Home My Page Projects Code Snippets Project Openings SML/NJ
Summary Activity Forums Tracker Lists Tasks Docs Surveys News SCM Files

SCM Repository

[smlnj] Diff of /sml/trunk/HISTORY
ViewVC logotype

Diff of /sml/trunk/HISTORY

Parent Directory Parent Directory | Revision Log Revision Log | View Patch Patch

revision 583, Thu Mar 23 21:52:30 2000 UTC revision 652, Tue Jun 6 02:14:56 2000 UTC
# Line 12  Line 12 
12  Tag: <post-commit CVS tag>  Tag: <post-commit CVS tag>
13  Description:  Description:
14  ----------------------------------------------------------------------  ----------------------------------------------------------------------
15    Name: Matthias Blume
16    Date: 2000/06/06 11:15:00 JST
17    Tag: blume-20000606-lazierpickle
18    Description:
19    
20    !!!! NEW BOOT FILES !!!!
21    
22    * The main purpose of this update is to make library pickles lazier in
23    order to reduce the initial space penalty for autoloading a library.
24    As a result, it is now possible to have $smlnj/compiler.cm
25    pre-registered.  This should take care of the many complaints or
26    inquiries about missing structure Compiler.  This required changes to
27    CM's internal data structures and small tweaks to some algorithms.
28    
29    As a neat additional effect, it is no longer necessary (for the sake
30    of lean heap image files) to distinguish between a "minimal" CM and a
31    "full" CM.  Now, there is only one CM (i.e., the "full" version:
32    $smlnj/cm.cm aka $smlnj/cm/full.cm), and it is always available at the
33    interactive top level. ($smlnj/cm/minimal.cm is gone.)
34    
35    To make the life of compiler-hackers easier, "makeml" now also
36    pre-registers $smlnj/cmb.cm (aka $smlnj/cmb/current.cm).  In other
37    words, after you bootstrap a new sml for the first time, you will not
38    have to autoload $smlnj/cmb.cm again afterwards.  (The first time
39    around you will still have to do it, though.)
40    
41    * A second change consists of major updates to the CM manual.  There
42    are now several appendices with summary information and also a full
43    specification of the CM description file syntax.
44    
45    * In directory src/system I added the script "allcross".  This script
46    invokes sml and cross-compiles the compiler for all supported
47    architectures.  (Useful when providing a new set of boot files.)
48    
49    * There seems to be a latent bug in my "lazy pickles" mechanism.  I
50    added a small tweak to pickle-util.sml to work around this problem,
51    but it is not a proper fix yet.  I will investigate further.  (The
52    effect of the bug was an inflation of library pickle size.)
53    
54    * Version number increased to 110.28.1 (to avoid compatibility problems).
55    
56    ----------------------------------------------------------------------
57    Name: Allen Leung
58    Date: 2000/05/25 17:28 EDT
59    Tag: leunga-20000525-ra
60    Description:
61    
62      Fixed a bug in freezing phase of the register allocator.
63    
64    ----------------------------------------------------------------------
65    Name: Allen Leung
66    Date: 2000/05/15 22:53 EDT
67    Tag: leunga-20000515-alpha-x86-ra
68    Description:
69    
70      1. Alpha
71    
72          Slight cleanup.  Removed the instruction SGNXL
73    
74      2. X86
75    
76          Added the following instructions to the instruction set:
77    
78            ROLx, RORx,
79            BTx, BTSx, BTLx, BTRx,
80            XCHGx, and variants with the LOCK prefix
81    
82      3. Register Allocation
83    
84          The module ra-rewrite-with-renaming has been improved.
85    
86      These have no effect on SML/NJ.
87    
88    ----------------------------------------------------------------------
89    Name: Matthias Blume
90    Date: 2000/05/15 16:20:00 JST
91    Tag: blume-20000515-lightrebuild
92    Description:
93    
94    1. I added an alternative to "-rebuild" to "makeml".  The difference is
95       that prior to calling CMB.make' the CM-variable "LIGHT" will be
96       defined.  In effect, the command will not build any cross-compiler
97       backends and therefore finish more quickly.
98    
99       The "fixpt" script also takes a "-light" switch to be able to use
100       this new facility while compiling for a fixpoint.
101    
102    2. I replaced all mentions of anchored paths in group owner specifications
103       with simple relative paths (usually starting with "..").
104       The rationale is that a library's internal workings should not be
105       compromised by the lack of some anchor.  (An anchor is necessary
106       for someone who wants to refer to the library by an anchored path,
107       but it should not be necessary to build the same library in the first
108       place.)
109    
110    3. I changed the way CM's tool mechanism determines the shell command
111       string used for things like ml-yacc etc. so that it does not break
112       when CM.Control.implicit_anchors is turned off.
