lkml.org 
[lkml]   [2019]   [Apr]   [25]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [RFC PATCH v2 00/17] Core scheduling v2

* Mel Gorman <mgorman@techsingularity.net> wrote:

> On Thu, Apr 25, 2019 at 11:55:08AM +0200, Ingo Molnar wrote:
> > > > Would it be possible to post the results with HT off as well ?
> > >
> > > What's the point here to turn HT off? The latency is sensitive to the
> > > relationship
> > > between the task number and CPU number. Usually less CPU number, more run
> > > queue wait time, and worse result.
> >
> > HT-off numbers are mandatory: turning HT off is by far the simplest way
> > to solve the security bugs in these CPUs.
> >
> > Any core-scheduling solution *must* perform better than HT-off for all
> > relevant workloads, otherwise what's the point?
> >
>
> I agree. Not only should HT-off be evaluated but it should properly
> evaluate for different levels of machine utilisation to get a complete
> picture.
>
> Around the same time this was first posted and because of kernel
> warnings from L1TF, I did a preliminary evaluation of HT On vs HT Off
> using nosmt -- this is sub-optimal in itself but it was convenient. The
> conventional wisdom that HT gets a 30% boost appears to be primarily based
> on academic papers evaluating HPC workloads on a Pentium 4 with a focus
> on embarassingly parallel problems which is the ideal case for HT but not
> the universal case. The conventional wisdom is questionable at best. The
> only modern comparisons I could find were focused on games primarily
> which I think hit scaling limits before HT is a factor in some cases.
>
> I don't have the data in a format that can be present everything in a clear
> format but here is an attempt anyway. This is long but the central point
> that when when a machine is lightly loaded, HT Off generally performs
> better than HT On and even when heavily utilised, it's still not a
> guaranteed loss. I only suggest reading after this if you have coffee
> and time. Ideally all this would be updated with a comparison to core
> scheduling but I may not get it queued on my test grid before I leave
> for LSF/MM and besides, the authors pushing this feature should be able
> to provide supporting data justifying the complexity of the series.

BTW., a side note: I'd suggest introducing a runtime toggle 'nosmt'
facility, i.e. turn a system between SMT and non-SMT execution runtime,
with full reversability between these states and no restrictions.

That should make both benchmarking more convenient (no kernel reboots and
kernel parameters to check), and it would also make it easier for system
administrators to experiment with how SMT and no-SMT affects their
typical workloads.

> Here is a tbench comparison scaling from a low thread count to a high
> thread count. I picked tbench because it's relatively uncomplicated and
> tends to be reasonable at spotting scheduler regressions. The kernel
> version is old but for the purposes of this discussion, it doesn't matter
>
> 1-socket Skylake (8 logical CPUs HT On, 4 logical CPUs HT Off)

Side question: while obviously most of the core-sched interest is
concentrated around Intel's HyperThreading SMT, I'm wondering whether you
have any data regarding AMD systems - in particular Ryzen based CPUs
appear to have a pretty robust SMT implementation.

