2008-08-01 |
Christoph Mallon | Skip a single downconv even if it has multiple users. |
tree | commitdiff |
2008-07-31 |
Michael Beck | - use double-linked lists instead of flexible arrays... |
tree | commitdiff |
2008-07-31 |
Michael Beck | - BugFix: when splitting by input, ensure than Z is... |
tree | commitdiff |
2008-07-30 |
Matthias Braun | simply deactivate edges instead of asserting it |
tree | commitdiff |
2008-07-30 |
Michael Beck | - inline_method moves copied constants to start block yet |
tree | commitdiff |
2008-07-30 |
Christoph Mallon | Sort the inputs of a Sync before it is being created... |
tree | commitdiff |
2008-07-30 |
Christoph Mallon | Reduce worst time complexity of mem parallelisation... |
tree | commitdiff |
2008-07-30 |
Michael Beck | - do not kill the start block when folding control... |
tree | commitdiff |
2008-07-30 |
Michael Beck | - opt_tail_recursion() must run before optimize_funccal... |
tree | commitdiff |
2008-07-30 |
Michael Beck | - commented out the wrong line |
tree | commitdiff |
2008-07-30 |
Michael Beck | - add compute_Confirm(), compute_Bad() and compute_Unkn... |
tree | commitdiff |
2008-07-29 |
Michael Beck | - BugFix: of by one error |
tree | commitdiff |
2008-07-29 |
Michael Beck | - using tarval_top as an alias for tarval_unreachable... |
tree | commitdiff |
2008-07-29 |
Michael Beck | Add NaN condition to Sub AND Cmp |
tree | commitdiff |
2008-07-29 |
Michael Beck | - BugFix: constants must be evaluated BEFORE congruence... |
tree | commitdiff |
2008-07-29 |
Michael Beck | - BugFix: fixed the "type is neither Top nor constant... |
tree | commitdiff |
2008-07-28 |
Michael Beck | - BugFix: Proj nodes must compute TOP if they are in... |
tree | commitdiff |
2008-07-26 |
Michael Beck | BugFix: |
tree | commitdiff |
2008-07-26 |
Michael Beck | - BugFix: must check the input of the user node y,... |
tree | commitdiff |
2008-07-26 |
Michael Beck | fixed warnings |
tree | commitdiff |
2008-07-26 |
Michael Beck | - BugFix: a node might fall 2 times, so we need a on_fa... |
tree | commitdiff |
2008-07-26 |
Michael Beck | BugFixes: |
tree | commitdiff |
2008-07-26 |
Michael Beck | - first working combo version (at least one example... |
tree | commitdiff |
2008-07-25 |
Michael Beck | add sorting of out edges, some small fixes |
tree | commitdiff |
2008-07-25 |
Michael Beck | - out edges now preserve the input position, needed... |
tree | commitdiff |
2008-07-25 |
Michael Beck | - fixed type_or_ent type: get rod of casts |
tree | commitdiff |
2008-07-24 |
Michael Beck | - implemented compute_Proj_Cond(), compute_Add() |
tree | commitdiff |
2008-07-24 |
Michael Beck | - add comment |
tree | commitdiff |
2008-07-24 |
Michael Beck | - renamed find_const() to find_const_or_confirm() |
tree | commitdiff |
2008-07-24 |
Michael Beck | - CondEval now really can handle Confirms |
tree | commitdiff |
2008-07-23 |
Michael Beck | add dumping hook |
tree | commitdiff |
2008-07-21 |
Michael Beck | - lattice elements can now contain Symbolic Constants |
tree | commitdiff |
2008-07-16 |
Michael Beck | fixed crop list handling |
tree | commitdiff |
2008-07-15 |
Michael Beck | - add compute for Sub and Cmp |
tree | commitdiff |
2008-07-14 |
Michael Beck | mostly finished version of combo |
tree | commitdiff |
2008-07-14 |
Matthias Braun | don't emit debug messages |
tree | commitdiff |
2008-07-10 |
Michael Beck | initial version of Cliff Click's Combo Optimization |
tree | commitdiff |
2008-07-09 |
Michael Beck | -BugFix: exp_gen sets must be cleaned |
tree | commitdiff |
2008-07-09 |
Michael Beck | - BugFix: Antic_in computation must always add exp_gen |
tree | commitdiff |
2008-07-08 |
Michael Beck | - add a doxygen comment |
tree | commitdiff |
2008-07-08 |
Michael Beck | fixed warnings |
tree | commitdiff |
2008-07-08 |
Michael Beck | removed unused (and buggy) trans_set, thios is NOT... |
tree | commitdiff |
2008-07-08 |
Michael Beck | for now, deactivate edges, they will crash with this... |
tree | commitdiff |
2008-07-08 |
Michael Beck | new GVN-PRE implementation |
tree | commitdiff |
2008-06-19 |
Michael Beck | - BugFix: whech checking Blocks, is_Block_dead() must... |
tree | commitdiff |
2008-06-18 |
Christoph Mallon | Avoid endless iteration when normalising returns. |
tree | commitdiff |
2008-06-18 |
Michael Beck | - make normalize_n_returns() more aggressive: remove... |
tree | commitdiff |
2008-06-18 |
Michael Beck | BugFix: must initialize Blocks with removable |
tree | commitdiff |
2008-06-18 |
Michael Beck | Hopefully better version of control flow optimization: |
tree | commitdiff |
2008-06-12 |
Michael Beck | workaround for dead CF: restart the optimization |
