Arne Elofsson <arne at elof.biokemi.su.se> writes:
> --Multipart_Fri_Feb__6_11:34:03_1998-1
> Content-Type: text/plain; charset=US-ASCII
>> Eugen Buehler <buehler at cbil.humgen.upenn.edu> writes:
>> > I would like to hear from anyone that is using Linux in
> > a large scale sequencing project, particularly for programs like
> > phred, phrap, the Staden package, acedb, blast, etc. I compiled
> > phrap on my new Dell at home and compared the time it took to
> > assemble a relatively large (2700 sequences) project. The results:
> >
> > Dell XPS Pentium II 300MHz, 64MB RAM, RH Linux 5.0: 11 minutes
> > Sun Ultra2 Creator 3D, 300MHz, 128MB RAM, Solaris 2.5: 40 minutes
> >
> This differnce is too big. OK solaris is slower than linux but the
> ultraII should be about the same speed as a pentium II
>> > Both were compiled using gcc, optimized.
>> You probably gain a factor of 2 or more using cc (-alotofflags) on
> solarisand about 30 % using pgcc on linux.
>> > Considering the
> > difference in price between these two machines, I'ld like to
> > consider using PC's running Linux if we buy more workstations.
> > Also of interest would be experiences people have networking
> > between Solaris and Linux through NFS filesystems.
>> No problems, however NFS prestanda of linux is really bad (sometimes a
> factor of 4 slower than on solaris). So write to local disks when you
> write a lot of data
The NFS is a bit slower in the 2.0.x kernels but is completely rewritten in
the 2.1.x versions, - but investing in a fast ethercard instead of these
cheap ones people usually put in PC's should give you a 'normal' performance.
-thomas
--
Sicheritz Ponten Thomas E. Department of Molecular Biology
Biomedical Center Uppsala University
BMC: +46 18 4714214 BOX 590 S-751 24 UPPSALA Sweden
Fax +46 18 557723 http://evolution.bmc.uu.se/~thomas
Molecular Tcl: http://evolution.bmc.uu.se/~thomas/tcl
Molecular Linux: http://evolution.bmc.uu.se/~thomas/mol_linux
De Chelonian Mobile ... The Turtle Moves ...