113    
114    ----------------------------------------------------------------------
115    Name: Matthias Blume
116    Date: 2000/05/12 18:20:00 JST
117    Tag: blume-20000512-ml-build
118    Description:
119    
120    Fixed a bug in config/_ml-build that prevented ml-yacc and ml-lex from
121    getting installed properly (by config/install.sh).
122    
123    ----------------------------------------------------------------------
124    Name: Matthias Blume
125    Date: 2000/05/12 17:30:00 JST
126    Tag: blume-20000512-anchors
127    Description:
128    
129    !!! NEW BOOT FILES !!!
130    
131    This change is in preparation of fading out support for "implicitly
132    anchored path names".  I went through all sources and used the
133    explicit (and relatively new) $-notation.  See system/README and the
134    CM manual for more info on this.
135    
136    I also modified the anchoring scheme for some things such as "smlnj",
137    "MLRISC", "cm", etc. to take advantage of the fact that explicit
138    anchors are more expressive: anchor name and first arc do not have to
139    coincide.  This entails the following user-visible change:
140    
141    You have to write $smlnj/foo/bar instead of smlnj/foo/bar.  In
142    particular, when you fire up sml with a command-line argument, say,
143    e.g.:
144    
145       sml '$smlnj/cmb.cm'
146    
147    At the ML toplevel prompt:
148    
149       CM.autoload "$smlnj/cmb.cm";
150    
151    There is also a new controller in CM.Control that can be used to turn
152    off all remaining support for implicit anchors by saying:
153    
154        CM.autoload "$smlnj/
155        #set CM.Control.implicit_anchors false;
156    
157    This causes CM to reject implicitly anchored paths.  This is (for the
158    time being) less permissive than the "final" version where there will
159    be no more such implicit anchors and relative paths will be just that:
160    relative.
161    
162    The next step (version after next version?) will be to make the
163    default for CM.Control.implicit_anchors false.  After the dust has
164    settled, I can then produce the "final" version of this...
165    
166    Note: Since bootstrapping is a bit tricky, I provided new boot files.
167    
168    ----------------------------------------------------------------------
169    Name: Matthias Blume
170    Date: 2000/05/11 16:30:00 JST
171    Tag: blume-20000511-sources
172    Description:
173    
174    The main change is that I added function CM.sources as a generalized
175    version of the earlier CM.makedepend.  This entails the following
176    additional changes:
177    
178      - CM.makedepend has been dropped.
179    
180      - CM manual has been updated.
181    
182      - TOOLS signature and API have been changed.
183    
184    ----------------------------------------------------------------------
185    Name: Allen Leung
186    Date: 2000/05/10 21:17 EDT
187    Tag: leunga-20000510-moby-c--ssa
188    Description:
189    
190      Various bug fixes and new features for C--, Moby and MLRISC optimizations.
191    None of these affect SML/NJ.
192    
193    1. Register Allocation
194    
195        a. A new ra spilling module (ra/ra-spill-with-renaming) is implemented.
196           This module tries to remove local (i.e. basic block level) redundancies
197           during spilling.
198    
199        b. A new framework for performing region based register allocation.
200           Not yet entirely functional.
201    
202    2. X86
203    
204       a. DefUse for POP was missing the stack pointer [found by Lal]
205       b. Reload for CALL was incorrect in X86Spill [found by John]
206       c. Various fixes in X86Spill so that it can be used correctly for
207          the new spilling module.
208    
209    3. SSA/IR
210    
211       a. New module ir/dj-dataflow.sml implements elimination based
212          data flow analysis.
213    
214    4. MLRiscGen
215    
216       a. Fix for gc type annotation
217    
218    5. MDGen
219    
220       Various fixes for machine description -> ml code translation.  For ssa
221       only.
222    
223    ----------------------------------------------------------------------
224    Name: Allen Leung
225    Date: 2000/05/08 22:17 EDT
226    Tag: leunga-20000508-labexp
227    Description:
228    
229      Fermin has found a few assembly problems with constant expressions
230      generated in LabelExp.  Mostly, the problems involve extra parentheses,
231      which choke on dumb assemblers.  This is his fix.
232    
233    ----------------------------------------------------------------------
234    Name: Dave MacQueen
235    Date: 2000/04/09 14:00 EDT
236    Tag: dbm-20000502-Version_110_28
237    Description:
238    
239    1. Updated src/compiler/TopLevel/main/version.sml to version 110.28
240    
241    2. Updated config/version to 110.28
242    
243    3. Updated config/srcarchiveurl
244    
245    3. New boot files!
246       ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.28/
247    
248    ----------------------------------------------------------------------
249    Name: Matthias Blume
250    Date: 2000/05/01 19:05:00 JST
251    Tag: blume-20000501-noweb
252    Description:
253    
254    A new noweb tool has been added.  The existing system is entirely
255    unaffected by this, but some CM users have asked for renewed noweb
256    support.  Everything is documented in the CM manual.