> Hmean 1 484.00 ( 0.00%) 519.95 * 7.43%*
> Hmean 2 925.02 ( 0.00%) 1022.28 * 10.51%*
> Hmean 4 1730.34 ( 0.00%) 2029.81 * 17.31%*
> Hmean 8 2883.57 ( 0.00%) 2040.89 * -29.22%*
> Hmean 16 2830.61 ( 0.00%) 2039.74 * -27.94%*
> Hmean 32 2855.54 ( 0.00%) 2042.70 * -28.47%*
> Stddev 1 1.16 ( 0.00%) 0.62 ( 46.43%)
> Stddev 2 1.31 ( 0.00%) 1.00 ( 23.32%)
> Stddev 4 4.89 ( 0.00%) 12.86 (-163.14%)
> Stddev 8 4.30 ( 0.00%) 2.53 ( 40.99%)
> Stddev 16 3.38 ( 0.00%) 5.92 ( -75.08%)
> Stddev 32 5.47 ( 0.00%) 14.28 (-160.77%)
>
> Note that disabling HT performs better when cores are available but hits
> scaling limits past 4 CPUs when the machine is saturated with HT off.
> It's similar with 2 sockets
>
> 2-socket Broadwell (80 logical CPUs HT On, 40 logical CPUs HT Off)
>
> smt nosmt
> Hmean 1 514.28 ( 0.00%) 540.90 * 5.18%*
> Hmean 2 982.19 ( 0.00%) 1042.98 * 6.19%*
> Hmean 4 1820.02 ( 0.00%) 1943.38 * 6.78%*
> Hmean 8 3356.73 ( 0.00%) 3655.92 * 8.91%*
> Hmean 16 6240.53 ( 0.00%) 7057.57 * 13.09%*
> Hmean 32 10584.60 ( 0.00%) 15934.82 * 50.55%*
> Hmean 64 24967.92 ( 0.00%) 21103.79 * -15.48%*
> Hmean 128 27106.28 ( 0.00%) 20822.46 * -23.18%*
> Hmean 256 28345.15 ( 0.00%) 21625.67 * -23.71%*
> Hmean 320 28358.54 ( 0.00%) 21768.70 * -23.24%*
> Stddev 1 2.10 ( 0.00%) 3.44 ( -63.59%)
> Stddev 2 2.46 ( 0.00%) 4.83 ( -95.91%)
> Stddev 4 7.57 ( 0.00%) 6.14 ( 18.86%)
> Stddev 8 6.53 ( 0.00%) 11.80 ( -80.79%)
> Stddev 16 11.23 ( 0.00%) 16.03 ( -42.74%)
> Stddev 32 18.99 ( 0.00%) 22.04 ( -16.10%)
> Stddev 64 10.86 ( 0.00%) 14.31 ( -31.71%)
> Stddev 128 25.10 ( 0.00%) 16.08 ( 35.93%)
> Stddev 256 29.95 ( 0.00%) 71.39 (-138.36%)
>
> Same -- performance is better until the machine gets saturated and
> disabling HT hits scaling limits earlier.

Interesting. This strongly suggests sub-optimal SMT-scheduling in the
non-saturated HT case, i.e. a scheduler balancing bug.

As long as loads are clearly below the physical cores count (which they
are in the early phases of your table) the scheduler should spread tasks
without overlapping two tasks on the same core.

Clearly it doesn't.

> SpecJBB 2005 is ancient but it does lend itself to easily scaling the
> number of active tasks so here is a sample of the performance as
> utilisation ramped up to saturation
>
> 2-socket
> Hmean tput-1 48655.00 ( 0.00%) 48762.00 * 0.22%*
> Hmean tput-8 387341.00 ( 0.00%) 390062.00 * 0.70%*
> Hmean tput-15 660993.00 ( 0.00%) 659832.00 * -0.18%*
> Hmean tput-22 916898.00 ( 0.00%) 913570.00 * -0.36%*
> Hmean tput-29 1178601.00 ( 0.00%) 1169843.00 * -0.74%*
> Hmean tput-36 1292377.00 ( 0.00%) 1387003.00 * 7.32%*
> Hmean tput-43 1458913.00 ( 0.00%) 1508172.00 * 3.38%*
> Hmean tput-50 1411975.00 ( 0.00%) 1513536.00 * 7.19%*
> Hmean tput-57 1417937.00 ( 0.00%) 1495513.00 * 5.47%*
> Hmean tput-64 1396242.00 ( 0.00%) 1477433.00 * 5.81%*
> Hmean tput-71 1349055.00 ( 0.00%) 1472856.00 * 9.18%*
> Hmean tput-78 1265738.00 ( 0.00%) 1453846.00 * 14.86%*
> Hmean tput-79 1307367.00 ( 0.00%) 1446572.00 * 10.65%*
> Hmean tput-80 1309718.00 ( 0.00%) 1449384.00 * 10.66%*
>
> This was the most surprising result -- HT off was generally a benefit
> even when the counts were higher than the available CPUs and I'm not
> sure why. It's also interesting with HT off that the chances of keeping
> a workload local to a node are reduced as a socket gets saturated earlier
> but the load balancer is generally moving tasks around and NUMA Balancing
> is also in play. Still, it shows that disabling HT is not a universal loss.

