Re: hyper threading.

From: Anthony Atkielski (atkielski.anthony_at_wanadoo.fr)
Date: 03/27/05

  • Next message: Anthony Atkielski: "Re: hyper threading." 
    Date: Sun, 27 Mar 2005 20:01:57 +0200
To: freebsd-questions@freebsd.org

    em1897@aol.com writes:

    > You know, you spout all of this wonderful theory without considering
    > the quality of the implementation.

    Somethings can be derived directly from theory. If you know the design
    of the hardware, you can predict that two processors will provide x%
    increment of throughput over a single processor, even if you don't
    actually measure them.

    In my case, I cite both theory and my own experience in measuring actual
    systems. The general principles of behavior of multiprocessor systems
    are well understood, although specific implementations vary. It is
    clear, based even on design data alone, that hyperthreading will
    generally improve throughput and should never diminish it (disregarding
    OS overhead). It is equally clear that the gain won't be as great as
    having physically independent processors, but the idea of putting more
    of the idle processor logic to work is a good one.

    > And a key point that you consistently overlook is that FreeBSD 5.x is
    > a particularly poor implementation of SMP. Linux and Dragonfly get 80%
    > improvement in performance with a 2nd processor, and FreeBSD doesn't.

    I'd need to see measurements to substantiate this.

    In general, when it comes to optimization, it's best not to fret too
    much over how many percentage points of processor power or throughput
    you gain or lose with specific configuration or implementation choices.
    If your system is running so close to the wire that five percent makes
    the difference between 100% busy and less than 100% busy, you need more
    hardware in any case.

    > The concept that the kernel is poorly implemented by userland is well
    > done is just not an assumption that you can make.

    Actually, it's not something that I spend a lot of time thinking about.
    Right now, my production system is never more than 0.4% busy. And if it
    were 99% busy, I'd be looking at faster hardware, no matter what OS or
    HT/MP options I might have implemented. 

    -- 
Anthony
_______________________________________________
freebsd-questions@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to "freebsd-questions-unsubscribe@freebsd.org"
  • Next message: Anthony Atkielski: "Re: hyper threading."

    Relevant Pages

    • Re: Security and EOL issues
      ... OS software resources are designed that reserved ram and disk space among other resources, to reflect what current hardware size is available. ... (There was a security patch a few years ago that could not be applied to NT4 as it required more resources then NT4 could provide. ... Installing air bags requires that the automobile manufacturer design, test, ... Computer Emergency Response Teams, and Digital Investigations. ...
      (Security-Basics)
    • Staff HW Engineer ~ Lead Us to ATCA & Beyond in Your End-to-End Board-Level HW Desig
      ... The senior level hardware engineer looking for the product realization ... and true ownership that comes with full end-to-end board-level hardware ... help us retain dominance in the design of high performance switching ...
      (comp.arch.embedded)
    • Re: 10khz DBSK decoder
      ... In an AVR, you may want to come closer to the 'hardwareish' thing: run the whole thing as a Costas loop or as a signal-square-and-PLL, and do integrate-and-dump. ... In retrospect, the ISR should have just taken ADC samples and shoved them into a queue, then set a flag. ... But I had never seen that design pattern, so it just ran with that big bloated ISR... ... You'll get more consistent timing if you can trigger your ADC from hardware and interrupt on the end of conversion pulse. ...
      (comp.dsp)
    • Re: For loop synthesis and repeated structures, any examples?
      ... I meant that the for loop IS an innefficient constuct ... in the sense that it would paralellise the hardware... ... has almost nothing to do with that design compromise. ... the multiplier, and capturing its output into the right ...
      (comp.lang.verilog)
    • Re: FPGA with 5V and PLCC package
      ... Their job will be to design software and not hardware systems. ... _graphical representation_ is the only appropriate way to do low level ... examples as gates. ...
      (comp.arch.fpga)