Hi Folks, Sorry for any cross-posting which may affect any of you!
I've been asked the following, but don't know the technicalities to provide an answer myself.
Given that a certain simulation job (10,000 rounds) takes 6-8 hours on an Intel Pentium 4 CPU 2.59GHz machine, what might it take (i.e. what speed factor) on a machine with
a) IntelR CoreT2 Duo E6400
b) IntelR CoreT2 Duo E6600
c) IntelR CoreT2 Duo E6700
(Ball-park will do). I haven't been given the GHz speeds for these other processors, but a quick Google suggests they're all around 2.6GHz, give or take a bit.
The computations for the simulation are numerically intensive; also much reading to & writing from RAM.
With thanks, Ted.
-------------------------------------------------------------------- E-Mail: (Ted Harding) Ted.Harding@manchester.ac.uk Fax-to-email: +44 (0)870 094 0861 Date: 09-Jul-08 Time: 21:43:45 ------------------------------ XFMail ------------------------------
On Wed, 2008-07-09 at 21:43 +0100, Ted.Harding@manchester.ac.uk wrote:
Given that a certain simulation job (10,000 rounds) takes 6-8 hours on an Intel Pentium 4 CPU 2.59GHz machine, what might it take (i.e. what speed factor) on a machine with
a) IntelR CoreT2 Duo E6400 speed factor of 1.265
b) IntelR CoreT2 Duo E6600 speed factor of 1.387
c) IntelR CoreT2 Duo E6700 speed factor of 1.441
:)
Seriously though..there is no way to answer this question without knowing a great deal about the simulation.
In theory clock for clock a single core of a Core2 due may in some circumstances be faster than a P4. But does your simulation thread across SMP machines so can you make full use of the 2nd core ?..or does the machine do other things at the same time or run multiple concurrent instances of the simulation and therefore will benefit anyway ?
Can you recompile the simulation for 64bit arch ? Is it the sort of thing that would benefit from this (or even just benefit from being able to address more ram than the 32bit arch is capable of)
You might see an improvement just because you have more memory bandwidth..DDR2 is up to 1066 Mhz now and dual channel mainboards are the norm. What memory is in your current P4 ?
Alternatively is the simulation portable enough to easily test on someone else's machine, plenty of people (me included) have dual or quad core versions of the processor you are looking at who could benchmark it for you.
On 09-Jul-08 21:39:21, Wayne Stallwood wrote:
On Wed, 2008-07-09 at 21:43 +0100, Ted.Harding@manchester.ac.uk wrote:
Given that a certain simulation job (10,000 rounds) takes 6-8 hours on an Intel Pentium 4 CPU 2.59GHz machine, what might it take (i.e. what speed factor) on a machine with
a) IntelR CoreT2 Duo E6400 speed factor of 1.265
b) IntelR CoreT2 Duo E6600 speed factor of 1.387
c) IntelR CoreT2 Duo E6700 speed factor of 1.441
:)
Thanks for that info -- it gives me an idea. I can't respond to your other points, since I don;t know the details, and all the machines in question are in New Zealand! (I was just sent the query in the form I Posted it)! Ted.
Seriously though..there is no way to answer this question without knowing a great deal about the simulation.
In theory clock for clock a single core of a Core2 due may in some circumstances be faster than a P4. But does your simulation thread across SMP machines so can you make full use of the 2nd core ?..or does the machine do other things at the same time or run multiple concurrent instances of the simulation and therefore will benefit anyway ?
Can you recompile the simulation for 64bit arch ? Is it the sort of thing that would benefit from this (or even just benefit from being able to address more ram than the 32bit arch is capable of)
You might see an improvement just because you have more memory bandwidth..DDR2 is up to 1066 Mhz now and dual channel mainboards are the norm. What memory is in your current P4 ?
Alternatively is the simulation portable enough to easily test on someone else's machine, plenty of people (me included) have dual or quad core versions of the processor you are looking at who could benchmark it for you.
main@lists.alug.org.uk http://www.alug.org.uk/ http://lists.alug.org.uk/mailman/listinfo/main Unsubscribe? See message headers or the web site above!
-------------------------------------------------------------------- E-Mail: (Ted Harding) Ted.Harding@manchester.ac.uk Fax-to-email: +44 (0)870 094 0861 Date: 09-Jul-08 Time: 23:55:04 ------------------------------ XFMail ------------------------------
On Wed, 2008-07-09 at 21:43 +0100, Ted.Harding@manchester.ac.uk wrote:
The computations for the simulation are numerically intensive; also much reading to & writing from RAM.
Frontside Bus speed and on chip cache will be the key factors I would guess then
-
Ted.Harding＠manchester.ac.uk -
Tim Fletcher -
Wayne Stallwood