Interesting indeed. Could there be some batch execution benefit, i.e. by
having fewer CPUs to execute on the tasks do not crowd out and trash
die/socket level caches as badly? With no-HT the workload had more
threads than CPUs to execute on and the tasks were forced into neat
queues of execution and cache trashing would be limited to the short
period after a task was scheduled in?

If this was on the 40-physical-core Broadwell system and the 'X' tput-X
roughly correlates to CPU utilization then this seems plausible, as the
improvements start roughly at the ~tput-40 bondary and increase
afterwards.

> netperf is inherently about two tasks. For UDP_STREAM, it shows almost
> no difference and it's within noise. TCP_STREAM was interesting
>
> Hmean 64 1154.23 ( 0.00%) 1162.69 * 0.73%*
> Hmean 128 2194.67 ( 0.00%) 2230.90 * 1.65%*
> Hmean 256 3867.89 ( 0.00%) 3929.99 * 1.61%*
> Hmean 1024 12714.52 ( 0.00%) 12913.81 * 1.57%*
> Hmean 2048 21141.11 ( 0.00%) 21266.89 ( 0.59%)
> Hmean 3312 27945.71 ( 0.00%) 28354.82 ( 1.46%)
> Hmean 4096 30594.24 ( 0.00%) 30666.15 ( 0.24%)
> Hmean 8192 37462.58 ( 0.00%) 36901.45 ( -1.50%)
> Hmean 16384 42947.02 ( 0.00%) 43565.98 * 1.44%*
> Stddev 64 2.21 ( 0.00%) 4.02 ( -81.62%)
> Stddev 128 18.45 ( 0.00%) 11.11 ( 39.79%)
> Stddev 256 30.84 ( 0.00%) 22.10 ( 28.33%)
> Stddev 1024 141.46 ( 0.00%) 56.54 ( 60.03%)
> Stddev 2048 200.39 ( 0.00%) 75.56 ( 62.29%)
> Stddev 3312 411.11 ( 0.00%) 286.97 ( 30.20%)
> Stddev 4096 299.86 ( 0.00%) 322.44 ( -7.53%)
> Stddev 8192 418.80 ( 0.00%) 635.63 ( -51.77%)
> Stddev 16384 661.57 ( 0.00%) 206.73 ( 68.75%)
>
> The performance difference is marginal but variance is much reduced
> by disabling HT. Now, it's important to note that this particular test
> did not control for c-states and it did not bind tasks so there are a
> lot of potential sources of noise. I didn't control for them because
> I don't think many normal users would properly take concerns like that
> into account. MMtests is able to control for those factors so it could
> be independently checked.

Interesting. This too suggests suboptimal scheduling: with just 2 tasks
there might be two major modes of execution: either the two tasks end up
on the same physical core or not. If the scheduler isn't entirely
consistent about this choice then we might see big variations in
execution, depending on whether running the two tasks on different
physical cores is better to performance or not.

This stddev artifact could be narrowed down further by using taskset to
force the benchmark on 2 logical CPUs, and by making those 2 CPUs HT
siblings or not we could see which execution is the more optimal one.

My prediction, which is easily falsifiable is that stddev noise should
reduce dramatically in such a 2-CPU restricted 'taskset' based affinity
jail, *regardless* of whether the two CPUs are actually on the same
physical core or not.

