Re: mutex overkill ?

Tech-Archive recommends: Speed Up your PC by fixing your registry

Only spinning - SMP seems to be just irrelevant here.

That "seems" incorrectly. As always.

Don't forget that critical section uses *LOCKED* increment and test, so
that the idea is similar to that of a spinlock. Therefore, the chance of
successfull acquisition without waiting on event does not seem to
increase, no matter how many CPUs are in the system

Profound ignorance. As always.

It looks like our self-appointed "Windows expert" (AKA Slava M.Usov)
decided to honour this particular thread with his participation.
Certainly, as usually, he speaks about "profound ignorance", which is
not surprizing at all. It would be great if, in addition to speaking
about "ignorance" he could say something on the actual topic of a
discussion, and explain what exactly, in his opinion, is wrong - there
are at least 3 different things that Slava M.Usov may theoretically
argue about, and, taking into consideration Slava M.Usov's propensity
to argue even about obvious things (for example, about the fact that
i8042prt is a port driver), it is hard to guess what exactly he
disagrees with


Anton Bassov

Slava M. Usov wrote:
"anton bassov" <soviet_bloke@xxxxxxxxxxx> wrote in message
news:1158635821.595214.141530@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

You are neglecting SMP and spinning in that analysis

Only spinning - SMP seems to be just irrelevant here.

That "seems" incorrectly. As always.

Don't forget that critical section uses *LOCKED* increment and test, so
that the idea is similar to that of a spinlock. Therefore, the chance of
successfull acquisition without waiting on event does not seem to
increase, no matter how many CPUs are in the system

Profound ignorance. As always.

The only reason why SMP *THEORETICALLY* may have some impact is because
the owner of the section has more CPU time available, so that it is able
to do the job and release the section faster than it would on the UP
machine. However, don't forget that each "constestant" has more CPU time
as well, so that its chance of reaching its maximal spin count and
entering wait on event is higher, compared to that on UP machine.

Faulty logic based on incorrect premises, predictably nonsensical
conclusion. As always.

S

.

Relevant Pages

  • Re: XP Prof 32 crashed; reinstalled XP, lost all files; need to restor
    ... You were incorrect in that you had no option and you are likely (given what ... Much in the same way a 'wipe' works by overwriting the area of the disk once ... Settings" directory - then you have a very good chance of recovery. ...
    (microsoft.public.windowsxp.general)
  • Re: Mail merge randomly shifting pages
    ... then we have no chance! ... But...are you saying that the text splits differently even when the text in ... appropriate to experiment with the various line and page break options. ... then maybe 1 incorrectly, a couple more correct then some incorrect. ...
    (microsoft.public.word.mailmerge.fields)
  • Re: mutex overkill ?
    ... that the idea is similar to that of a spinlock. ... Therefore, the chance of ... don't forget that each "constestant" has more CPU time ... Faulty logic based on incorrect premises, ...
    (microsoft.public.win32.programmer.kernel)
  • Re: My my, how time flies ...... its been about "1 hour" -- anyone cracked CryptoSMS yet?
    ... judging from the things he's said in ... there's a good chance that the ciphers are implemented ... are incorrect? ...
    (sci.crypt)