath of being moved to a CPU speed throttled for power saving. Please give it a whorl and report back anything you find, positive or negative - though it should all be positive. If you have benchmarks you want to throw at it, even better.

EDIT: After the initial enthusiasm, it appears this DOES have a detrimental effect on interactivity so I will be looking for another change in the near future with yet another release.

Enjoy!
お楽しみください

BFS 465, linux-4.3-ck1

Finally a resync to mainline linux, with linux-4.3.

BFS by itself:

4.3-sched-bfs-465.patch

-ck branded linux-4.3-ck1 patches:

4.3-ck1 patches

In addition to the usual collection of resyncs and minor updates, this includes 3 patches courtesy of Alfred Chen who maintained the fort while I was too busy to work on a resync for linux-4.2. (THANKS!) His changes fix a warning that happens on some system on startup, hopefully fix the long standing hang on heavy file access with truly unlocked block flush unplugging, and a build problem.

With such a large resync and update, the usual warnings apply regarding instability, file system corruption and unwanted impregnations.

EDIT: Build fix: bfs465-nohz-buildfix.patch

Enjoy!
お楽しみください