<div><br></div><div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Mon, Jul 4, 2022 at 1:13 PM Wojciech Kudla <<a href="mailto:wkudla.kernel@gmail.com">wkudla.kernel@gmail.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-style:solid;padding-left:1ex;border-left-color:rgb(204,204,204)"><div dir="ltr"><div><div><div>Thanks for your input, Vitaly. I'd be interested to find out more about the nature of the HW noise you observed in your benchmarks as our results were very consistent and it was pretty straightforward to pinpoint the culprit as JNI call overhead. Maybe it was just easier for us because we disallow C- and P-state transitions and put a lot of effort to eliminate platform jitter in general. Were you maybe running on a CPU model that doesn't support constant TSC? I would also suggest retrying with LAPIC interrupts suppressed (with: cli/sti) to maybe see if it's the kernel and not the hardware.</div></div></div></div></blockquote><div dir="auto">This was on a Broadwell Xeon chipset with constant tsc. All the typical jitter sources were reduced: C/P states disabled in bios, max turbo enabled, IRQs steered away, core isolated, etc. By the way, by noise I don’t mean the results themselves were noisy - they were constant run to run. I just meant the delta between normal vs critical JNI entrypoints was very minimal - ie “in the noise”, particularly with rdtsc.</div><div dir="auto"><br></div><div dir="auto">I can try to remeasure on newer Intel but see below …</div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-style:solid;padding-left:1ex;border-left-color:rgb(204,204,204)"><div dir="ltr"><div><div><div dir="auto"><br><br></div>100% agree on rdtsc(p) and snippets. There are some narrow usecases were one can get some substantial speed ups with direct access to prefetch or by abusing misprediction to keep icache hot. These scenarios are sadly only available with inline assembly. I know of a few shops that go to the length of forking Graal, etc to achieve that but am quite convinced such capabilities would be welcome and utilized by many more groups if they were easily accessible from java.</div></div></div></blockquote><div dir="auto">I’m of the firm (and perhaps controversial for some :)) opinion these days that Java is simply the wrong platform/tool for low latency cases that warrant this level of control. There’re very strong headwinds even outside of JNI costs. And the “real” problem with JNI, besides transition costs, is lack of inlining into the native calls. So even if JVM transition costs are fully eliminated, there’s still an optimization fence due to lost inlining (not unlike native code calling native fns via shared libs).</div><div dir="auto"><br></div><div dir="auto">That’s not say that perf regressions are welcomed - nobody likes those :).</div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-style:solid;padding-left:1ex;border-left-color:rgb(204,204,204)"><div dir="ltr"><div><div dir="auto"><br><br></div>Thanks,<br></div>W.<br></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Mon, Jul 4, 2022 at 5:51 PM Vitaly Davidovich <<a href="mailto:vitalyd@gmail.com" target="_blank">vitalyd@gmail.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-style:solid;padding-left:1ex;border-left-color:rgb(204,204,204)"><div dir="auto">I’d add rdtsc(p) wrapper functions to the list. These are usually either inline asm or compiler intrinsic in the JNI entrypoint. In addition, any native libs exposed via JNI that have “trivial” functions are also candidates for faster calling conventions. There’re sometimes way to mitigate the call overhead (eg batching) but it’s not always feasible.</div><div dir="auto"><br></div><div dir="auto">I’ll add that last time I tried to measure the improvement of Java criticals for clock_gettime (and rdtsc) it looked to be in the noise on the hardware I was testing on. It got the point where I had to instrument the critical and normal JNI entrypoints to confirm the critical was being hit. The critical calling convention isn’t significantly different *if* basic primitives (or no args at all) are passed as args. JNIEnv*, IIRC, is loaded from a register so that’s minor. jclass (for static calls, which is what’s relevant here) should be a compiled constant. Critical call still has a GCLocker check. So I’m not actually sure what the significant difference is for “lightweight” (ie few primitive or no args, primitive return types) calls.