On 3/11/06, Prof Brian Ripley <[EMAIL PROTECTED]> wrote:
> Here is a summary of some results on a dual Opteron 252 running FC3
>
> 64-bit gcc 3.4.5
> R's blas34.83 3.45 38.56
> ATLAS 36.70 3.28 40.14
> ATLAS multithread 76.85 5.39 82.29
> Goto 1 thread
Here is a summary of some results on a dual Opteron 252 running FC3
64-bit gcc 3.4.5
R's blas34.83 3.45 38.56
ATLAS 36.70 3.28 40.14
ATLAS multithread 76.85 5.39 82.29
Goto 1 thread 36.17 3.44 39.76
Goto multithread 178.06 345.97 467.99
A
I don't think this calculation is memory-bound at all and I would be
surprised if changing to a 32-bit environment would change things. I
do have a 32-bit chroot environment on these machines (needed for
things like wine and acroread) so I'll try that out but I think I will
need to use Atlas as th
On Fri, 3 Mar 2006, Douglas Bates wrote:
> I have been timing a particular model fit using lmer on several
> different computers and came up with a peculiar result - the model fit
> is considerably slower on a dual-core Athlon 64 using Goto's
> multithreaded BLAS than on a single-core processor.
Paul,
I think what you're seeing is the performance gap between 64-bit binary and
32-bit binary on x86_64. I believe Prof. Ripley had mentioned this several
times in the past.
I do remember back when I was playing with optimized BLAS with R on 32-bit
Linux that I've seen something similar to wha
Doug
This is probably not your reason, but I am finding my dual core Athlon
64 is much slower running 64 bit Linux and R than it was running 32 bit
Linux and R. All the programs are bigger. (Some, like the clock applet,
are a lot bigger for no obvious reason.) The difference is enough to
put
I have been timing a particular model fit using lmer on several
different computers and came up with a peculiar result - the model fit
is considerably slower on a dual-core Athlon 64 using Goto's
multithreaded BLAS than on a single-core processor.
Here is the timing on a single-core Athlon 64 3000
