Talk:Symmetric multiprocessor system
![]() | This is the talk page of a redirect that has been merged and now targets the page: • Symmetric multiprocessing Because this page is not frequently watched, present and future discussions, edit requests and requested moves should take place at: • Talk:Symmetric multiprocessing Merged page edit history is maintained in order to preserve attributions. |
![]() | This redirect does not require a rating on Wikipedia's content assessment scale. It is of interest to the following WikiProjects: | ||||||||||
|
So why not just say this in symmetric multiprocessing?
[edit]Why not just move whatever's here to symmetric multiprocessing and make this a redirect? Guy Harris (talk) 19:27, 30 October 2011 (UTC)
I agree, this needs to be folded in with the more-complete symmetric multiprocessing page, especially if this is the page that comes up first when I do a search for SMP, when it is incomplete as compared to the other page of nearly the same name that it shadows. mmpozulp (talk) 03:06, 3 December 2013 (UTC)
Sorry but I am completely disagree with you for two reasons. First of all "symmetric multiprocessor" and "symmetric multiprocessing" are two different concepts and so that the two "voices" need to be separated. When you search for "SMP" you must find "symmetric multiprocessor" because this term is refereed an hardware architecture (see for instance SMP definition by IBM in Glossary pg. 246 [1] (see also other references). Second, symmetric multiprocessing is a type of processing obtained by using an SMP system or also a NUMA system (clustering of SMPs). If you merge SMP concept in multiprocessing definition why not put inside also NUMA system definition ? Ferry24.Milan (talk)
- I.e., a "symmetric multiprocessor" is a system where, purely at the hardware level, all processors are equivalent and can take on any role, even if the OS assigns one or more processors different roles (for example, only running kernel-mode code on a "master" processor), while "symmetric multiprocessing" is what you have when a symmetric multiprocessor is running an OS that doesn't assign particular roles to particular processors? Guy Harris (talk) 09:22, 6 December 2013 (UTC)
A general definition of SMP is based on its capability rather on its possible use
I report the IBM definition:
- "SMP. Symmetric Multiprocessor. Flat multi-processor system where each processor has equal access to all of memory, I/O, and interrupts " [1]
Other definition (Introduction to SMP Systems - Arwed Starke) :
- "The abbreviation SMP stands for tightly coupled, shared memory symmetric multiprocessor system. As set of equal CPUs accesses a common physical memory (and I/O ports) via a shared front side bus " [2].
From a IBM patent:
- "In a symmetric multi-processor (SMP) computer, all of the processing units are generally identical, that is, they all use a common set or subset of instructions and protocols to operate, and generally have the same architecture " [3]
From Intel (MultiProcessor Specification - System Overview)
- "... The MP specification’s model of multiprocessor systems incorporates a tightly-coupled, shared-memory architecture with a distributed interprocessor and I/O interrupt capability. It is fully symmetric; that is, all processors are functionally identical and of equal status, and each processor can communicate with every other processor. There is no hierarchy, no master-slave relationship, no geometry that limits communication only to “neighboring” processors. The model is symmetric in two important respects:
- Memory symmetry.
- Memory is symmetric when all processors share the same memory space and access that space by the same addresses...
- I/O symmetry.
- I/O is symmetric when all processors share access to the same I/O subsystem (including I/O ports and interrupt controllers) and any processor can receive interrupts from any source... " [4] - Ferry24.Milan (talk)
From a general point of view, the answer to the previous Guy Harris question is "yes" - Ferry24.Milan (talk)
References
- ^ a b "An Introduction to the New IBM e-server pSeries High Performance Switch" - Glossary pg. 246 - http://www.redbooks.ibm.com/redbooks/pdfs/sg246978.pdf
- ^ Locking in OS Kernels for SMP Systems - 1.1 Introduction to SMP Systems p. 2 - http://irl.cs.ucla.edu/~yingdi/web/paperreading/smp_locking.pdf
- ^ http://www.google.nl/patents/US6349369
- ^ Intel MultiProcessor Specification - 2. System Overview - http://pdos.csail.mit.edu/6.828/2007/readings/ia32/MPspec.pdf
Why does heterogeneous computing redirect here?
[edit]This articles states that a symmetric multiprocessor system "is not a heterogeneous computing system." — Preceding unsigned comment added by 50.72.53.26 (talk) 06:42, 11 April 2014 (UTC)
- It is an obvious mistake (already corrected) Ferry24.Milan (talk). — Preceding undated comment added 20:46, 16 April 2014 (UTC)
Blank-and-redirect
[edit]As discussed at Talk:Symmetric_multiprocessing#Merger_proposal, consensus is to merge this page with Symmetric multiprocessing. I'm replacing this page with a blank and redirect to that page.
Looking back in this page's history, it looks like it was created by a single user and largely filled with content duplicating Symmetric multiprocessing. This same user did something similar, creating the Cache memory page when equivalent content already existed at CPU cache and Cache coherency. These may have been good faith edits, but their results was disruptive. I have already replaced Cache memory with a blank-and-redirect.
I don't believe there is any salvageable content here that isn't already on Symmetric multiprocessing.