I currently do this to successfully get the current epoch time in nanos:
Instant inst = Instant.now();
long time = inst.getEpochSecond();
time *= 1000000000l;
time += inst.getNano();
However, it's a bit too slow for my use case, taking around 1us each call after the JVM has warmed up.
Is there a faster way to do it?
I'm happy with a solution that gives me the microseconds since epoch, as long as it's faster than the above.
What may work is to run:
long n1 = System.nanoTime();
long m = System.currentTimeMillis();
long n2 = System.nanoTime();
a number of times until the difference between n1 and n2 is less than the resolution you want (it's about 400 ns on my PC after a couple of iterations).
You can then use the difference between n1 (or n2 or an average of the 2...) and m * 1e6 as an offset that you need to add to System.nanoTime() to get the current epoch nanos.
Disclaimer:
System.nanoTime doc explicitly states that the resolution is at least that of System.currentTimeMillis(), which may be > 1 ms. So no guarantee that you will get microsecond resolution.
Corollary: this probably doesn't work in all environments (you may never get n2-n1 small enough - or it may be 0 just because the resolution of your system is too low).
System.nanoTime() may be out of sync over long periods - so this is a trade off between precision and performance.
You also need to account for possible arithmetic overflow.
See also: Current time in microseconds in java
Related
I am asked to store the time right before my algorithm start, and time when it ends, and also need to provide the difference between them (end time - start time).
But the System.currentTimeMillis() function generates values that are too long:
start=1497574732045
end=1497574732168
Is there a way to make this value just 3 digits like "123" but also be as precise as using the System.currentTimeMillis() function?
as the currentTimeMillis() description says:-
Returns the current time in milliseconds. Note that while the unit of time of the return value is a millisecond, the granularity of the value depends on the underlying operating system and may be larger. For example, many operating systems measure time in units of tens of milliseconds.
Returns:
the difference, measured in milliseconds, between the current time and midnight, January 1, 1970 UTC.
in your case use this simple trick and you will get the desired result.
Long startTime= Long.parseLong("1497674732168");
Long endTime= Long.parseLong("1497574732168");
System.out.println("start time is"+new Date(startTime)+"end time is"+new Date(endTime));
If you need to store the start and end times separately, there are only two ways (I can think of) to make the values smaller.
Firstly, System.currentTimeMillis() counts from January 1, 1970 UTC. But if your clock is never going to run previous to "now", you can subtract a fixed amount of time. I chose 1497580000000 as it's definitely in the past at the time I wrote this and its a nice even number.
Second, divide the value by any amount of precision you are willing to lose. In your case you might not want to even do that, but here I chose 100.
The numbers returned look small now, but they will continue to get bigger as the difference between the current time and 1497580000000 become more pronounced.
The preferred solution is to not do any of this at all, but just store the long value if you can.
You'll never magic a large precise number into only 3 decimal digits. Not without quantum mechanics.
{
long start = 1497584001010L;
long end = 1497584008000L;
System.out.println("Diff: " + (end - start));
int compactStart = compact(start);
int compactEnd = compact(end);
System.out.println("Compact Start: " + compactStart);
System.out.println("Compact End: " + compactEnd);
System.out.println("Diff: " + (expand(compactEnd) - expand(compactStart)));
}
private int compact(long millis) {
return (int)((millis - 1497580000000L)/100);
}
private long expand(int millis) {
return (millis + 1497584000000L)*100;
}
Result...
Diff: 6990
Compact Start: 40010
Compact End: 40080
Diff: 7000
Note 7000 doesn't equal 6990 because of the intentional precision loss.
According to java.lang.System API
currentTimeMillis() Returns the current time in milliseconds
nanoTime() Returns the current value of the running Java Virtual
Machine's high resolution time source, in nanoseconds.
Strictly speaking a nanosecond is 1e-9 and millisecond is 1e-3. Therefore, a duration in nanosecs must be a multiple of 1e6 of the same duration in millisecs. This is not the case in practice, what is the reason?
scala> System.nanoTime / System.currentTimeMillis
res0: Long = 107
System.nanoTime() has an arbitrary start point; it's not unix epoch. From the Javadoc:
The value returned represents nanoseconds since some fixed but arbitrary origin time
So what you're actually calculating there is:
(unknownOffset + offsetFromEpochInNanos) / offsetFromEpochInMillis
which will almost certainly not be 1e6, unless unknownOffset happens to be arbitrarily zero.
If you can remove the effect of the unknown offset by subtracting the two times, you can see that the ratio is around 1e6:
long nanoStart = System.nanoTime();
long milliStart = System.currentTimeMillis();
Thread.sleep(2000);
long nanoEnd = System.nanoTime();
long milliEnd = System.currentTimeMillis();;
long nanoDelta = nanoEnd - nanoStart;
long milliDelta = milliEnd - milliStart;
System.out.println((double) nanoDelta / milliDelta);
Output (running 5 times):
1000058.3725
1000045.4705
999549.1579210395
1000046.101
1000038.1045
Ideone demo
So, pretty close to 1e6.
Note that it might not be this, because System.currentTimeMillis() doesn't progress smoothly, owing to corrections for clock skew. However, these should be infrequent, so most of the time when you run this code, you'll see roughly 1e6.
I want to return microseconds from linux as java only has wall clock times to millisecond accuracy on systems with a monotonic clock.
