Re: [ccp4bb]: Some naive questions about AltiVec

[William Scott <wgscott@chemistry.ucsc.edu>]

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

Hi Derek:

The situation is actually a bit worse in that an individual in another 
group here shelled out the $1300 only to find the Absoft 
vectorization/altivec basically wouldn't work on either CCP4 or CNS.  
Absoft was utterly unhelpful in getting any assistance.  Since this guy 
is our local computer specialist, and is extremely knowledgable and 
competent, it wasn't a question of lack of his experience, etc.  The 
problem clearly resides with Absoft.

However in terms of distributing between two processors, upgrading from 
the 2.95 to 3.1 compilers made a tremendous difference, at least in my 
hands (well over a factor of 2 speed increase in CNS refinements, for 
example).

For now I would urge people to think twice before investing in Absoft's 
products.  It may be that they have improved their product and customer 
service in the year since we had this experience, but please find out 
before they burn you.  Others here have reported similar situations.

For me, computer speed is not the bottleneck in my research program, at 
least when compared to getting good crystals, etc.  I'd hate to see 
CCP4 send them money, especially when that $1300, or its Sterling 
equivalent, could be much better spent, for example, on various 
solvents at the Ring of Bells.

If computational speed is really the bottleneck, I think people are 
probably better off with a 2.8 GHz processor and a Linux operating 
system.

All the best,

Bill


On Tuesday, January 14, 2003, at 12:33 AM, Derek Logan wrote:

>  I was wondering if anyone had attempted to optimize CCP4 code for the 
> G4 processor, in particular to exploit the AltiVec/Velocity Engine.

> The problem is that there only seems to be one automatic vectorization 
> program, i.e. VAST from Veridian. The combined FORTRAN/C optimizer 
> costs $1000! Absoft bundle VAST for FORTRAN with their ProFortran 
> compiler for $1300. This is obviously a bit much for an individual 
> lab, so I wondered if the CCP4 developers, or any other developers and 
> implementers out there, had thought about this issue. Obviously it 
> means producing major amounts of tailor-made code for a specific 
> operating system, which is maybe not true to the CCP4 philosophy.
>