Re: [ccp4bb]: Comments on G4 Ti

[Anastassis Perrakis <perrakis@nki.nl>]

***  For details on how to be removed from this list visit the  ***
***          CCP4 home page http://www.ccp4.ac.uk         ***

Dear all -

Happy new year to everybody to start with - it's always a good start
with yet another successful CCP4 meeting... ;-)

Taking the opportunity from the OSX question, here is some oil for the 
more general fire of
'which computer to choose'. Over the holiday season I benchmarked our 
collection of
different computers. I used - what else - ARP/wARP. ARP/wARP is a good 
real-life benchmark:
It runs lots of floating point intensive Refmac, has lots of I/O for 
the maps and runs the integer/float
intensive autobuilding routines which also eat up lots of memory. So, i 
consider it to
be very relevant as a 'crystallographic bencmark'.

I hope that these get handy to some people and sorry that they are not 
really relevant to the
'which notebook' question since in choosing an notebook I personally 
feel there are
other parameters than crystallographic performance that play a major 
role ...(for me) !

In the list below '1' indicates the fastest and i.e. '2' indicates that 
a machine is twice as slow.
I also give an indication of how 'old' the machine is to make the 
comparison a bit more fare.
I don't include a price indication since it's not fare, different 
graphics boards, maintenance
contracts, etc, etc, but, ask your local dealer ;-))

P4 2.8 GHz Xeon, Intel Compilers:		1.00 (1 month)
P4 2.8 GHz Xeon, CCP4 distribution 	1.43	(1 month)
ES 40 0.67 GHz EV667, -O1 optimised	1.69 (55 months)
P4 1.9 GHz, CCP4 distribution			2.73 (6 months)
PIII 1.8 GHz, CCP4 distribution			3.19 (12 months)
ES40 0.67 GHz EV667, CCP4 distr.		3.51 (55 months)
OSX PPC 1GHz, CCP4 distr.			5.11 (3 months)
SGI R10k 0.2 GHz					7.39 (60 months)

So, besides the obvious merits of OSX machines (which led me to have an 
iMAC
both at home and at work in my desk) any Pentium 4 is biting the hell 
out of it (for computation)
at the same or slightly lesser cost.

I am also VERY impressed by the fact the the 3-year old Compaq
is still *almost* as fast as the newest P4, for what we do ! Plus you 
dont have to mess with
Intel compilers which are a serious headache alleviated only for CCP4 
due to
the work of Clemens (try out his ccp4_cv.sh script and his patches!!!)
..... plus the fact that besides the very heavy load we put on it it has 
never 'crashed',
which is not what I would say for Linux - although redHat 8.0 has 
proved so far very stable)


	Tassos

On Thursday, Jan 2, 2003, at 23:43 Europe/Amsterdam, 
<Zwart@EMBL-Hamburg.DE> wrote:

> ***  For details on how to be removed from this list visit the  ***
> ***          CCP4 home page http://www.ccp4.ac.uk         ***
>
> Quoting "Alex W. Smith" <alex.smith@asu.edu>:
>
>> Hello all:
>>
>> I am thinking about purchasing a new G4 power book and would be 
>> interested
>> in any comments current users of the new OSX may have.  I would
>> particularly
>> appreciate any comments on how CCP4, mosflm, O and assorted other
>> crystallography programs run on the OSX platform.
>>
>> Regards,
>>
>> Alex Smith
>
>
> I have a nice Mac OSX work horse and i am quite happy with ccp4 and 
> arp warp.
> it runs without any problems. If i am not mistaken, most of the 
> upsalla stuff
> (Moleman, mapman etc) runs as well on Mac OSX.
>
> On the whole, i am quite happy with it. The only drawback is that i am 
> an
> Xtalview fan, which is not (yet?????) available for Mac OSX. For this 
> reason, I
> still have a linux machine occupying some desk space.
> O does run though, but i can't give any feedback on that ....
>
> greetings,
>
> Peter
>
>
>
>