My exposure to jni is limited so apologies if it's a silly question.
I believe I can either make a call in the c layer to gettimeofday and return the value as jlong:
private native long getMicros();
Or perhaps alternatively take a pointer to an address and then write the value to this address:
private native void getMicros(Long ptr);
The latter throws up lots of questions in my mind like "how does c know what the binary format of jlong is" and "how would I even do this!".
I just wondered if the latter might be faster than returning a value back across the jni layer.
Any thoughts most welcome.
http://docs.oracle.com/javase/6/docs/api/java/lang/System.html#nanoTime()
"Returns the current value of the most precise available system timer, in nanoseconds.
This method can only be used to measure elapsed time and is not related to any other notion of system or wall-clock time. The value returned represents nanoseconds since some fixed but arbitrary time (perhaps in the future, so values may be negative). This method provides nanosecond precision, but not necessarily nanosecond accuracy. No guarantees are made about how frequently values change. Differences in successive calls that span greater than approximately 292 years (263 nanoseconds) will not accurately compute elapsed time due to numerical overflow.
For example, to measure how long some code takes to execute:"
long startTime = System.nanoTime();
// ... the code being measured ...
long estimatedTime = System.nanoTime() - startTime;
java.sql.Timestamp constructor go like this:
public Timestamp(long time) {
super((time/1000)*1000);
nanos = (int)((time%1000) * 1000000);
if (nanos < 0) {
nanos = 1000000000 + nanos;
super.setTime(((time/1000)-1)*1000);
}
}
It basically accepts time in millisecond and then extracts the last 3 digits and makes it nanos. So for a millisecond value of 1304135631 421, I'm getting Timestamp.getnanos() as
421000000. This is plain calculation (adding 6 zeroes at the end)... does not seems to be optimum.
A better way could have been Timestamp constructor that accepts time in nanoseconds and then calculates the nanosecond value out of that.
If you run the below program, you'll see the difference between actual nanoseconds and the one returned by Timestamp way of calculating nanosecods.
long a = System.currentTimeMillis();
for(;;){
long b = System.currentTimeMillis();
Timestamp tm = new Timestamp(System.currentTimeMillis());
System.out.println(tm.getTime());
System.out.println(tm.getNanos());
System.out.println("This is actual nanos" + System.nanoTime()%1000000000);
System.out.println("--------------------------");
if(b-a >= 1)
break;
}
So all the discussion about Timestamp that says it stores time up to nanoseconds , does not seems to be so correct.. Isn't?
The time in millis does not represent the time in nanos. More precise it simply can't be. You're supposed to use Timestamp#setNanos() to set the real nanos.
long timeInMillis = System.currentTimeMillis();
long timeInNanos = System.nanoTime();
Timestamp timestamp = new Timestamp(timeInMillis);
timestamp.setNanos((int) (timeInNanos % 1000000000));
// ...
Since the introduction of java.time.*, there is a new factory method in java.sql.Timestamp: Timestamp.from(Instant.now()) will do the job (with nanoseconds precision). There is also Timestamp.toInstant() to convert it the other way around.
Although it's an old post, I would like to add that the docs of Timestamp does state that it "holds fractional seconds by allowing the specification of
fractional seconds to a precision of nanaoseconds". The confusing part is "hold". This seems confusing at first but if understood correctly, it actually does not state that it holds nanaoseconds
value.It says it "holds" fractional value and allows it to be a "precision" of nanoseconds. Precision should be understood in terms of representation
of total number of digits. So it essentially means that the part is actually fractional (still milliseconds) but is multiplied by 1000000 to represent it as nanoseconds.
The accepted answer (by ever helpful BaluC) sums it up nicely.
I like OpenJPA's implementation of TimestampHelper. It use static initializers to keep track of elapsed nanoseconds between calls to make a timestamp.
I'm in and android widget and checking elapsed time between two calls of System.nanoTime() and the number is huge. How do you measure elapsed time with this? it should be a fraaction of a second and instead its much more. Thanks
The System.nanoTime() returns a time value whose granularity is a nanosecond; i.e. 10-9 seconds, as described in the javadoc. The difference between two calls to System.nanoTime() that are a substantial fraction of a second apart is bound to be a large number.
If you want a time measure with a larger granularity, consider System.currentTimeMillis() ... or just divide the nanosecond values by an appropriate power of 10 to suit your application.
Note that on the Android platform there are 3 distinct system clocks that support different "measures" of time; see SystemClock. If you are programming explicitly for the Android platform, you should read the javadoc and decide which measure is most appropriate to what you are doing.
For your information, "nano-" is one of the standard prefixes defines by the International System of Units (SI) - see http://physics.nist.gov/cuu/Units/prefixes.html.
If you really think that "they" got it wrong and that "nano-" is too small, you could always write a letter to the NIST. I'm sure someone would appreciate it ... :-)
One seconds contains 1,000,000,000 nanoseconds, so as long as your number is in that range, it's reasonable.
If you want it in fractional form, just take your value / 10^9 where value is your difference in nanoTime()s.
long nanoSeconds = 500000000;
float seconds = nanoSeconds / 1000000000;
Log.i("NanoTime", nanoSeconds + " ns is the same as " + seconds + " seconds");
Your output would be:
07-27 11:35:47.196: INFO/NanoTime(14237): 500000000 ns is the same as 0.5 seconds