257    
258    New (plugin) libraries:
259    
260       noweb-tool.cm
261       nw-ext.cm
262    
263    ----------------------------------------------------------------------
264    Name: Dave MacQueen
265    Date: 2000/04/30 12:40PM EDT
266    Tag: dbm-20000430-bug_fixes
267    Description:
268    
269    1. Fix for bug 1498
270       smlnj/src/system/Basis/Implementation/Unsafe/object.sig
271       smlnj/src/system/Basis/Implementation/Unsafe/object.sml
272         added toRealArray function
273       smlnj/src/compiler/MiscUtil/print/ppobj.sml
274         added check for tag Obj.RealArray to array printing case in ppObj
275    
276    2. Fix for bug 1510
277       smlnj/src/compiler/Semant/types/typesutil.sml
278         fixed definition of dummyargs (used by equalTycon) so that
279         dummy args are distinct types
280    
281    ----------------------------------------------------------------------
282    Name: Matthias Blume
283    Date: 2000/04/30 01:00:00 JST
284    Tag: blume-20000430-versions
285    Description:
286    
287    1. CM version numbering added.  This is an implementation of Lal's
288       proposal for adding version numbers and version checking to .cm
289       files.  Lal said that his proposal was just that -- a proposal.
290       For the time being I went ahead and implemented it so that people
291       can comment on it.  Everything is completely backward-compatible
292       (except for the stable library format, i.e., new bootfiles!).
293    
294       As usual, see the CM manual for details.
295    
296    2. An alternative syntax for anchored paths has been implemented.
297       Dave has recently voiced the same concerns that I had when I did
298       this, so there should be some support.  My take is that eventually
299       I will let support for the current syntax (where anchors are
300       "implicit") fade out in favor of the new, explicit syntax.
301       In order to be backward-compatible, both old and new syntax are
302       currently supported.
303    
304       Again, see the CM manual for details.
305    
306    3. Parallel make is trying to be slightly smarter:  When the master
307       process finds a "bottleneck", i.e., when there is only one
308       compilation unit that can be compiled and everybody else is
309       waiting on it, then it will simply compile it directly instead
310       of clumsily telling one of the slaves to do it.
311    
312    4. Support for "unsharing" added.  This is necessary in order to be
313       able to have two different versions of the same library running
314       at the same time (e.g., for trying out a new MLRISC while still
315       having the old MLRISC linked into the current compiler, etc.)
316       See the CM manual.
317    
318    5. Simple "makedepend" functionality added for generating Makefile
319       dependency information.  (This is rather crude at the moment.
320       Expect some changes here in the future.)
321    
322    6. ".fun" added as a recognized suffix for ML files. Also documented
323       explicitly in the manual that the fallback behavior (unknown suffix
324       -> ML file) is not an official feature!
325    
326    7. Small changes to the pickler for stable libraries.
327    
328    8. Several internal changes to CM (for cleanup/improvement).
329    
330    
331    !!!! NEW BINFILES !!!!
332    
333    ----------------------------------------------------------------------
334    Name: Matthias Blume
335    Date: 2000/04/28 17:30:00 JST
336    Tag: blume-20000428-pathconfig
337    Description:
338    
339    1. I changed config/install.sh to remove duplicate entries from the
340       lib/pathconfig file at the end.  Moreover, the final version of
341       lib/pathconfig is sorted alphabetically.  The same (sorting) is done
342       in src/system/installml.
343    
344    2. The config/install.sh script now consistently uses relative
345       pathnames in lib/pathconfig whenever the anchor is in the lib
346       directory.  (So far this was true for the libraries that come
347       pre-compiled and bundled as part of the bootfiles but not for
348       libraries that are compiled by the script itself.)
349    
350    ----------------------------------------------------------------------
351    Name: Matthias Blume
352    Date: 2000/04/26 13:10:00 JST
353    Tag: blume-20000426-fun_suffix
354    Description:
355    
356    Added ".fun" as a recognized file name suffix (for ML code).
357    
358    ----------------------------------------------------------------------
359    Name: Allen Leung
360    Date: 2000/04/25 17:00:00 EST
361    Tag: leunga-20000425-alpha-ra
362    Description:
363    
364    1. Alpha
365    
366        PSEUDOARITH was missing in AlphaRewrite.  This causes an endless loop
367    in C--.
368    
369    2. RA
370    
371       Added a flag "ra-dump-size" to print out the size of the flowgraph
372       and the interference graph.
373    
374    ----------------------------------------------------------------------
375    Name: Dave MacQueen
376    Date: 2000/04/25/
377    Tag: dbm-20000425-mlyacc_doc_examples
378    Description:
379      Updated mlyacc.tex sections 5 and 7 for SML '97 and CM.
380      Updated all three examples in src/ml-yacc/examples to run
381      under 110.* using CM.make.