> hackbench is the most obvious loser. This is for processes communicating
> via pipes.
>
> Amean 1 0.7343 ( 0.00%) 1.1377 * -54.93%*
> Amean 4 1.1647 ( 0.00%) 2.1543 * -84.97%*
> Amean 7 1.6770 ( 0.00%) 3.1300 * -86.64%*
> Amean 12 2.4500 ( 0.00%) 4.6447 * -89.58%*
> Amean 21 3.9927 ( 0.00%) 6.8250 * -70.94%*
> Amean 30 5.5320 ( 0.00%) 8.6433 * -56.24%*
> Amean 48 8.4723 ( 0.00%) 12.1890 * -43.87%*
> Amean 79 12.3760 ( 0.00%) 17.8347 * -44.11%*
> Amean 110 16.0257 ( 0.00%) 23.1373 * -44.38%*
> Amean 141 20.7070 ( 0.00%) 29.8537 * -44.17%*
> Amean 172 25.1507 ( 0.00%) 37.4830 * -49.03%*
> Amean 203 28.5303 ( 0.00%) 43.5220 * -52.55%*
> Amean 234 33.8233 ( 0.00%) 51.5403 * -52.38%*
> Amean 265 37.8703 ( 0.00%) 58.1860 * -53.65%*
> Amean 296 43.8303 ( 0.00%) 64.9223 * -48.12%*
> Stddev 1 0.0040 ( 0.00%) 0.0117 (-189.97%)
> Stddev 4 0.0046 ( 0.00%) 0.0766 (-1557.56%)
> Stddev 7 0.0333 ( 0.00%) 0.0991 (-197.83%)
> Stddev 12 0.0425 ( 0.00%) 0.1303 (-206.90%)
> Stddev 21 0.0337 ( 0.00%) 0.4138 (-1127.60%)
> Stddev 30 0.0295 ( 0.00%) 0.1551 (-424.94%)
> Stddev 48 0.0445 ( 0.00%) 0.2056 (-361.71%)
> Stddev 79 0.0350 ( 0.00%) 0.4118 (-1076.56%)
> Stddev 110 0.0655 ( 0.00%) 0.3685 (-462.72%)
> Stddev 141 0.3670 ( 0.00%) 0.5488 ( -49.55%)
> Stddev 172 0.7375 ( 0.00%) 1.0806 ( -46.52%)
> Stddev 203 0.0817 ( 0.00%) 1.6920 (-1970.11%)
> Stddev 234 0.8210 ( 0.00%) 1.4036 ( -70.97%)
> Stddev 265 0.9337 ( 0.00%) 1.1025 ( -18.08%)
> Stddev 296 1.5688 ( 0.00%) 0.4154 ( 73.52%)
>
> The problem with hackbench is that "1" above doesn't represent 1 task,
> it represents 1 group and so the machine gets saturated relatively
> quickly and it's super sensitive to cores being idle and available to
> make quick progress.

hackbench is also super sensitive to the same group of ~20 tasks being
able to progress at once, and hence is pretty noisy.

The flip-over between hackbench being able to progress effectively and a
half-scheduled group hindering all the others seems to be super
non-deterministic and can be triggered by random events both within
hackbench, and other things happening on the machine.

So while hackbench is somewhat artificial in its intensity and load
levels, it still matches messaging server peak loads so it's still
consider it an imporant metric of scheduling quality.

I'm wondering whether the scheduler could do anything to reduce the
non-determinism of hackbench.

BTW., note that 'perf bench scheduling' is a hackbench work-alike:

dagon:~/tip> perf bench sched messaging
# Running 'sched/messaging' benchmark:
# 20 sender and receiver processes per group
# 10 groups == 400 processes run

Total time: 0.158 [sec]

It also has a threaded variant (which is a hackbench-pthread work-alike):

dagon:~/tip> perf bench sched messaging --thread --group 20
# Running 'sched/messaging' benchmark:
# 20 sender and receiver threads per group
# 20 groups == 800 threads run

Total time: 0.265 [sec]

