Serious high-performance and lock-free algorithms (by LMAX devs)

[akg]

Here is a paper I found extremely useful when studying Lock Free data structures. It focuses on linked lists, but has a nice exposition of the idea: http://www.cse.yorku.ca/~ruppert/papers/lfll.pdf

[espeed]

Datomic (http://datomic.com) is also built on an immutable, append-only, lock-free design.

[judofyr]

Direct link to the PDF: https://www.dropbox.com/s/dnf2bxtsvu37z3v/pdf-thing.pdf

[Nrsolis]

I'd add that this (http://www.akkadia.org/drepper/cpumemory.pdf) is a good accompaniment to the slides.

[dochtman]

Thanks! InfoQ "presentations" are so crappy...

[bbrtyth]

Thanks!

[aperture]

A great talk, but quite sad that it's on infoq.com , given the difficulty of getting pdf, registration, and that everything is run through .swf so those with flash are left without alternatives. Tidbit of viewing the source gives on lines 989 and 990 some insight on controlling slides, since they seem to load based on time delays. :P It's also quite the array that's created, but I suppose that's all part of making a website.

[mbucc]

This talk is very informative.

The historical bit about NASA using event-sourced design for the Apollo Program (1969) was pretty interesting. He gives IBM big props for hitting 2,300 transactions/second in the mid-60's (with IMS aka DB1).

@17:15 "We've forgetting a lot of this good stuff in our modern designs."

@17:34 "Transaction queues, pulling things off, uncontended, and processing them."

@17:50 "Some of the systems we have today are woeful and can't even get close to that, considering the hardware we have today, and it's ... how we are writing contended designs"

[tomp]

Can anyone tell me, how can the InfoQ presentations be controlled (i.e. next/previous slide)?

[willvarfar]

I haven't worked it out other than clicking on the video timeline; I think the dashes on the timeline on the video show when the slides advance.

In this particular talk, I think the talk is more valuable than skimming the slides. Almost better to listen to than to watch.

[sp332]

If you register, you can download the slides separately.

[SlipperySlope]

I used the Disrupter lock-free queue in my multi-threaded Java application last year. No problems using it and after I replaced the previous concurrent queue, that portion of the code is no longer a profiling hotspot.

It was not in Maven at the time, so I copied the Disrupter packages into my own utilities Maven module.

[willvarfar]

In the middle of the talk, they mention specifically recent changes to disruptor to avoid degrading massively if you have more producers than physical cores.

Anything to get you to watch the talk all the way through ;)