[daip] Re: aips performance (fwd)

[Eric Greisen]

Lincoln writes:
 > Hi Eric,
 > 
 > I've also been in touch with Andy Briggs at ROE.  No luck.
 > I'll close this out and turn my questions to inside SAO after one more 
 > directed to you, if you don't mind.
 > 
 > Our better machines have EIDE RAID0, 3.2-3.6 GHz PIV or Xeon, 2.1-4.0 GB 
 > of RAM.  I would expect to do better than our observed aipsmark 108 - 135 
 > - in light of your aipsmark=190 (PIV 3.2 GHz).
 > 
 > I'd appreciate it if you would please note any other off-the-cuff specs 
 > for your machine beyond processor type and speed.  After that I'll lobby 
 > for us to start testing w/ and w/o the server, w/ and w/o an electric 
 > cattle prod in the CPU socket, etc.  Our underperformance is pretty 
 > serious.
 > 
 > Best,
 > Lincoln
 > 
 > 
 > -- 
 > Lincoln J. Greenhill               Smithsonian Astrophysical Observatory
 > Office:     1 617-495-7194         60 Garden St, Cambridge, MA 02138
 > Cell:       1 617 851 8918         E-mail:  greenhill at cfa.harvard.edu
 > FAX:        1 617-495-7345

At a cost I am told of $1000

primate<479>$ cat /proc/cpuinfo
processor       : 0
vendor_id       : GenuineIntel
cpu family      : 15
model           : 4
model name      : Intel(R) Pentium(R) 4 CPU 3.20GHz
stepping        : 3
cpu MHz         : 3192.534
cache size      : 2048 KB
fdiv_bug        : no
hlt_bug         : no
f00f_bug        : no
coma_bug        : no
fpu             : yes
fpu_exception   : yes
cpuid level     : 5
wp              : yes
flags           : fpu vme de pse tsc msr pae mce cx8 apic mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe nx lm pni monitor ds_cpl est cid xtpr
bogomips        : 6388.14

New primate
   Intel load modules              Y2K 207-211   (3.224-3.243)
   GNU   3.4.4                         316       (cpu 4.934)

or an aips mark (00) of 191 w the Intel load modules served from our
server or an aips mark of 127 with the GNU compiler served directly on
primate.

There are two large disks
/dev/sda3             129G   63G   60G  52% /export/home/primate
/dev/hdb1             151G   57G   94G  38% /export/home/primate2
but I run the test with the input and output disks set to the the same
value.  (scratch files may go elsewhere)  The print file and message
file must be local to the machine.

The options we use are in the FDEFAULT.SH files we ship.

Are you sure you are not accidentally using the GNU compiler?

Eric Greisen