tree | commitdiff |
2008-06-11 |
Michael Beck | - add a version of remove_critical_edges() that handles... |
tree | commitdiff |
2008-06-06 |
Michael Beck | - export classify_pointer() |
tree | commitdiff |
2008-06-02 |
Matthias Braun | handle Bads more gracefully |
tree | commitdiff |
2008-06-01 |
Michael Beck | call optimize_cf() to delete dead-code that might arise... |
tree | commitdiff |
2008-05-30 |
Michael Beck | inline_method(): |
tree | commitdiff |
2008-05-30 |
Michael Beck | handle a call with less arguments than needed gracefull... |
tree | commitdiff |
2008-05-27 |
Michael Beck | BugFix: phi node might be exchanged, read the next... |
tree | commitdiff |
2008-05-27 |
Michael Beck | - BugFix: set the mark on the right block |
tree | commitdiff |
2008-05-27 |
Michael Beck | BugFix: fixed Phi-list manipulation |
tree | commitdiff |
2008-05-26 |
Michael Beck | add some comments, small cosmetic |
tree | commitdiff |
2008-05-26 |
Michael Beck | must use Lg, not Ne to check for != |
tree | commitdiff |
2008-05-26 |
Michael Beck | use block Phi lists and Block marks to get rid of block... |
tree | commitdiff |
2008-05-26 |
Michael Beck | C99 features removed |
tree | commitdiff |
2008-05-25 |
Matthias Braun | fix a few warnings |
tree | commitdiff |
2008-05-25 |
Matthias Braun | commit the beginning of advanced boolopt (had that... |
tree | commitdiff |
2008-05-25 |
Michael Beck | - can now handle const-like Confirms |
tree | commitdiff |
2008-05-21 |
Michael Beck | typos fixed |
tree | commitdiff |
2008-05-17 |
Michael Beck | - fixed indentation |
tree | commitdiff |
2008-05-16 |
Michael Beck | small code improvements |
tree | commitdiff |
2008-05-16 |
Michael Beck | BugFix: invalidate more data |
tree | commitdiff |
2008-05-15 |
Michael Beck | Improved inline heuristic: |
tree | commitdiff |
2008-05-13 |
Michael Beck | BugFix:loop-info must run on copy |
tree | commitdiff |
2008-05-13 |
Michael Beck | inline_leave_functions(): uses new collect_calls2(... |
tree | commitdiff |
2008-05-09 |
Michael Beck | - give bonus to small functions and leaves |
tree | commitdiff |
2008-05-09 |
Michael Beck | BugFix: add missing assure_cf_loop() for copy graphs |
tree | commitdiff |
2008-05-09 |
Michael Beck | inline_functions now inlines calls inside loops first |
tree | commitdiff |
2008-05-05 |
Michael Beck | add a parameter to limit the maximum size of methods... |
tree | commitdiff |
2008-05-05 |
Michael Beck | fixed comment |
tree | commitdiff |
2008-04-18 |
Matthias Braun | follow_mem_chain_Store should respect alias relation... |
tree | commitdiff |
2008-04-16 |
Christoph Mallon | Fix typos in comments. |
tree | commitdiff |
2008-04-14 |
Michael Beck | - get_irg_initial_exec()/set_irg_initial_exec() added |
tree | commitdiff |
2008-04-11 |
Michael Beck | limited support for inlining K&R functions added |
tree | commitdiff |
2008-04-10 |
Michael Beck | disable inline of functions which call alloca(), causes... |
tree | commitdiff |
2008-04-10 |
Michael Beck | - inline_method() now sets current_ir_graph itself |
tree | commitdiff |
2008-04-10 |
Michael Beck | BugFix: - there is only ONE proj_m to replace |
tree | commitdiff |
2008-04-07 |
Michael Beck | issue a debug output when inlining |
tree | commitdiff |
2008-04-07 |
Michael Beck | - BugFix: SSA construction requires current block set ... |
tree | commitdiff |
2008-04-07 |
Michael Beck | fixed unsaved checkin ... |
tree | commitdiff |
2008-04-07 |
Michael Beck | - optimize Tuple nodes away before calling scalar repla... |
tree | commitdiff |
2008-04-07 |
Michael Beck | - uses new SSA constructor interface, shortening and... |
tree | commitdiff |
2008-04-05 |
Michael Beck | completely removed old verbosity based debug output |
tree | commitdiff |
2008-04-05 |
Michael Beck | switch scalar replacement to new style debug output |
tree | commitdiff |
2008-04-05 |
Michael Beck | after inlining, the loop-info is inconsistent |
tree | commitdiff |
2008-04-05 |
Michael Beck | - init function for inlining added |
tree | commitdiff |
2008-04-04 |
Michael Beck | Improved inliner: rerun scalar_replacement() if address... |
tree | commitdiff |
2008-04-04 |
Michael Beck | BugFix: store the weigths to the right location |
tree | commitdiff |
2008-04-04 |
Michael Beck | fixed obstack initialization, removed unused variable |
tree | commitdiff |
2008-04-04 |
Michael Beck | - only local visible methods get the big benefice |
tree | commitdiff |
2008-04-04 |
Michael Beck | calculate benefice on real callee |
tree | commitdiff |
2008-04-04 |
Michael Beck | no bonus for recursive inlining |
tree | commitdiff |
next |