382    
383    ----------------------------------------------------------------------
384    Name: Allen Leung
385    Date: 2000/04/20 23:04:00 EST
386    Tag: leunga-20000420-ssa-c---stuff
387    Description:
388    
389      This update synchronizes my repository with Yale's.  Most of these
390    changes, however, do not affect SML/NJ at all (the RA is an exception).
391    
392    1. Register Allocator
393    
394       a. An improvement in the interference graph construction:
395          Given a copy
396    
397                s <- t
398    
399          no interference edge between s and t is added for this definition of s.
400    
401       b. I've added two new spill heuristic modules that Fermin and I developed
402          (in the new library RA.cm). These are unused in SML/NJ but maybe
403          useful for others (Moby?)
404    
405    2. X86
406    
407       a. Various fixes in the backend provided by Fermin [C--] and Lal.
408    
409    3. Alpha
410    
411       a. Added the BSR instruction and code generation that goes with it [C--]
412       b. Other fixes too numerous to recount provided by Fermin [C--]
413    
414    4. Regmaps
415    
416       a. The regmaps are not initialized with the identity physical bindings
417          at creation time.  This is unneeded.
418    
419    5. MLRISC Optimizations
420    
421       a. The DJ-Graph module can now compute the iterated dominance frontiers
422          intersects with liveness incrementally in linear time! Woohoo!
423          This is now used in my new SSA construction algorithm.
424    
425       b. THe branch reorganization module is now smarter about linear chains of
426          basic blocks.
427    
428    
429    ----------------------------------------------------------------------
430    Name: Matthias Blume
431    Date: 2000/04/12 13:52:00 JST
432    Tag: blume_main_v110p27_1
433    Description:
434    
435    Changed install.sh script to handle archive files without version number
436    and to use "boot.<arch>-<os>" instead of "sml.boot.<arch>-<os>" for the
437    name of the boot file archive.
438    
439    ----------------------------------------------------------------------
440    Name: Dave MacQueen
441    Date: 2000/04/09 14:00 EDT
442    Tag: dbm-20000410-Version_110_27
443    Description:
444    
445    1. Updated src/compiler/TopLevel/main/version.sml to version 110.27
446    
447    2. Updated src/config/version to 110.27
448    
449    3. New boot files!
450    
451    ----------------------------------------------------------------------
452    Name: Allen Leung
453    Date: 2000/04/09 19:09:00 EST
454    Tag: leunga-20000409-misc
455    Description:
456    
457    1.  Yet another fix for x86 assembly for idivl, imull, mull and friends.
458    
459    2.  Miscellaneous improvements to MLRISC (unused in sml/nj)
460    
461    ----------------------------------------------------------------------
462    Name: Stefan
463    Date: 2000/04/07 10:00:00 EDT
464    Tag: monnier-20000406-branch-handling
465    Description:
466    
467    Improved handling of branches (mostly those generated from
468    polymorphic equality), removed switchoff and changed the
469    default optimization settings (more cpsopt and less flintopt).
470    
471    ----------------------------------------------------------------------
472    Name: Allen Leung
473    Date: 2000/04/06 01:30:00 EST
474    Tag: leunga-20000406-peephole-x86-SSA-2
475    Description:
476    
477       Forgot a few files.
478    
479    ----------------------------------------------------------------------
480    Name: Allen Leung
481    Date: 2000/04/06 00:36:00 EST
482    Tag: leunga-20000406-peephole-x86-SSA
483    Description:
484    
485    1.  New Peephole code
486    
487    2.  Minor improvement to X86 instruction selection
488    
489    3.  Various fixes to SSA and machine description -> code translator
490    
491    ----------------------------------------------------------------------
492    Name: Matthias Blume
493    Date: 2000/04/05 12:30:00 JST
494    Tag: blume_main_v110p26p2_3
495    Description:
496    
497    This update just merges three minor cosmetic updates to CM's sources
498    to get ready for the 110.27 code freeze on Friday.  No functionality
499    has changed.
500    
501    ----------------------------------------------------------------------
502    Name: Allen Leung
503    Date: 2000/04/04 19:39:00 EST
504    Tag: leunga-20000404-x86-asm
505    Description:
506    
507    1.  Fixed a problem in X86 assembly.
508    
509        Things like
510    
511           jmp %eax
512           jmp (%eax)
513    
514        should be output as
515    
516           jmp *%eax
517           jmp *(%eax)
518    
519    2.  Assembly output
520    
521          Added a new flag
522    
523              "asm-indent-copies" (default to false)
524    
525          When this flag is on, parallel copies will be indented an extra level.
526    
527    ----------------------------------------------------------------------
528    Name: Allen Leung
529    Date: 2000/04/04 03:18:00 EST
530    Tag: leunga-20000404-C--Moby
531    Description:
532    
533        All of these fixes are related to C--, Moby, and my own optimization
534        stuff; so they shouldn't affect SML/NJ.
535    
536    1.  X86
537    
538        Various fixes related floating point, and extensions.
539    
540    2.  Alpha
541    
542        Some extra patterns related to loads with signed/zero extension
543        provided by Fermin.
