Another good writeup on iostat and EBS, from Ilya Grigorik

This is very cool! Applying Mechanical Sympathy optimization techniques to Jetty, specifically: "False sharing" on the BlockingArrayQueue data structure resolved; a new ArrayTernaryTrie data structure to improve header field storage, making it faster to build. look up, efficient on RAM, cheap to GC, and more cache-friendly than a traditional trie; and a branchless hex-to-byte conversion statement. The results are a 30%-faster microbenchmark on amd64, with 50% less Young Gen garbage collections. Lovely to see low-level infrastructure libs like Jetty getting this kind of optimization.

Using new Intel Core i7 instructions to speed up string manipulation. Fascinating stuff. SSE ftw

Interesting trend; Intel moved from a btree to an array-based data structure for their logical-block address indirection map, in order to reduce worst-case latencies (via Martin Thompson)

Excellent stuff. Using "sljit", a stackless platform-independent JIT compiler, this compiles Perl-compatible regular expressions to machine code on ARM, x86, MIPS and PowerPC platforms, resulting in 'similar matching speed to DFA based engines (like re2) on common patterns' with Perl compatibility. 'This work has been released as part of PCRE 8.20 and above. Now (PCRE 8.31), nearly all PCRE features are supported including UTF-8/16 and partial matching.'

'If I were to sum up the past 6 years I've spent optimizing JRuby it would be with the following phrase: Get Rid Of Hash Lookups.'

This has been a particular theme of some recent optimization hacks I've been working on. Hashes may be O(1) to read, on average, but that doesn't necessarily mean they're the right tool for performance...

(via Declan McGrath)

some good advice for large-scale Python performance: prun and guppy for profiling, namedtuples for memory efficiency, and picloud for trivial EC2-based scale-out. (via Nelson)

huh, I'd never heard of sorting networks before

great tips for system-level and app-level profiling on Linux from Padraig

A high-performance compressor optimized for binary data -- 'designed to transmit data to the processor cache faster than a traditional, non-compressed, direct memory fetch via memcpy()' (via Bill de hOra)

very interesting; modern multicore x86 architectures use a NUMA memory architecture, which can cause a dip into swap, even when there appears to be plenty of free RAM available

'a Perl script that generates an OpenOffice.org spreadsheet which loads up SpamAssassin configuration and known spam and ham messages. Once loaded, you can tweak individual SpamAssassin scores in the spreadsheet itself and see their effect on spam/ham classification in real-time. The script also shows you the number of false positives and negatives for a set of scores in real-time.' by Raj Mathur <raju at linux-delhi.org>

Good presentation on C compiler optimization, via Cal Henderson. 'People often write less readable code because they think it will produce faster code. Unfortunately, in most cases, the code will not be faster.' I particularly like 'Fancy-Schmancy Algorithms': 'If you have 10-100 elements, use a list, not a red-black tree; Fancy data structures help on paper, but rarely in reality. (More space overhead in the data structure, less L2 cache left for actual data.)'