</div><div dir="auto"><br></div><div dir="auto">In general, I do think it’d be nice if there was a faster native call sequence, even if it comes with a caveat emptor and/or special requirements on the callee (not unlike the requirements for criticals). I think Vladimir Ivanov was working on “snippets” that allowed dynamic construction of a native call, possibly including assembly. Not sure where that exploration is these days, but that would be a welcome capability.</div><div dir="auto"><br></div><div dir="auto">My $.02. Happy 4th of July for those celebrating!</div><div dir="auto"><br></div><div dir="auto">Vitaly</div><div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Mon, Jul 4, 2022 at 12:04 PM Maurizio Cimadamore <<a href="mailto:maurizio.cimadamore@oracle.com" target="_blank">maurizio.cimadamore@oracle.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-style:solid;padding-left:1ex;border-left-color:rgb(204,204,204)">
<div>
<p>Hi,<br>
while I'm not an expert with some of the IO calls you mention
(some of my colleagues are more knowledgeable in this area, so I'm
sure they will have more info), my general sense is that, as with
getrusage, if there is a system call involved, you already pay a
hefty price for the user to kernel transition. On my machine this
seem to cost around 200ns. In these cases, using JNI critical to
shave off a dozen of nanoseconds (at best!) seems just not worth
it.</p>
<p>So, of the functions in your list, the ones in which I *believe*
dropping transitions would have the most effect are (if we exclude
getpid, for which another approach is possible) clock_gettime and
getcpu, I believe, as they might use vdso [1], which typically
brings the performance of these call closer to calls to shared lib
functions.<br>
</p>
<p>If you have examples e.g. where performance of recvmsg (or
related calls) varies significantly between base JNI and critical
JNI, please send them our way; I'm sure some of my colleagues
would be intersted to take a look.<br>
</p>
<p>Popping back a couple of levels, I think it would be helpful to
also define what's an acceptable regression in this context. Of
course, in an ideal world, we'd like to see no performance
regression at all. But JNI critical is an unsupported interface,
which might misbehave with modern garbage collectors (e.g. ZGC)
and that requires quite a bit of internal complexity which might,
in the medium/long run, hinder the evolution of the Java platform
(all these things have _some_ cost, even if the cost is not
directly material to developers). In this vein, I think calls like
clock_gettime tend to be more problematic: as they complete very
quickly, you see the cost of transitions a lot more. In other
cases, where syscalls are involved, the cost associated to
transitions are more likely to be "in the noise". Of course if we
look at absolute numbers, dropping transitions would always yield
"faster" code; but at the same time, going from 250ns to 245ns is
very unlikely to result in visible performance difference when
considering an application as a whole, so I think it's critical
here to decide _which_ use cases to prioritize.<br>
</p>
<p>I think a good outcome of this discussion would be if we could
come to some shared understanding of which native calls are truly
problematic (e.g. clock_gettime-like), and then for the JDK to
provide better (and more maintainable) alternatives for those
(which might even be faster than using critical JNI).<br>
</p>
<p>Thanks<br>
Maurizio<br>
</p>
<p>[1] - <a href="https://man7.org/linux/man-pages/man7/vdso.7.html" target="_blank">https://man7.org/linux/man-pages/man7/vdso.7.html</a><br>
</p></div><div>
<div>On 04/07/2022 12:23, Wojciech Kudla
wrote:<br>
</div>
<blockquote type="cite">
<div dir="ltr">
<div>
<div>Thanks Maurizio,<br>
<br>
</div>
I raised this case mainly about clock_gettime and
recvmsg/sendmsg, I think we're focusing on the wrong things
here. Feel free to drop the two syscalls from the discussion
entirely, but the main usecases I have been presenting
throughout this thread definitely stand.<br>
<br>
</div>
<div>Thanks<br>
</div>
<br>
</div>
<br>
<div class="gmail_quote">
<div dir="ltr" class="gmail_attr">On Mon, Jul 4, 2022 at 10:54
AM Maurizio Cimadamore <<a href="mailto:maurizio.cimadamore@oracle.com" target="_blank">maurizio.cimadamore@oracle.com</a>>
wrote:<br>
</div>
<blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-style:solid;padding-left:1ex;border-left-color:rgb(204,204,204)">
