[Pw_forum] Re: Woodcrest vs Opteron performance in pwscf calc.

[Huiqun Zhou]

Hi, Alexander,

A short answer is: Woodcrest (2.66 GHz) > Dempsey (3.2 GHz) > Opteron 280 
(2.6 GHz).

The compute nodes with woodcrest and dempsey in my cluster have 4GB memory, 
one SATA
disk. The nodes with opteron 280 are SUN Fire x4100, each has 8GB memory and 
two SAS
disks.

The test case is total energy calculation of MgAl2O4 (calcium ferrite 
structure). The system has
28 atoms (orthohombic system, Z=4). It took the same number of iterations to 
reach SCF
convergency on the 3 different machines and got physically same results. The 
number in
parantheses is the elapsed time given by PWscf.

Here are numbers
(1) woodcrest (2.66 GHz):
1 core : 3m57s (3m55.86s)
2 cores: 2m11s (2m10.44s)
4 cores: 1m23s (1m17.73s)

(2) dempsey (3.2 GHz)
1 core : (6m26.90s)
2 cores: (3m16.47s)
4 cores: (1m39.74s)

(3) opteron 280 (2.6 GHz)
1 core : 7m13s (7m09.71s)
2 cores: 3m56s (3m52.70s)
4 cores: 2m26s (2m16.72s)

It seems that woodcrest and dempsey are much faster than opteron. The 
scalability of
dempsey is the best, woodcrest is the worst. Despite of the amazing 
performance per
core of woodcrest, it drops to the same level of its predecessor, dempsey, 
when taking
the machine as a unit to evaluate its performance.

But remember one thing: the number for opteron may not be fair. I compiled 
the program
using Intel fortran, Intel MPI 2.0. However, I ever used both Intel and 
PathScale to
compile FFTW and its test cases on opteron machine, I didn't find any 
impressive
differences.

Hope it help you.


Huiqun Zhou


----- Original Message ----- 
From: "Alexander Shaposhnikov" <shaposh at isp.nsc.ru>
To: "Huiqun Zhou" <hqzhou at nju.edu.cn>
Sent: Wednesday, August 02, 2006 2:57 PM
Subject: Woodcrest vs Opteron performance in pwscf calc.


> Hello dear sir,
>
> i have noticed from your posts to PW_forum that you
> have access to both newest Intel Woodcrest platform and
> AMD Opteron. Could you compare the performance of these two
> platforms, for single-threaded, and parallel 4-core pwscf calculations?
>
> Thank you in advance,
> Best Regards,
> A. Shaposhnikov
>
>
>
>