544    
545    3.  Assembly
546    
547        When generating assemby, resolve the value of client defined constants,
548        instead of generating symbolic values.  This is controlled by the
549        new flag "asm-resolve-constants", which is default to true.
550    
551    4.  Machine Descriptions
552    
553        a. The precedence parser was slightly broken when parsing infixr symbols.
554        b. The type generalizing code had the bound variables reversed, resulting
555           in a problem during arity raising.
556        c. Various fixes in machine descriptions.
557    
558    ----------------------------------------------------------------------
559    Name: Matthias Blume
560    Date: 2000/04/03 16:05:00 JST
561    Tag: blume_main_v110p26p2_2
562    Description:
563    
564    I eliminated coreEnv from compInfo.  Access to the "Core" structure is
565    now done via the ordinary static environment that is context to each
566    compilation unit.
567    
568    To this end, I arranged that instead of "structure Core" as "structure
569    _Core" is bound in the pervasive environment.  Core access is done via
570    _Core (which can never be accidentially rebound because _Core is not a
571    legal surface-syntax symbol).
572    
573    The current solution is much cleaner because the core environment is
574    now simply part of the pervasive environment which is part of every
575    compilation unit's context anyway.  In particular, this eliminates all
576    special-case handling that was necessary until now in order to deal
577    with dynamic and symbolic parts of the core environment.
578    
579    Remaining hackery (to bind the "magic" symbol _Core) is localized in the
580    compilation mananger's bootstrap compiler (actually: in the "init group"
581    handling).  See the comments in src/system/smlnj/init/init.cmi for
582    more details.
583    
584    I also tried to track down all mentions of "Core" (as string argument
585    to Symbol.strSymbol) in the compiler and replaced them with a
586    reference to the new CoreSym.coreSym.  Seems cleaner since the actual
587    name appears in one place only.
588    
589    Binfile and bootfile format have not changed, but the switchover from
590    the old "init.cmi" to the new one is a bit tricky, so I supplied new
591    bootfiles anyway.
592    
593    ----------------------------------------------------------------------
594    Name: Allen Leung
595    Date: 2000/04/02 21:17:00 EST
596    Tag: leunga-20000402-mltree
597    Description:
598    
599       1. Renamed the constructor CALL in MLTREE by popular demand.
600       2. Added a bunch of files from my repository.  These are currently
601          used by other non-SMLNJ backends.
602    
603    ----------------------------------------------------------------------
604    Name: Allen Leung
605    Date: 2000/03/31 21:15:00 EST
606    Tag: leunga-20000331-aliasing
607    Description:
608    
609    This update contains a rewritten (and hopefully more correct) module
610    for extracting aliasing information from CPS.
611    
612       To turn on this feature:
613    
614            Compiler.Control.CG.memDisambiguate := true
615    
616       To pretty print the region information with assembly
617    
618           Compiler.Control.MLRISC.getFlag "asm-show-region" := true;
619    
620       To control how many levels of aliasing information are printed, use:
621    
622           Compiler.Control.MLRISC.getInt "points-to-show-level" := n
623    
624       The default of n is 3.
625    
626    ----------------------------------------------------------------------
627    Name: David MacQueen
628    Date: 2000/03/31 11:15:00 EST
629    Tag: dbm-20000331-runtime_fix
630    Description:
631    
632    This update contains:
633    
634    1. runtime/c-lib/c-libraries.c
635       includes added in revision 1.2 caused compilation errors on hppa-hpux
636    
637    2. fix for bug 1556
638       system/Basis/Implementation/NJ/internal-signals.sml
639    
640    ----------------------------------------------------------------------
641    Name: Matthias Blume
642    Date: 2000/03/31 18:00:00 JST
643    Tag: blume_main_v110p26p2_1
644    Description:
645    
646    This update contains:
647    
648    1. A small change to CM's handling of stable libraries:
649       CM now maintains one "global" modmap that is used for all stable
650       libraries.  The use of such a global modmap maximizes sharing and
651       minimizes the need for re-traversing parts of environments during
652       modmap construction.  (However, this has minor impact since modmap
653       construction seems to account for just one percent or less of total
654       compile time.)
655    
656    2. I added a "genmap" phase to the statistics.  This is where I got the
657       "one percent" number (see above).
658    
659    3. CM's new tool parameter mechanism just became _even_ better. :)
660       - The parser understands named parameters and recursive options.
661       - The "make" and "shell" tools use these new features.
662         (This makes it a lot easier to cascade these tools.)
663       - There is a small syntax change: named parameters use a
664    
665           <name> : ( <option> ... )            or
666           <name> : <string>
667    
668         syntax.  Previously, named parameters were implemented in an
669         ad-hoc fashion by each tool individually (by parsing strings)
670         and had the form
671    
672           <name>=<string>
673    
674       See the CM manual for a full description of these issues.
675    
676    ----------------------------------------------------------------------
677    Name: Matthias Blume
678    Date: 2000/03/30 18:00:00 JST
679    Tag: blume_main_v110p26p2_0
680    Description:
681    
682    !!!!! WARNING !!!!!!
683    !!  New binfiles  !!
684    !!!!!!!!!!!!!!!!!!!!
685    
686    This update contains:
687    
688    1. Moderate changes to CM:
689    
690       - Changes to CM's tools mechanism.  In particular, it is now possible
691       to have tools that accept additional "command line" parameters
692       (specified in the .cm file at each instance where the tool's class is
693       used).
694    
695       This was done to accomodate the new "make" and "shell" tools which
696       facilitate fairly seemless hookup to portions of code managed using
697       Makefiles or Shell scripts.
698    
699       There are no classes "shared" or "private" anymore.  Instead, the
700       sharing annotation is now a parameter to the "sml" class.
701    
702       There is a bit of generic machinery for implementing one's own
703       tools that accept command-line parameters.  However, I am not yet fully
704       satisfied with that part, so expect changes here in the future.
705    
706       All existing tools are described in the CM manual.
707    
708       - Slightly better error handling.  (CM now surpresses many followup
709       error messages that tended to be more annoying than helpful.)
710    
711    2. Major changes to the compiler's static environment data structures.
712    
713       - no CMStaticEnv anymore.
714            - no CMEnv, no "BareEnvironment" (actually, _only_ BareEnvironment,
715              but it is called Environment), no conversions between different
716              kinds of static environments
717    
718       - There is still a notion of a "modmap", but such modmaps are generated
719         on demand at the time when they are needed.  This sounds slow, but I
720         sped up the code that generates modmaps enough for this not to lead to
721         a slowdown of the compiler (at least I didn't detect any).
722    
723       - To facilitate rapid modmap generation, static environments now
724         contain an (optional) "modtree" structure.  Modtree annotations are
725         constructed by the unpickler during unpickling.  (This means that
726         the elaborator does not have to worry about modtrees at all.)
727         Modtrees have the advantage that they are compositional in the same
728         way as the environment data structure itself is compositional.
729         As a result, modtrees never hang on to parts of an environment that
730         has already been rendered "stale" by filtering or rebinding.
731    
732       - I went through many, many trials and errors before arriving at the
733         current solution.  (The initial idea of "linkpaths" did not work.)
734         But the result of all this is that I have touched a lot of files that
735         depend on the "modules" and "types" data structures (most of the
736         elaborator). There were a lot of changes during my "linkpath" trials
737         that could have been reverted to their original state but weren't.
738         Please, don't be too harsh on me for messing with this code a bit more
739         than what was strictly necessary...  (I _did_ resist the tempation
740         of doing any "global reformatting" to avoid an untimely death at
741         Dave's hands. :)
742    
743       - One positive aspect of the previous point:  At least I made sure that
744         all files that I touched now compile without warnings (other than
745         "polyEqual").
746    
747       - compiler now tends to run "leaner" (i.e., ties up less memory in
748         redundant modmaps)
749    
750    ----------------------------------------------------------------------
751    Name: Allen Leung
752    Date: 2000/03/29 18:00:00
753    Tag: leunga-20000327-mlriscGen_hppa_alpha_x86
754    Boot files (optional): ftp://react-ilp.cs.nyu.edu/leunga/110.26.1-sml.boot.x86-unix-20000330.tar.gz
755    Description:
756    
757       This update contains *MAJOR* changes to the way code is generated from CPS
758    in the module mlriscGen, and in various backend modules.
759    
760    CHANGES
761    =======
762    
763    1. MLRiscGen: forward propagation fix.
764    
765       There was a bug in forward propagation introduced at about the same time
766       as the MLRISC x86 backend, which prohibits coalescing to be
767       performed effectively in loops.
768    
769       Effect: speed up of loops in RISC architectures.
770               By itself, this actually slowed down certain benchmarks on the x86.
771    
772    2. MLRiscGen:  forward propagating addresses from consing.
773    
774       I've changed the way consing code is generated.  Basically I separated
775       out the initialization part:
776    
777            store tag,   offset(allocptr)
778            store elem1, offset+4(allocptr)
779            store elem2, offset+8(allocptr)
780            ...
781            store elemn, offset+4n(allocptr)
782    
783       and the address computation part:
784    
785            celladdr <- offset+4+alloctpr
786    
787       and move the address computation part
788    
789       Effect:  register pressure is generally lower as a result.  This
790                makes compilation of certain expressions much faster, such as
791                long lists with non-trivial elements.
792    
793                 [(0,0), (0,0), .... (0,0)]
794    
795    3. MLRiscGen: base pointer elimination.