<div>
<p>Hi Wojtek,<br>
thanks for sharing this list, I think this is a good
starting point to understand more about your use case.</p>
<p>Last week I've been looking at "getrusage" (as you
mentioned it in an earlier email), and I was surprised to
see that the call took a pointer to a (fairly big) struct
which then needed to be initialized with some thread-local
state:</p>
<p><a href="https://man7.org/linux/man-pages/man2/getrusage.2.html" target="_blank">https://man7.org/linux/man-pages/man2/getrusage.2.html</a></p>
<p>I've looked at the implementation, and it seems to be
doing memset on the user-provided struct pointer, plus all
the fields assignment. Eyeballing the implementation, this
does not seem to me like a "classic" use case where
dropping transition would help much. I mean, surely
dropping transitions would help shaving some nanoseconds
off the call, but it doesn't seem to me that the call
would be shortlived enough to make a difference. Do you
have some benchmarks on this one? I did some [1] and the
call overhead seemed to come up at 260ns/op - w/o
transition you might perhaps be able to get to 250ns, but
that's in the noise?<br>
</p>
<p>As for getpid, note that you can do (since Java 9):<br>
<br>
ProcessHandle.current().pid();<br>
<br>
I believe the impl caches the result, so it shouldn't even
make the native call.<br>
</p>
<p>Maurizio</p>
<p>[1] - <a href="http://cr.openjdk.java.net/~mcimadamore/panama/GetrusageTest.java" target="_blank">http://cr.openjdk.java.net/~mcimadamore/panama/GetrusageTest.java</a><br>
</p>
<div>On 02/07/2022 07:42, Wojciech Kudla wrote:<br>
</div>
<blockquote type="cite">
<div dir="ltr">
<div>
<div>Hi Maurizio,<br>
<br>
</div>
Thanks for staying on this.<br>
<br>
> Could you please provide a rough list of the
native calls you make where you believe critical JNI
is having a real impact in the performance of your
application?<br>
</div>
<div><br>
From the top of my head:<br>
</div>
<div>clock_gettime<br>
</div>
<div>recvmsg<br>
</div>
<div>recvmmsg</div>
<div>sendmsg<br>
</div>
<div>sendmmsg</div>
<div>select<br>
</div>
<div>getpid</div>
<div>getcpu<br>
</div>
<div>getrusage<br>
</div>
<div><br>
</div>
<div>> Also, could you please tell us whether any of
these calls need to interact with Java arrays?<br>
</div>
<div>No arrays or objects of any type involved.
Everything happens by the means of passing raw
pointers as longs and using other primitive types as
function arguments.<br>
</div>
<div><br>
> In other words, do you use critical JNI to remove
the cost associated with thread transitions, or are
you also taking advantage of accessing on-heap memory
_directly_ from native code?<br>
</div>
<div>Criticial JNI natives are used solely to remove the
cost of transitions. We don't get anywhere near java
heap in native code.<br>
<br>
</div>
<div>In general I think it makes a lot of sense for Java
as a language/platform to have some guards around
unsafe code, but on the other hand the popularity of
libraries employing Unsafe and their success in more
performance-oriented corners of software engineering
is a clear indicator there is a need for the JVM to
provide access to more low-level primitives and
mechanisms. <br>
</div>
<div>I think it's entirely fair to tell developers that
all bets are off when they get into some non-idiomatic
scenarios but please don't take away a feature that
greatly contributed to Java's success.<br>
<br>
</div>
<div>Kind regards,<br>
</div>
<div>Wojtek<br>
</div>
</div>
<br>
<div class="gmail_quote">
<div dir="ltr" class="gmail_attr">On Wed, Jun 29, 2022
at 5:20 PM Maurizio Cimadamore <<a href="mailto:maurizio.cimadamore@oracle.com" target="_blank">maurizio.cimadamore@oracle.com</a>>
wrote:<br>
</div>
<blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-style:solid;padding-left:1ex;border-left-color:rgb(204,204,204)">
<div>
<p>Hi Wojciech,<br>
picking up this thread again. After some internal
discussion, we realize that we don't know enough
about your use case. While re-enabling JNI
critical would obviously provide a quick fix,
we're afraid that (a) developers might end up
depending on JNI critical when they don't need to
(perhaps also unaware of the consequences of
depending on it) and (b) that there might actually
be _better_ (as in: much faster) solutions than
using critical native calls to address at least
some of your use cases (that seemed to be the case
with the clock_gettime example you mentioned).
