On 01/09/2013 08:06 PM, Vincent Diepeveen wrote: > > On Jan 10, 2013, at 1:47 AM, Ellis H. Wilson III wrote: >> [snip] > >> My attention to them has revolved more around the latency issues that >> arise when you add a HW raid controller and the SATA protocol between >> the CPU and the SSD. This is exacerbated if the access patterns do >> not >> remain completely random but converge over the lifetime of the run >> -- in >> that case you may run into the accesses localizing onto a single >> SSD in >> the RAID array and bottlenecking performance towards the tail-end >> of the >> execution. > > Ellis, ever heard of striping?
Yup, and I've also heard that stripe sizes vary wildly based on what the vendor allows and what the admin has configured. Many of these chem apps start in a huge space and then converge on a really tight one (which is why so many iterations are needed), so it's not unthinkable, especially if we're slicing the data over many machines already, that the converged-upon space ends up being on just one or two stripes in the array. Good question though, ellis _______________________________________________ Beowulf mailing list, Beowulf@beowulf.org sponsored by Penguin Computing To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf
