- Apr 15, 2005
-
-
rtoy authored
o Add function for ppc to compute the number of CPU cycles per timebase ticks. Save this in a global variable. code/time.lisp: o Adjust cycle-count/float for ppc to use the cycles-per-tick to compute the actual number of CPU cycles, so we can display cycles instead of ticks.
-
- Oct 09, 2004
-
-
rtoy authored
that seems to depend on what CPU is being used. Jon Boone says it's 64 on a G5. It's some other weird value for a G4. o Change TIME output to say "time-base cycles" instead of "CPU cycles" for PPC, so the user isn't confused when (time (sleep 1)) returns something unexpected.
-
rtoy authored
incremente once every 16 clock cycles on a 400 MHz iMac G3. We assume that's true for other ppc's.
-
- Jul 26, 2003
-
-
gerd authored
(get-time-consing): Use it. (%time): Don't print if *in-get-time-consing*.
-
- Jul 24, 2003
-
-
gerd authored
calling (TIME NIL). * src/code/time.lisp (*time-consing*, *last-time-consing*): New vars. (get-time-consing): New function. (%time): Call it to get the additional consing overhead of %time. Subtract *time-consing* from the result unless null. Set *last-time-consing*.
-
- Feb 25, 2003
-
-
toy authored
-
emarsden authored
uses the CPUID + RDTSC instructions on Pentium, and reads the %TICK register on UltraSPARC. Accessible via the VM::READ-CYCLE-COUNTER VOP that returns two (unsigned-byte 32) values, that are the lower and upper components of a 64-bit cycle count (actually 63 bits for UltraSPARC). Basic support for counting the number of CPU cycles has been added to the TIME macro.
-
- Jan 06, 2003
-
-
toy authored
have the semi-colon prefix.
-
- Nov 05, 2002
-
-
cracauer authored
Make (time ...) and the profiler do precise measuring of space allocation. It will also not overflow or bomb out when consing amounts cross most-positive fixnum. The new profiler also has an interface to plug in your own print function (also dictates sorting or results). This is written on gencgc/x86 but tests indicated the fallsbacks for other platforms work. The dfixnum package included here is sketchy.
-
- Jun 07, 2000
-
-
dtc authored
multiple of 1/3600 as required by ANSI CL, rather than just multiples of 1/60.
-
- Jan 29, 1998
-
-
dtc authored
use of this to implement get-internal-run-time and get-system-info. This allows the run time to be obtained on Solaris 2.4, and is reportedly faster than getrusage on later versions of Solaris.
-
- Jul 12, 1996
-
-
ram authored
-
- Oct 31, 1994
-
-
ram authored
-
- Feb 11, 1994
-
-
cvs2git authored
-
- Nov 13, 1993
-
-
wlott authored
encode-universal-time to be sane and to correctly deal with timezones. Well, at least they are closer to dealing correctly with timezones.
-
- Aug 25, 1993
-
-
ram authored
-
- Jan 13, 1993
-
-
cvs2git authored
-
- Aug 05, 1992
-
-
ram authored
number-consing and generic arithmetic. Also, rearranged the computation so that the time is correctly computed for up to 457 days, instead of only 71 minutes.
-
- Mar 14, 1992
-
-
ram authored
is displayed.
-
- Feb 20, 1992
-
-
ram authored
-
- Feb 19, 1992
-
-
wlott authored
-
- Feb 14, 1992
-
-
wlott authored
-
- Aug 30, 1991
-
-
ram authored
syscalls. Use SYSTEM:GET-SYSTEM-INFO for os-independent operation.
-
- Aug 23, 1991
-
-
ram authored
when it isn't.
-
- Feb 08, 1991
-
-
ram authored
-
- Feb 04, 1991
-
-
ram authored
call to minimize the probability of bignum results. Changed both GET-INTERNAL-REAL-TIME and GET-INTERNAL-RUN-TIME to assume that the syscall does ERRNO checking and to have a locally unsafe policy.
-
- Aug 24, 1990
-
-
wlott authored
-
- May 30, 1990
-
-
cvs2git authored
-
- Mar 05, 1990
-
-
cvs2git authored
-
- Feb 22, 1990
-
-
ram authored
Made a simple tweak to DECODE-UNIVERSAL-TIME that, in compliance with the new standard, ignores daylight-savings-time when the user supplies the :time-zone argument.
-
- Feb 07, 1990
-
-
ram authored
%TIME with the form wrapped into a lambda. This helps interpreted uses of TIME, mainly by reducing the spurious elapsed time for preprocessing the huge form, but should also produce somewhat more accurate results, since the code in %TIME always runs compiled. Theoretically this could have a slight negative effect on the accuracy of compiled uses by introducing a call, but this should be well down in the noise, since call overhead should be well less than the clock resolution.
-
- Feb 06, 1990
-
-
ram authored
-