796    
797        As part of the linkage mechanism, we generate the sequence:
798    
799         L:  ...  <- start of the code fragment
800    
801         L1:
802             base pointer <- linkreg - L1 + L
803    
804         The base pointer was then used for computing relocatable addresses
805       in the code fragment.  Frequently (such as in lots of continuations)
806       this is not needed.  We now eliminate this sequence whenever possible.
807    
808         For compile time efficiency, I'm using a very stupid local heuristic.
809       But in general, this should be done as a control flow analysis.
810    
811       Effect:  Smaller code size.  Speed up of most programs.
812    
813    4. Hppa back end
814    
815         Long jumps in span dependence resolution used to depend on the existence
816      of the base pointer.
817    
818         A jump to a long label L was expanded into the following sequence:
819    
820          LDIL %hi(L-8192), %r29
821          LDO  %lo(L-8192)(%r29), %r29
822          ADD  %r29, baseptr, %r29
823          BV,n %r0(%r29)
824    
825         In the presence of change (3) above, this will not work.  I've changed
826       it so that the following sequence of instructions are generated, which
827       doesn't mention the base pointer at all:
828    
829             BL,n  L', %r29           /* branch and link, L' + 4 -> %r29 */
830        L':  ADDIL L-(L'+4), %r29     /* Compute address of L */
831             BV,n  %r0(%r29)          /* Jump */
832    
833    5. Alpha back end
834    
835          New alpha instructions LDB/LDW have been added, as per Fermin's
836       suggestions.   This is unrelated to all other changes.
837    
838    6. X86 back end
839    
840         I've changed andl to testl in the floating point test sequence
841         whenever appropriate.  The Intel optimization guide states that
842         testl is perferable to andl.
843    
844    7. RA (x86 only)
845    
846         I've improved the spill propagation algorithm, using an approximation
847       of maximal weighted independent sets.   This seems to be necessary to
848       alleviate the negative effect in light of the slow down in (1).
849    
850         I'll write down the algorithm one of these days.
851    
852    8. MLRiscGen: frequencies
853    
854         I've added an annotation that states that all call gc blocks have zero
855       execution frequencies.  This improves register allocation on the x86.
856    
857    BENCHMARKS
858    ==========
859    
860       I've only perform the comparison on 110.25.
861    
862       The platforms are:
863    
864        HPPA  A four processor HP machine (E9000) with 5G of memory.
865        X86   A 300Hhz Pentium II with 128M of memory, and
866        SPARC An Ultra sparc 2 with 512M of memory.
867    
868       I used the following parameters for the SML benchmarks:
869    
870                 @SMLalloc
871         HPPA    256k
872         SPARC   512k
873         X86     256k
874    
875    COMPILATION TIME
876    ----------------
877       Here are the numbers comparing the compilation times of the compilers.
878       I've only compared 110.25 compiling the new sources versus
879       a fixpoint version of the new compiler compiling the same.
880    
881                     110.25                                  New
882               Total  Time in RA  Spill+Reload   Total  Time In RA Spill+Reload
883         HPPA   627s    116s        2684+3584     599s    95s       1003+1879
884         SPARC  892s    173s        2891+3870     708s    116s      1004+1880
885         X86    999s    315s       94006+130691   987s    296s    108877+141957
886    
887                   110.25         New
888                Code Size      Code Size
889         HPPA   8596736         8561421
890         SPARC  8974299         8785143
891         X86    9029180         8716783
892    
893       So in summary, things are at least as good as before.   Dramatic
894       reduction in compilation is obtained on the Sparc; I can't explain it,
895       but it is reproducible.  Perhaps someone should try to reproduce this
896       on their own machines.
