Re: Lahey Fortran and Core 2 duo
Harvey Gratt wrote:
Interesting issue. I have access to two Dell Precision M4300 workstation
laptops. On both machines I installed my copy of Lahey Fortran Pro 5.5
(LF9555). I also have an old Dell Dimension D8300 desktop with the
compiler on it.
Even though the laptops are suppose to much, much quicker from all
aspects (cpu, HDD, bus, memory, etc), a 6 dof simulation runs over 3
times slower on the "faster" laptops.
Are there any known issues with Lahey Fortran and Core 2 Duo (Santa
Rosa) machines?
You don't give much specific to go on, so I resort to web searches for data.
I wouldn't call the typical 1.8Ghz CPU of the M4300 faster than the
typical 3.2Ghz CPU of D8300, even with the typical 30% higher
instruction per clock rate of SSE code, which you may not have selected
(is your compiler too old for SSE?). As you didn't say anything about
threading, or OpenMP, I'll guess you're leaving 1 core idle. It's a
known issue that a single thread uses only 1 core.
Even if the laptop uses DDR2-667 and the desktop DDR2-533, it doesn't
necessarily make any difference with only 1 thread running. There
probably aren't any commercial Fortran compilers yet which optimize for
the Core 2 rather than the older CPUs, not that we know enough about
your code to say it would make a difference.
I don't have any data about the disks to go on, except that I've never
seen a laptop disk compete on speed.
.
Relevant Pages
- Re: Lahey Fortran and Core 2 duo
... Even though the laptops are suppose to much, much quicker from all aspects, a 6 dof simulation runs over 3 times slower on the "faster" laptops. ... Are there any known issues with Lahey Fortran and Core 2 Duo machines? ... I wouldn't call the typical 1.8Ghz CPU of the M4300 faster than the typical 3.2Ghz CPU of D8300, even with the typical 30% higher instruction per clock rate of SSE code, which you may not have selected. ... As you didn't say anything about threading, or OpenMP, I'll guess you're leaving 1 core idle. ... (comp.lang.fortran) - Re: Lahey Fortran and Core 2 duo
... Even though the laptops are suppose to much, much quicker from all aspects, a 6 dof simulation runs over 3 times slower on the "faster" laptops. ... Are there any known issues with Lahey Fortran and Core 2 Duo machines? ... I wouldn't call the typical 1.8Ghz CPU of the M4300 faster than the typical 3.2Ghz CPU of D8300, even with the typical 30% higher instruction per clock rate of SSE code, which you may not have selected. ... (comp.lang.fortran) - Re: Lahey Fortran and Core 2 duo
... Even though the laptops are suppose to much, much quicker from all aspects, a 6 dof simulation runs over 3 times slower on the "faster" laptops. ... Are there any known issues with Lahey Fortran and Core 2 Duo machines? ... I wouldn't call the typical 1.8Ghz CPU of the M4300 faster than the typical 3.2Ghz CPU of D8300, even with the typical 30% higher instruction per clock rate of SSE code, which you may not have selected. ... As you didn't say anything about threading, or OpenMP, I'll guess you're leaving 1 core idle. ... (comp.lang.fortran) - Re: Lahey Fortran and Core 2 duo
... Even though the laptops are suppose to much, much quicker from all aspects, a 6 dof simulation runs over 3 times slower on the "faster" laptops. ... Are there any known issues with Lahey Fortran and Core 2 Duo machines? ... I wouldn't call the typical 1.8Ghz CPU of the M4300 faster than the typical 3.2Ghz CPU of D8300, even with the typical 30% higher instruction per clock rate of SSE code, which you may not have selected. ... (comp.lang.fortran) - Re: Lahey Fortran and Core 2 duo
... Even though the laptops are suppose to much, much quicker from all aspects, a 6 dof simulation runs over 3 times slower on the "faster" laptops. ... Are there any known issues with Lahey Fortran and Core 2 Duo machines? ... I wouldn't call the typical 1.8Ghz CPU of the M4300 faster than the typical 3.2Ghz CPU of D8300, even with the typical 30% higher instruction per clock rate of SSE code, which you may not have selected. ... (comp.lang.fortran) |
|