Why is 64 bit faster. RE: [Beowulf] Win64 Clusters!!!!!!!!!!!!
Robert G. Brown
rgb at phy.duke.edu
Thu Apr 12 10:01:02 EDT 2007
On Thu, 12 Apr 2007, Mark Hahn wrote:
>> 2.) If you install the 32 bit version of an os, say linux, instead of
>> the 64 bit version and then run 32 bit apps, do you get the speedup?
> no. in any case, I doubt there's any reason to install 32b linux,
> since a 64b kernel _can_ support 32b processes (which actually see a somewhat
> bigger practical address space...)
I'd modify this in two ways. In most cases these days, an automated 64b
install will install both versions of libraries anyway to enable 32b
binaries to run, so a 64b kernel and distro can still run 32b at least
some of the time without a recompile (with the usual YMMV warnings).
Second, when the opteron was first released, it was touted as the
world's fastest 32 bit CPU, because it was. One can argue about whether
or not "any" new generation CPU wouldn't have run faster (32 bit or 64
bit, where for obvious reasons they chose to engineer it as 64 bit) but
either way "empirically" a 32b operating system on a 64b CPU can run
faster at equivalent clock than the same operating system on a 32b CPU.
That doesn't make Mark's reply incorrect, only gives it context (and
shows a bit of the complexity associated with the problem. The
discussion continues to focus on "local" parts of a "global" engineering
problem. One can always second guess whether any particular part of a
CPUs design is optimal for all purposes, but the fact is that that part
is INTEGRATED with the rest of the design with many trade-offs in a
global optimization problem. Unless you are a hardware engineer working
for Intel, AMD, Motorola, IBM it is unlikely that you can know or
understand >>all<< of the reasons that went into why and how a design
ended up the way that it did, although we all know some of them and
although it is great fun to discuss them in or out of context.
Robert G. Brown http://www.phy.duke.edu/~rgb/
Duke University Dept. of Physics, Box 90305
Durham, N.C. 27708-0305
Phone: 1-919-660-2567 Fax: 919-660-2525 email:rgb at phy.duke.edu
Beowulf mailing list, Beowulf at beowulf.org
To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf
More information about the Beowulf