Mandrakesoft announced today the release of the first Linux system capable of real-time computing on Itanium 2 based parallel computing platforms (Symmetric Multiprocessing and cluster computing).
Real-time Linux system for parallel computing for Itanium 2
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
Is there even (big) demand for this? I thought Itanium pretty much died due to it’s price and beeing a pain to program for.
Not quite dead yet, but it will limp along for awhile yet. With most Linux apps running fine on it, it is a viable tux platform. (Some scientific apps run very well on Itanic.) And with HP porting VMS to it, all thoughs users will go Itanic also or have to switch to a different OS, which will only happen in a very cold day in hell. Most VMS boxs stay up for years, they make BSD look unstable. :0
just look at the sgi altix “boxes”, they are itanium2 powered.
my university will get a 2k+ cpu supercomputer basing on it. so i guess there IS a high demand for it in general HPC, especially considering that there are not many other cpu and hardware i/o architectures that are built for such tasks.
It isn’t that Itanium is a bad performer – far from it – but that it doesn’t run x86 binaries very well and it isn’t priced for desktop use.
You’d be hard-pressed to genuinely have the requirement for that much crunch at home.
Altix is prety much the ONLY Itanium-based system which sells, and it’s known in HPC to be pretty poor value for money. Opteron outsold Itanium within its month of release, and the main big iron vendors aren’t interested in Itanic – IBM have their POWER chips, Cray have embraced Opteron, and Sun will still sell you Ultrasparc if you want it.
We have a spare dual itanium 2 compute node that Intel gave us for free to show how great Itanic is. Know what? It sucks, hard.
We have a spare dual itanium 2 compute node that Intel gave us for free to show how great Itanic is. Know what? It sucks, hard.
You can always mail them to me if you don’t like them
In every Itanium-related thread there appear people who seem to have some perverse joy in saying “Itanic is dying! Muahhaha!”. This is really funny.
Why would people want to do real time computing on Itanium 2?
Wouldn’t the reduced throughput, inability to cluster and multiprocessing problems caused by hard real time requirements make it useless for number crunching?
What scientific problems require it?
I know there must be reason for Mandrake to do this, I just don’t understand who would use it.
This is part of a larger project, see http://www.hyades-itea.org/
Here is a paper outlining their goals: http://www.lifl.fr/west/publi/PMSD04rr06.pdf Appearantly, they are working on an architecture, called ARTiS, that allows some CPUs to be decdicated to realtime tasks while others are able to address non-realtime tasks. So, I suppose this could be useful for control systems that must simultaniously control hardware using sensor feedback (realtime) while having a seperate process that searches for some globally optimized feedback. (e.g., flying a plane while searching for an optimal flight path that accomidates realtime weather and the locations of other aircraft) Doing this one one computer, rather than having two seperate computers could lead to smaller, lighter, cheaper, architecturally simpler control systems.
http://www.intel.com/products/roadmap/server.htm
look at the July – December 2005 plans for Itanuim. think itanium 3. intel is devoting more time and money into itanuim that ever. think about it