897    
898    SML BENCHMARKS
899    --------------
900    
901        On the average, all benchmarks perform at least as well as before.
902    
903          HPPA         Compilation Time     Spill+Reload      Run Time
904                     110.25  New            110.25    New   110.25  New
905    
906          barnesHut  3.158  3.015  4.75%    1+1       0+0   2.980  2.922   2.00%
907              boyer  6.152  5.708  7.77%    0+0       0+0   0.218  0.213   2.34%
908       count-graphs  1.168  1.120  4.32%    0+0       0+0  22.705 23.073  -1.60%
909                fft  0.877  0.792 10.74%    1+3       1+3   0.602  0.587   2.56%
910        knuthBendix  3.180  2.857 11.32%    0+0       0+0   0.675  0.662   2.02%
911             lexgen  6.190  5.290 17.01%    0+0       0+0   0.913  0.788  15.86%
912               life  0.803  0.703 14.22%   25+25      0+0   0.153  0.140   9.52%
913              logic  2.048  2.007  2.08%    6+6       1+1   4.133  4.008   3.12%
914         mandelbrot  0.077  0.080 -4.17%    0+0       0+0   0.765  0.712   7.49%
915             mlyacc 22.932 20.937  9.53%  154+181    32+57  0.468  0.430   8.91%
916            nucleic  5.183  5.060  2.44%    2+2       0+0   0.125  0.120   4.17%
917      ratio-regions  3.357  3.142  6.84%    0+0       0+0  116.225 113.173 2.70%
918                ray  1.283  1.290 -0.52%    0+0       0+0   2.887  2.855   1.11%
919             simple  6.307  6.032  4.56%   28+30      5+7   3.705  3.658   1.28%
920                tsp  0.888  0.862  3.09%    0+0       0+0   7.040  6.893   2.13%
921               vliw 24.378 23.455  3.94%  106+127    25+45  2.758  2.707   1.91%
922      --------------------------------------------------------------------------
923       Average                     6.12%                                   4.09%
924    
925          SPARC        Compilation Time     Spill+Reload      Run Time
926                     110.25  New            110.25    New   110.25  New
927    
928          barnesHut  3.778  3.592  5.20%    2+2       0+0   3.648  3.453    5.65%
929              boyer  6.632  6.110  8.54%    0+0       0+0   0.258  0.242    6.90%
930       count-graphs  1.435  1.325  8.30%    0+0       0+0  33.672 34.737   -3.07%
931                fft  0.980  0.940  4.26%    3+9       2+6   0.838  0.827    1.41%
932        knuthBendix  3.590  3.138 14.39%    0+0       0+0   0.962  0.967   -0.52%
933             lexgen  6.593  6.072  8.59%    1+1       0+0   1.077  1.078   -0.15%
934               life  0.972  0.868 11.90%   26+26      0+0   0.143  0.140    2.38%
935              logic  2.525  2.387  5.80%    7+7       1+1   5.625  5.158    9.05%
936         mandelbrot  0.090  0.093 -3.57%    0+0       0+0   0.855  0.728   17.39%
937             mlyacc 26.732 23.827 12.19%  162+189    32+57  0.550  0.560   -1.79%
938            nucleic  6.233  6.197  0.59%    3+3       0+0   0.163  0.173   -5.77%
939      ratio-regions  3.780  3.507  7.79%    0+0       0+0 133.993 131.035   2.26%
940                ray  1.595  1.550  2.90%    1+1       0+0   3.440  3.418    0.63%
941             simple  6.972  6.487  7.48%   29+32      5+7   3.523  3.525   -0.05%
942                tsp  1.115  1.063  4.86%    0+0       0+0   7.393  7.265    1.77%
943               vliw 27.765 24.818 11.87%  110+135    25+45  2.265  2.135    6.09%
944      ----------------------------------------------------------------------------
945       Average                     6.94%                                    2.64%
946    
947          X86          Compilation Time     Spill+Reload      Run Time
948                     110.25  New            110.25    New   110.25  New
949    
950          barnesHut  5.530  5.420  2.03%  593+893   597+915   3.532  3.440   2.66%
951              boyer  8.768  7.747 13.19%  493+199   301+289   0.327  0.297  10.11%
952       count-graphs  2.040  2.010  1.49%  298+394   315+457  26.578 28.660  -7.26%
953                fft  1.327  1.302  1.92%  112+209   115+210   1.055  0.962   9.71%
954        knuthBendix  5.218  5.475 -4.69%  451+598   510+650   0.928  0.932  -0.36%
955             lexgen  9.970  9.623  3.60% 1014+841  1157+885   0.947  0.928   1.97%
956               life  1.183  1.183  0.00%  162+182   145+148   0.127  0.103  22.58%
957              logic  3.285  3.512 -6.45%  514+684   591+836   5.682  5.577   1.88%
958         mandelbrot  0.147  0.143  2.33%   38+41     33+54    0.703  0.690   1.93%
959             mlyacc 35.457 32.763  8.22% 3496+4564 3611+4860  0.552  0.550   0.30%
960            nucleic  7.100  6.888  3.07%  239+168   201+158   0.175  0.173   0.96%
961      ratio-regions  6.388  6.843 -6.65% 1182+257   981+300  120.142 120.345 -0.17%
962                ray  2.332  2.338 -0.29%  346+398   402+494   3.593  3.540   1.51%
963             simple  9.912  9.903  0.08% 1475+941  1579+1168  3.057  3.178  -3.83%
964                tsp  1.623  1.532  5.98%  266+200   250+211   8.045  7.878   2.12%
965               vliw 33.947 35.470 -4.29% 2629+2774 2877+3171  2.072  1.890   9.61%
966      ----------------------------------------------------------------------------
967       Average                     1.22%                                     3.36%
968    
969    ----------------------------------------------------------------------
970  Name: Allen Leung  Name: Allen Leung
971  Date: 2000/03/23 16:25:00  Date: 2000/03/23 16:25:00
972  Tag: leunga-20000323-fix_x86_alpha  Tag: leunga-20000323-fix_x86_alpha

Legend:
Removed from v.583  
changed lines
  Added in v.652

root@smlnj-gforge.cs.uchicago.edu
ViewVC Help
Powered by ViewVC 1.0.0