thanks stephan , thierry and henrik I will use your code. Alexander I would prefer to profile slow code, but sometimes profiling even fast code gives more info about the internals of the system and what is fast and what is slow and why.
On Sun, Nov 22, 2015 at 10:39 PM EuanM <euan...@gmail.com> wrote: > I was profiling yesterday, and discovered these issues: > > Issue: > ===== > Time and DateAndTime both show human-readable time information to > micro-second precision. Time also holds information to nano-second > precision. > > Both objects are only providing information to the nearest millisecond. > > The digits beyond milliseconds are always (on my PC) filled in as '902'. > > In the nanoseconds instvar of the Time objects, the nanoseconds are > always filled in as '000'. > > > Root cause: > ========= > When Times (and DateAndTimes) are created, they use the > Time>>primUTCMicrosecondsClock method to generate the time > > primUTCMicrosecondsClock > "Answer the number of micro-seconds ellapsed since Squeak epoch. > That is since 00:00 on the morning of January 1, 1901 UTC. > At least a 60-bit unsigned integer is used internally which is enough > for dates up to year 38435. > Essential. See Object documentation whatIsAPrimitive. " > > <primitive: 240> > self primitiveFailed > > Potential for solution: > ================ > [The following is taken from Stack Overflow, > > http://stackoverflow.com/questions/2607263/how-precise-is-the-internal-clock-of-a-modern-pc > ] > Modern PCs have a hardware High Precision Event Timer (HPET) which > specifies a clock speed of 10MHz > http://www.intel.com/hardwaredesign/hpetspec_1.pdf > Even newer PC architectures take this and put it on the Northbridge > controller and the HPET can tun at 100MHz or even greater. At 10Mhz we > should be able to get a resolution of 100 nano-seconds and at 100MHZ > we should be able to get 10 nano-second resolution. > > The following operating systems are known not to be able to use HPET: > Windows XP, Windows Server 2003, and earlier Windows versions, older > Linux versions > The following operating systems are known to be able to use HPET: > Windows Vista, Windows 2008, Windows 7, x86 based versions of Mac OS > X, Linux operating systems using the 2.6 kernel and FreeBSD. > > Final Question: > =========== > Is this something we should pass on to the VM teams? > > On 22 November 2015 at 20:23, EuanM <euan...@gmail.com> wrote: > > Because profiling longer loops takes longer. > > > > Why profile a 32768-cycle loop if a 2048-cycle loop would work? > > > > On 22 November 2015 at 20:21, Alexandre Bergel <alexandre.ber...@me.com> > wrote: > >> Hi! > >> > >> Why do you need to profile fast code? You can always put your code in a > loop. > >> > >> Alexandre > >> > >> > >>> On Nov 22, 2015, at 4:22 PM, Dimitris Chloupis <kilon.al...@gmail.com> > wrote: > >>> > >>> yeah probably , your code returns nothing too. So what we suppose to > use to profile fast code ? > >>> > >>> On Sun, Nov 22, 2015 at 9:12 PM Nicolai Hess <nicolaih...@gmail.com> > wrote: > >>> 2015-11-22 20:01 GMT+01:00 Dimitris Chloupis <kilon.al...@gmail.com>: > >>> Why > >>> > >>> [CheckboxMorph new] timeProfile > >>> > >>> returns nothing ? > >>> > >>> too fast? > >>> [ 3+3 ] timeProfile > >>> return nothing too. > >>> > >>> try > >>> [CheckboxMorph new openInWorld] timeProfile. > >>> > >>> > >>> I also tried it in the profile tool without the block and still > displays nothing . Is there any special requirement for profiling code ? > >> > >> -- > >> _,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;: > >> Alexandre Bergel http://www.bergel.eu > >> ^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;. > >> > >> > >> > >> > >