I'm trying to distill the most important scheduler micro-benchmarks into
'perf bench':

dagon:~/tip> perf bench sched

# List of available benchmarks for collection 'sched':

messaging: Benchmark for scheduling and IPC
pipe: Benchmark for pipe() between two processes
all: Run all scheduler benchmarks

which is still stuck at a very low count of 2 benchmarks currently.
:-)

> Kernel building which is all anyone ever cares about is a mixed bag
>
> 1-socket
> Amean elsp-2 420.45 ( 0.00%) 240.80 * 42.73%*
> Amean elsp-4 363.54 ( 0.00%) 135.09 * 62.84%*
> Amean elsp-8 105.40 ( 0.00%) 131.46 * -24.73%*
> Amean elsp-16 106.61 ( 0.00%) 133.57 * -25.29%*
>
> 2-socket
> Amean elsp-2 406.76 ( 0.00%) 448.57 ( -10.28%)
> Amean elsp-4 235.22 ( 0.00%) 289.48 ( -23.07%)
> Amean elsp-8 152.36 ( 0.00%) 116.76 ( 23.37%)
> Amean elsp-16 64.50 ( 0.00%) 52.12 * 19.20%*
> Amean elsp-32 30.28 ( 0.00%) 28.24 * 6.74%*
> Amean elsp-64 21.67 ( 0.00%) 23.00 * -6.13%*
> Amean elsp-128 20.57 ( 0.00%) 23.57 * -14.60%*
> Amean elsp-160 20.64 ( 0.00%) 23.63 * -14.50%*
> Stddev elsp-2 75.35 ( 0.00%) 35.00 ( 53.55%)
> Stddev elsp-4 71.12 ( 0.00%) 86.09 ( -21.05%)
> Stddev elsp-8 43.05 ( 0.00%) 10.67 ( 75.22%)
> Stddev elsp-16 4.08 ( 0.00%) 2.31 ( 43.41%)
> Stddev elsp-32 0.51 ( 0.00%) 0.76 ( -48.60%)
> Stddev elsp-64 0.38 ( 0.00%) 0.61 ( -60.72%)
> Stddev elsp-128 0.13 ( 0.00%) 0.41 (-207.53%)
> Stddev elsp-160 0.08 ( 0.00%) 0.20 (-147.93%)
>
> 1-socket matches other patterns, the 2-socket was weird. Variability was
> nuts for low number of jobs. It's also not universal. I had tested in a
> 2-socket Haswell machine and it showed different results
>
> Amean elsp-2 447.91 ( 0.00%) 467.43 ( -4.36%)
> Amean elsp-4 284.47 ( 0.00%) 248.37 ( 12.69%)
> Amean elsp-8 166.20 ( 0.00%) 129.23 ( 22.24%)
> Amean elsp-16 63.89 ( 0.00%) 55.63 * 12.93%*
> Amean elsp-32 36.80 ( 0.00%) 35.87 * 2.54%*
> Amean elsp-64 30.97 ( 0.00%) 36.94 * -19.28%*
> Amean elsp-96 31.66 ( 0.00%) 37.32 * -17.89%*
> Stddev elsp-2 58.08 ( 0.00%) 57.93 ( 0.25%)
> Stddev elsp-4 65.31 ( 0.00%) 41.56 ( 36.36%)
> Stddev elsp-8 68.32 ( 0.00%) 15.61 ( 77.15%)
> Stddev elsp-16 3.68 ( 0.00%) 2.43 ( 33.87%)
> Stddev elsp-32 0.29 ( 0.00%) 0.97 (-239.75%)
> Stddev elsp-64 0.36 ( 0.00%) 0.24 ( 32.10%)
> Stddev elsp-96 0.30 ( 0.00%) 0.31 ( -5.11%)
>
> Still not a perfect match to the general pattern for 2 build jobs and a
> bit variable but otherwise the pattern holds -- performs better until the
> machine is saturated. Kernel builds (or compilation builds) are always a
> bit off as a benchmark as it has a mix of parallel and serialised tasks
> that are non-deterministic.

