James Harris wrote:
> Oddly, no replies as yet. Will see if others are interested. The test
> below was on an Intel chip. Testing on an AMD CPU shows similar (but
> not as marked) effect.
:) I also waited for some answers ...
The align will affect the timing of all instructions after the
first RDTSC. My time measurement work an a dedicated test area
filled with nops, aligned to cache-bounds and includes debugger
overhead, so I can step/trace my code under test and see the
re****ted TSC-difference.
But this is just useful for comparing algos or find obvious
dependencies early, it wont tell anything about code duration when
this test code become part of a greater scenario somewhere else ...
The different figures you get without align may come from cache-
bound-crossing of your test code and if I look in more detail:
+71,+91,+112,+126 could this be burst-read penalties caused by
the 2nd serialising ?
btw: I don't have a CPUID before the second RDTSC
so I get an almost contant 14 cycles count for an empty test.
__
wolfgang
> On 22 Apr, 02:30, James Harris <james.harri...@[EMAIL PROTECTED]
> wrote:
> > If I run the code below (which is based on an Intel performance
> > measuring note) the times for each of the three inter-rdtsc sequences
> > is almost always 122 cycles. However, if the align directive is
> > removed, sometimes the cycle counts vary markedly.
> >
> > Anyone else seen this effect?
> >
> > --
> > (Reposting here to avoid the moderator delay on another group.)
> > James
> >
> > An example of running the code without the align directive:
> >
> > Calibration times 122, 122, 122
> > Calibration times 122, 213, 122
> > Calibration times 122, 193, 122
> > Calibration times 122, 213, 122
> > Calibration times 122, 234, 122
> > Calibration times 122, 193, 122
> > Calibration times 122, 193, 122
> > Calibration times 122, 248, 122
> > Calibration times 122, 193, 122
> > Calibration times 122, 213, 122
> >
> > With the align directive:
> >
> > Calibration times 122, 122, 122
> > Calibration times 122, 122, 122
> > Calibration times 122, 122, 122
> > Calibration times 122, 122, 122
> > Calibration times 122, 122, 122
> > Calibration times 122, 122, 122
> > Calibration times 122, 122, 122
> > Calibration times 122, 122, 122
> > Calibration times 122, 122, 122
> > Calibration times 122, 122, 122
> >
> > ;;;;;;;;
> > ;Calibrate the time stamp counter measurement
> >
> > align 32
> > tsc_calibrate:
> > mov eax, [subtime_1] ;Preload cache
> > xor eax, eax
> > cpuid
> > rdtsc
> > mov [subtime_1], eax
> > xor eax, eax
> > cpuid
> > rdtsc
> > sub eax, [subtime_1]
> > mov [subtime_1], eax
> >
> > mov eax, [subtime_2] ;Preload cache
> > xor eax, eax
> > cpuid
> > rdtsc
> > mov [subtime_2], eax
> > xor eax, eax
> > cpuid
> > rdtsc
> > sub eax, [subtime_2]
> > mov [subtime_2], eax
> >
> > mov eax, [subtime_3] ;Preload cache
> > xor eax, eax
> > cpuid
> > rdtsc
> > mov [subtime_3], eax
> > xor eax, eax
> > cpuid
> > rdtsc
> > sub eax, [subtime_3]
> > mov [subtime_3], eax
> >
> > ret
>
|
