|
| 1 | +Release Notes for MiniSat 2.2.0 |
| 2 | +=============================== |
| 3 | + |
| 4 | +Changes since version 2.0: |
| 5 | + |
| 6 | + * Started using a more standard release numbering. |
| 7 | + |
| 8 | + * Includes some now well-known heuristics: phase-saving and luby |
| 9 | + restarts. The old heuristics are still present and can be activated |
| 10 | + if needed. |
| 11 | + |
| 12 | + * Detection/Handling of out-of-memory and vector capacity |
| 13 | + overflow. This is fairly new and relatively untested. |
| 14 | + |
| 15 | + * Simple resource controls: CPU-time, memory, number of |
| 16 | + conflicts/decisions. |
| 17 | + |
| 18 | + * CPU-time limiting is implemented by a more general, but simple, |
| 19 | + asynchronous interruption feature. This means that the solving |
| 20 | + procedure can be interrupted from another thread or in a signal |
| 21 | + handler. |
| 22 | + |
| 23 | + * Improved portability with respect to building on Solaris and with |
| 24 | + Visual Studio. This is not regularly tested and chances are that |
| 25 | + this have been broken since, but should be fairly easy to fix if |
| 26 | + so. |
| 27 | + |
| 28 | + * Changed C++ file-extention to the less problematic ".cc". |
| 29 | + |
| 30 | + * Source code is now namespace-protected |
| 31 | + |
| 32 | + * Introducing a new Clause Memory Allocator that brings reduced |
| 33 | + memory consumption on 64-bit architechtures and improved |
| 34 | + performance (to some extent). The allocator uses a region-based |
| 35 | + approach were all references to clauses are represented as a 32-bit |
| 36 | + index into a global memory region that contains all clauses. To |
| 37 | + free up and compact memory it uses a simple copying garbage |
| 38 | + collector. |
| 39 | + |
| 40 | + * Improved unit-propagation by Blocking Literals. For each entry in |
| 41 | + the watcher lists, pair the pointer to a clause with some |
| 42 | + (arbitrary) literal from the clause. The idea is that if the |
| 43 | + literal is currently true (i.e. the clause is satisfied) the |
| 44 | + watchers of the clause does not need to be altered. This can thus |
| 45 | + be detected without touching the clause's memory at all. As often |
| 46 | + as can be done cheaply, the blocking literal for entries to the |
| 47 | + watcher list of a literal 'p' is set to the other literal watched |
| 48 | + in the corresponding clause. |
| 49 | + |
| 50 | + * Basic command-line/option handling system. Makes it easy to specify |
| 51 | + options in the class that they affect, and whenever that class is |
| 52 | + used in an executable, parsing of options and help messages are |
| 53 | + brought in automatically. |
| 54 | + |
| 55 | + * General clean-up and various minor bug-fixes. |
| 56 | + |
| 57 | + * Changed implementation of variable-elimination/model-extension: |
| 58 | + |
| 59 | + - The interface is changed so that arbitrary remembering is no longer |
| 60 | + possible. If you need to mention some variable again in the future, |
| 61 | + this variable has to be frozen. |
| 62 | + |
| 63 | + - When eliminating a variable, only clauses that contain the variable |
| 64 | + with one sign is necessary to store. Thereby making the other sign |
| 65 | + a "default" value when extending models. |
| 66 | + |
| 67 | + - The memory consumption for eliminated clauses is further improved |
| 68 | + by storing all eliminated clauses in a single contiguous vector. |
| 69 | + |
| 70 | + * Some common utility code (I/O, Parsing, CPU-time, etc) is ripped |
| 71 | + out and placed in a separate "utils" directory. |
| 72 | + |
| 73 | + * The DIMACS parse is refactored so that it can be reused in other |
| 74 | + applications (not very elegant, but at least possible). |
| 75 | + |
| 76 | + * Some simple improvements to scalability of preprocessing, using |
| 77 | + more lazy clause removal from data-structures and a couple of |
| 78 | + ad-hoc limits (the longest clause that can be produced in variable |
| 79 | + elimination, and the longest clause used in backward subsumption). |
0 commit comments