Interesting.

Here too I'm wondering whether the scheduler could do something to
improve the saturated case: which *is* an important workload, as kernel
hackers tend to over-load their systems a bit when building kernel, to
make sure the system is at least 100% utilized. ;-)

Probably not though, without injecting too much policy. We could perhaps
repurpose SCHED_BATCH to be even more batch scheduling, i.e. to reduce
the non-determinism of the over-loaded machines in an even more assertive
manner? As long as there's enough RAM and no serios async IO the kbuild
gets perturbed by (which should be true these days) this should be
possible to do. We could then stick SCHED_BATCH into the kbuild process,
to make more than 0.01% of kernel developers use it. Win-win. :-)

> With the NASA Parallel Benchmark (NPB, aka NAS) it's trickier to do a
> valid comparison. Over-saturating NAS decimates performance but there
> are limits on the exact thread counts that can be used for MPI. OpenMP
> is less restrictive but here is an MPI comparison anyway comparing a
> fully loaded HT On with fully loaded HT Off -- this is crucial, HT Off
> has half the level of parallelisation
>
> Amean bt 771.15 ( 0.00%) 926.98 * -20.21%*
> Amean cg 445.92 ( 0.00%) 465.65 * -4.42%*
> Amean ep 70.01 ( 0.00%) 97.15 * -38.76%*
> Amean is 16.75 ( 0.00%) 19.08 * -13.95%*
> Amean lu 882.84 ( 0.00%) 902.60 * -2.24%*
> Amean mg 84.10 ( 0.00%) 95.95 * -14.10%*
> Amean sp 1353.88 ( 0.00%) 1372.23 * -1.36%*
>
> ep is the embarassingly parallel problem and it shows with half the cores
> with HT off, we take a 38.76% performance hit. However, even that is not
> universally true as cg for example did not parallelise as well and only
> performacne 4.42% worse even with HT off.

Very interesting. I'm wondering what kind of workload 'ep' is exactly,
and would love to have a work-alike in 'perf sched bench'.

Do these benchmarks over-saturate by default, and is this really
representative of how all the large compute cluster folks are *using*
MPI?

I thought the more common pattern was to closely tailor MPI parallelism
to available (logical) cores parallelism, to minimize shared cache
trashing in an oversubscribed scenario, but I could be wrong.

> I can show a comparison with equal levels of parallelisation but with
> HT off, it is a completely broken configuration and I do not think a
> comparison like that makes any sense.

I would still be interested in that comparison, because I'd like
to learn whether there's any true *inherent* performance advantage to
HyperThreading for that particular workload, for exactly tuned
parallelism.

Even if nobody is going to run the NPB/NAS benchmark that way.

> I didn't do any comparison that could represent Cloud. However, I think
> it's worth noting that HT may be popular there for packing lots of virtual
> machines onto a single host and over-subscribing. HT would intuitively
> have an advantage there *but* it depends heavily on the utilisation and
> whether there is sustained VCPU activity where the number of active VCPUs
> exceeds physical CPUs when HT is off. There is also the question whether
> performance even matters on such configurations but anything cloud related
> will be "how long is a piece of string" and "it depends".

Intuitively I'd guess that because all the cloud providers are pushing
for core-sched HT is probably a win in cloud benchmarks, if not for the
pesky security problems. ;-)

> So there you have it, HT Off is not a guaranteed loss and can be a gain
> so it should be considered as an alternative to core scheduling. The case
> where HT makes a big difference is when a workload is CPU or memory bound
> and the number of active tasks exceeds the number of CPUs on a socket
> and again when number of active tasks exceeds the number of CPUs in the
> whole machine.

Fascinating measurements, thanks a lot Mel for doing these!

This is super useful.

Thanks,

Ingo

\
 
 \ /
  Last update: 2019-04-25 20:53    [W:0.178 / U:10.928 seconds]
©2003-2018 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site