Could you please provide a rough list of the
native calls you make where you believe critical
JNI is having a real impact in the performance of
your application? Also, could you please tell us
whether any of these calls need to interact with
Java arrays? In other words, do you use critical
JNI to remove the cost associated with thread
transitions, or are you also taking advantage of
accessing on-heap memory _directly_ from native
code?</p>
<p>Regards<br>
Maurizio<br>
</p>
<div>On 13/06/2022 21:38, Wojciech Kudla wrote:<br>
</div>
<blockquote type="cite">
<div dir="ltr">
<div>
<div>
<div>
<div>
<div>Hi Mark,<br>
<br>
</div>
Thanks for your input and apologies for
the delayed response.<br>
<br>
> If the platform included, say, an
intrinsified System.nanoRealTime()<br>
method that returned
clock_gettime(CLOCK_REALTIME), how much
would<br>
that help developers in your unnamed
industry?<br>
<br>
</div>
Exposing realtime clock with nanosecond
granularity in the JDK would be a great
step forward. I should have made it clear
that I represent fintech corner
(investment banking to be exact) but the
issues my message touches upon span areas
such as HPC, audio processing, gaming, and
defense industry so it's not like we have
an isolated case.<br>
<br>
> In a similar vein, if people are
finding it necessary to “replace parts<br>
of NIO with hand-crafted native code” then
it would be interesting to<br>
understand what their requirements are<br>
<br>
</div>
As for the other example I provided with
making very short lived syscalls such as
recvmsg/recvmmsg the premise is getting
access to hardware timestamps on the ingress
and egress ends as well as enabling batch
receive with a single syscall and otherwise
exploiting features unavailable from the JDK
(like access to CMSG interface,
scatter/gather, etc).<br>
</div>
<div>There are also other examples of calls
that we'd love to make often and at lowest
possible cost (ie. getrusage) but I'm not
sure if there's a strong case for some of
these ideas, that's why it might be worth
looking into more generic approach for
performance sensitive code.<br>
</div>
<div>Hope this does better job at explaining
where we're coming from than my previous
messages.<br>
</div>
<div><br>
</div>
Thanks,<br>
</div>
W<br>
</div>
<br>
<div class="gmail_quote">
<div dir="ltr" class="gmail_attr">On Tue, Jun 7,
2022 at 6:31 PM <<a href="mailto:mark.reinhold@oracle.com" target="_blank">mark.reinhold@oracle.com</a>>
wrote:<br>
</div>
<blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-style:solid;padding-left:1ex;border-left-color:rgb(204,204,204)">2022/6/6
0:24:17 -0700, <a href="mailto:wkudla.kernel@gmail.com" target="_blank">wkudla.kernel@gmail.com</a>:<br>
>> Yes for System.nanoTime(), but
System.currentTimeMillis() reports<br>
>> CLOCK_REALTIME.<br>
> <br>
> Unfortunately System.currentTimeMillis()
offers only millisecond<br>
> granularity which is the reason why our
industry has to resort to<br>
> clock_gettime.<br>
<br>
If the platform included, say, an intrinsified
System.nanoRealTime()<br>
method that returned
clock_gettime(CLOCK_REALTIME), how much would<br>
that help developers in your unnamed industry?<br>
<br>
In a similar vein, if people are finding it
necessary to “replace parts<br>
of NIO with hand-crafted native code” then it
would be interesting to<br>
understand what their requirements are. Some
simple enhancements to<br>
the NIO API would be much less costly to
design and implement than a<br>
generalized user-level native-call
intrinsification mechanism.<br>
<br>
- Mark<br>
</blockquote>
</div>
</blockquote>
</div>
</blockquote>
</div>
</blockquote>
</div>
</blockquote>
</div>
</blockquote>
</div>
</blockquote></div></div>-- <br><div dir="ltr">Sent from my phone</div>
</blockquote></div>
</blockquote></div></div>-- <br><div dir="ltr" class="gmail_signature" data-smartmail="gmail_signature">Sent from my phone</div>