On TechRepublic: Five super-secret features in Windows 7
BNET Business Network:
BNET
TechRepublic
ZDNet
TalkBack 32 of 39:
Next »
« Previous
Ya ya ya: something you overlook
MP operating systems use something called "affinity" to
guarantee that a task does not execute on a different CPU from
its parent (can be altered). In fact, no matter how many threads
are created by a given task, they must all operate in the local
machine space owned by their parent task. I believe that in the
absence of any guidance from a progammer, the OS linker is
strongly dedicated to the idea that tasks not be separated from
the parent that called them. I did not write any part of *Office*
and so I would not dare to comment, but I would be that it is
designed to run in a parent environment (if for no other reason
than to take advantage of the shareable objects).

That said, the original poster might be right. However, the
original poster asserts that the multicore CPU is actually slower
because each single core is clocking slower. That could not be
further from the truth. The rate of a given clock is definite. If a
utility says 3201MHz is your clock speed, that is the number of
cycles generated by the clock. No changing that to 1/2 speed
on two cores.

This is because a multicore CPU cannot operate if it has more
than one time signal, or has more than one arbitrator of a
common signal. In either of those two cases, it would be
possible for each core to have a significantly different idea of the
bus time. If the cores did not agree on the bus time, they could
easily overwrite i/o or data, and the OS would crash.

Not that I am an expert.
Posted by: gjsherr   Posted on: 05/17/06 You are currently: a Guest | Members login | Terms of Use
Reply to Story Reply to Message

Alert moderator to an offensive message

Subscribe to this discussion via Email or RSS

Script  Yagotta B. Kidding | 05/16/06
Intel and AMD both have blinders on.  No_Ax_to_Grind | 05/16/06
unless they can execute x86 instructions faster then Intel/AMD can natively  mattdevo | 05/16/06
Your right, and that is why they are fools.  No_Ax_to_Grind | 05/16/06
Good point  John Zern | 05/17/06
but that's just it...  st!lborn | 05/16/06
So you're telling us  Yagotta B. Kidding | 05/16/06
We sure are...  andrnils | 05/16/06
Let me lend you a clue.  No_Ax_to_Grind | 05/16/06
The price of oranges in Seville  Yagotta B. Kidding | 05/16/06
And I take it you have never looked at an Xbox 360.  No_Ax_to_Grind | 05/16/06
Not to mention sangria  Yagotta B. Kidding | 05/16/06
XBOX 360 does not use the cell  eb276 | 05/16/06
Let me give you a clue  eb276 | 05/16/06
congrats  nECrO_z | 05/16/06
Let me lend YOU a clue as well  nECrO_z | 05/16/06
I agree No_Ax  crocd | 05/16/06
Already out...  No_Ax_to_Grind | 05/16/06
LOL. Cell chip? Cell is this years version of the Emotion Engine  Prognosticator | 05/16/06
Yeah, the x86 instruction set is perfect  No_Ax_to_Grind | 05/16/06
The triumph of thrust over aerodynamics  Yagotta B. Kidding | 05/16/06
You seem to be confused.  No_Ax_to_Grind | 05/16/06
No takers  Yagotta B. Kidding | 05/16/06
Bravo  nECrO_z | 05/16/06
It's a pity  Yagotta B. Kidding | 05/17/06
While you're playing your xbox and ps3 I'll be working  rschror | 05/16/06
AMD is missing Virtualization  mighetto | 05/16/06
No they are not  toadlife | 05/16/06
Thats true.  AssaultedNuts | 05/16/06
Agreed AMD virtualization is already in box  rschror | 05/16/06
Not quite right  balsover | 05/16/06
Ya ya ya: something you overlook  gjsherr | 05/17/06
Intel's claim of performance was proven fraudulent  sharikou | 05/16/06
Where you been hiding?  Prognosticator | 05/16/06
With a name like rhetoricpolice are you much better?  rschror | 05/16/06
Sharikou was proven fraudulent as well  doh123 | 05/17/06
Competition is Great!  jpr75_z | 05/16/06
Will this use the AM2 socket?  compunetinc | 05/16/06
This is great progress ofcourse. I wish that it wouldn't be for a x86  michael_t | 05/16/06

What do you think?

SponsoredWhite Papers, Webcasts, and Downloads

advertisement
advertisement