2 __ _____ _ __ ___(_) ___ _ __ |___ / |___ /| || |
3 \ \ / / _ \ '__/ __| |/ _ \| '_ \ |_ \ |_ \| || |_
4 \ V / __/ | \__ \ | (_) | | | | ___) | ___) |__ _|
5 \_/ \___|_| |___/_|\___/|_| |_| |____(_)____(_) |_|
7 The "Desktop Grid needs love too" release (also called "Xmas release").
8 Most important changes:
9 * Big speedup through lazy evaluation of the linear models
10 * Supernovae mode: Compile everything in one unit to improve inlining
11 * Simix network module for internal cleanups
12 * Load DAX of applications into SimDag
13 * Lot of small cleanups and other bug fixes
16 __ _____ _ __ ___(_) ___ _ __ |___ / |___ / |___ /
17 \ \ / / _ \ '__/ __| |/ _ \| '_ \ |_ \ |_ \ |_ \
18 \ V / __/ | \__ \ | (_) | | | | ___) | ___) | ___) |
19 \_/ \___|_| |___/_|\___/|_| |_| |____(_)____(_)____/
21 The "Need for Speed" release.
22 Big speedup through some function inlining.
25 __ _____ _ __ ___(_) ___ _ __ |___ / |___ / |___ \
26 \ \ / / _ \ '__/ __| |/ _ \| '_ \ |_ \ |_ \ __) |
27 \ V / __/ | \__ \ | (_) | | | | ___) | ___) | / __/
28 \_/ \___|_| |___/_|\___/|_| |_| |____(_)____(_)_____|
30 The "Simplicity does not preceed complexity, but follows it" release.
31 Most important changes:
32 * surf and simix modules reworked for simplification.
33 It lays the ground for future extensions and improvements of SimGrid
34 * SMPI was further improved, altrough not ready for production use yet.
36 On the way, we gained a bit more than 5% on the classical master/slave
37 example. More to come on this in future releases.
39 __ _____ _ __ ___(_) ___ _ __ |___ / |___ / / |
40 \ \ / / _ \ '__/ __| |/ _ \| '_ \ |_ \ |_ \ | |
41 \ V / __/ | \__ \ | (_) | | | | ___) | ___) || |
42 \_/ \___|_| |___/_|\___/|_| |_| |____(_)____(_)_|
45 This dot release is mainly a maintainance one. Most important changes:
46 * We fixed a large amount of bugs all around
48 * We sanitized the way configuration is handled internally.
49 Try passing --cfg-help to binaries compiled against this version.
51 * SMPI is in better shape: lot of bugs fixing & usability improvements
52 It may be worth trying it (even if all bugs are not gone yet)
54 This version may have a bit more of memleaks than 3.3. This will be
55 fixed in a latter release.
59 / ___|(_)_ __ ___ / ___|_ __(_) __| |
60 \___ \| | '_ ` _ \| | _| '__| |/ _` |
61 ___) | | | | | | | |_| | | | | (_| |
62 |____/|_|_| |_| |_|\____|_| |_|\__,_|
64 __ _____ _ __ ___(_) ___ _ __ |___ / |___ /
65 \ \ / / _ \ '__/ __| |/ _ \| '_ \ |_ \ |_ \
66 \ V / __/ | \__ \ | (_) | | | | ___) | ___) |
67 \_/ \___|_| |___/_|\___/|_| |_| |____(_)____/
69 _ _ _ _ | |___ __ _ __| |_ ___ __| |
70 | || | ' \| / -_) _` (_-< ' \/ -_) _` |
71 \_,_|_||_|_\___\__,_/__/_||_\___\__,_|
74 homepage: http://simgrid.gforge.inria.fr/
75 download: http://gforge.inria.fr/frs/?group_id=12
77 At least, after 2 years of hard work, we managed to release the 3.3
78 version of the SimGrid framework. There is so many changes that the
79 changelog only lists the most important ones, leaving alone the small
80 improvements, bug fixing and new gadgets.
84 * New simulation models, and improvement of the GTNetS wrapper
86 * Large memory savings (mainly in parser)
87 * Faster (twice faster is not uncommon, and from 20 hours to 2
88 minutes on very large scenarios)
89 * Much better scalability (tested up to 250,000 processes)
91 * Complete regression testing framework so that you can trust the tool
92 * Lot of neat new modules in the XBT toolbox
94 This version was throughfully tested on linux 32bits and 64bits
95 (debian), as well as Mac OSX (leopard).
97 Unfortunately, our windows-guy left, and we cannot release the windows
98 version at the same time than the other archs. Any help would be
101 Some of the 96 included test suites are known to fail, but everything
102 should work anyway (don't panic):
103 * The amok module does not work in real deployment ATM, but I don't
104 see this as release critical since I'm not aware of anyone needing
106 * Some tests about the ability of GRAS to receive messages from
107 exotic platforms fail because I've lost access to these platforms
109 * the example/gras/pmm sometimes fails (about 1/10 of cases). I'm
110 quite puzzled, but I suspect a bug in the test, not in the lib.
111 * the tesh auto-tests "fail" on Mac OSX. This is because "rm -rf" is
112 sometimes too verbose (when the OS creates hidden files, I
113 suspect), but tesh definitly work as the rest on this arch.
115 We hope to manage to do more timely releases in the future, even if
116 that may turn out difficult since big stuff is comming (I don't say
117 much here for the suspense ;)
119 We hope you'll enjoy this new version, and please report any feedback
120 on the <simgrid-user@lists.gforge.inria.fr> list.
122 Martin Quinson (for Da SimGrid Team)