Memory Management Technique Speeds Apps By 20%

Dotnaught writes "A paper (PDF) to be presented later this month at the IEEE International Parallel and Distributed Processing Symposium in Atlanta describes a new approach to memory management that allows software applications to run up to 20% faster on multicore processors. Yan Solihin, associate professor of electrical and computer engineering at NCSU and co-author of the paper, says that using the technique is just a matter of linking to a library in a program that makes heavy use of memory allocation. The technique could be especially valuable for programs that are difficult to parallelize, such as word processors and Web browsers." Informationweek has a few more details from an interview with Solihin.

Comments

Re:Nothing to see here....

[Anonymous Coward]

new malloc()

I see what you did there.

Re:Summary

[14erCleaner]

The speedup comes from using a memory-allocation library (PHKMalloc) that does extensive and expensive checking to avoid programmer errors, then basically hides most of its overhead in the second thread (so that the allocation thread would be mostly doing sanity checking). For most programs, this probably won't help performance any. It's an old trick in parallel processing research: pick a slow algorithm, then speed it up via parallelism, rather than starting out with an efficient solution.

I once submitted an April Fool's joke to this effect to a moderated website; I claimed superlinear speedup for sorting by starting with a bubble sort, then "modifying it for parallelism" until it morphed into a quicksort. Alas, the moderator rejected it...

Re:It's programmers that need parallelization

[__aasqbs9791]

...If I still coded much anymore it would drive me to drink.

Maybe that's my problem? If I started drinking maybe I could handle it [programming for other people] again.

Re:Nothing to see here....

[Angst Badger]

This trick absolutely cannot be used in real life - it's useful only when the operating system runs exactly one process, a scenario that occurs only in research papers.

On the contrary, this opens up whole new possibilities for MS-DOS!

Re:20%?!

[guyminuslife]

I was aware that malloc() had a price tag attached, but free()? That's misleading advertising.

Re:Beware the key term there:

[ILostMyOldUserID]

Checkbox with a compiler?!! I denounce you as a Windows user - hand in your geek card!

Re:Beware the key term there:

[Anonymous Coward]

He could be talking about Eclipse, but I think that's just as bad.

Re:Beware the key term there:

[Idiomatick]

I just assumed Windows needed the help more, my bad!