For Java, individual JVMs have been seen to run happily with good scaling at 16 CPUs now, and maybe more. For database applications like Oracle and DB2 there have been reports of linear scale at 72 CPUs. So, Amdahl's law (and I met him - he's a great man hardly given the credit due him!) really applies to a particular application or algorithm with both serial and parallel components. As the number of CPUs rises, the serial portion tends to dominate - but that's not relevant for *separate* applications!
The confrontation between Gandalf and the Nazgul king is one of the most dramatic scenes in the trilogy - you would hope that it would get included in the movie sooner or later. That scene also showed up in the pre-opening ads for RoTK on TV.
Also it looks like they dropped the scene (in the theater) of when that orc with the nasty skin condition (well, that's redundanct, for an orc) - the one saying 'this is the Age or Orcs' - gets off'ed in the battle of Pellenor fields.
Actually, this is more like IBM's VM, but not exactly like that either - read the posts here and you'll see it does NOT create virtual machines (each of which requires its own operating system). LPAR gives you only a small and fixed number of OS contexts on a box: a z900 goes up only to 16! Virtualization via VM lets you have hundreds or several thousand, Zones lets you have hundreds or several thousands with less overhead.
FWIW: LPARs were introduced by IBM in 1987 (plus or minus a year), and it was imitating Amdahl's MDF feature.
The resource management part has been in Solaris for quite a while, letting you control how much CPU (and other resources) an application can get. What the containers add is the ability to provide isolated environments (namespace, filesystems, process list, security and fault containment contexts) so each container thinks it has its own instance of Solaris.
For the authoritative source, see Melinda Varian's "History of VM and the VM User Community" at http://pucc.princeton.edu/~melinda which describes this in detail.
The bundle was called VM/370 before being renamed (and enhanced) as VM/SP; subsequently there were VM/XA, VM/ESA and current z/VM. Quite correct that the user community was instrumental in keeping it alive and that SHARE and the universities were key in this. And, it was (and mostly remains) available in source code to its licensees. There was a bitter fight with IBM over that but they came around (the VM team always had their hearts in the right place; it was the other guys needing convincing)
The CP portions of CP/67 of
VM/370 and their descendants weren't as derivative of CTSS as CMS because CTSS wasn't a hypervisor.
There are some mistakes in the parent post: VM was not developed in academia, it was originally developed by IBM at IBM labs in Cambridge Massachusetts (not Cambridge University - hence the common confusion about who created VM), and later in Kingston and Endicott, New York.
Also, while it was distributed in source format for years, it was not "open source" as it we now describe it. An active user community distributed enhancements to VM, but the OS belonged to IBM at all times. Unfortunately, in 1983 IBM introduced a policy called "Object Code Only" (OCO) that planned to remove all source code. The VM user community fought hard against this, and managed to (mostly) retract this policy for the VM OS itself. It's should be noted that IBM is most fond, to this day, of open-sourcing other people's OSes, not their own (you don't get source to z/OS, DB2, etc, and it's not because it contains other people's intellectual property)
IBM did have a version of Unix for mainframes, like UTS: it was called AIX/ESA, and for better or worse it never went anywhere.
Slashdot Mirror
For Java, individual JVMs have been seen to run happily with good scaling at 16 CPUs now, and maybe more. For database applications like Oracle and DB2 there have been reports of linear scale at 72 CPUs. So, Amdahl's law (and I met him - he's a great man hardly given the credit due him!) really applies to a particular application or algorithm with both serial and parallel components. As the number of CPUs rises, the serial portion tends to dominate - but that's not relevant for *separate* applications!
The confrontation between Gandalf and the Nazgul king is one of the most dramatic scenes in the trilogy - you would hope that it would get included in the movie sooner or later. That scene also showed up in the pre-opening ads for RoTK on TV.
Also it looks like they dropped the scene (in the theater) of when that orc with the nasty skin condition (well, that's redundanct, for an orc) - the one saying 'this is the Age or Orcs' - gets off'ed in the battle of Pellenor fields.
FWIW: LPARs were introduced by IBM in 1987 (plus or minus a year), and it was imitating Amdahl's MDF feature.
You bet - Look at http://wwws.sun.com/software/solaris/10/
The resource management part has been in Solaris for quite a while, letting you control how much CPU (and other resources) an application can get. What the containers add is the ability to provide isolated environments (namespace, filesystems, process list, security and fault containment contexts) so each container thinks it has its own instance of Solaris.
The bundle was called VM/370 before being renamed (and enhanced) as VM/SP; subsequently there were VM/XA, VM/ESA and current z/VM. Quite correct that the user community was instrumental in keeping it alive and that SHARE and the universities were key in this. And, it was (and mostly remains) available in source code to its licensees. There was a bitter fight with IBM over that but they came around (the VM team always had their hearts in the right place; it was the other guys needing convincing)
The CP portions of CP/67 of VM/370 and their descendants weren't as derivative of CTSS as CMS because CTSS wasn't a hypervisor.
Also, while it was distributed in source format for years, it was not "open source" as it we now describe it. An active user community distributed enhancements to VM, but the OS belonged to IBM at all times. Unfortunately, in 1983 IBM introduced a policy called "Object Code Only" (OCO) that planned to remove all source code. The VM user community fought hard against this, and managed to (mostly) retract this policy for the VM OS itself. It's should be noted that IBM is most fond, to this day, of open-sourcing other people's OSes, not their own (you don't get source to z/OS, DB2, etc, and it's not because it contains other people's intellectual property)
IBM did have a version of Unix for mainframes, like UTS: it was called AIX/ESA, and for better or worse it never went anywhere.