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 578, Tue Mar 14 05:16:29 2000 UTC revision 642, Thu May 11 07:30:29 2000 UTC
# Line 13  Line 13 
13  Description:  Description:
14  ----------------------------------------------------------------------  ----------------------------------------------------------------------
15  Name: Matthias Blume  Name: Matthias Blume
16    Date: 2000/05/11 16:30:00 JST
17    Tag: blume-20000511-sources
18    Description:
19    
20    The main change is that I added function CM.sources as a generalized
21    version of the earlier CM.makedepend.  This entails the following
22    additional changes:
23    
24      - CM.makedepend has been dropped.
25    
26      - CM manual has been updated.
27    
28      - TOOLS signature and API have been changed.
29    
30    ----------------------------------------------------------------------
31    Name: Allen Leung
32    Date: 2000/05/10 21:17 EDT
33    Tag: leunga-20000510-moby-c--ssa
34    Description:
35    
36      Various bug fixes and new features for C--, Moby and MLRISC optimizations.
37    None of these affect SML/NJ.
38    
39    1. Register Allocation
40    
41        a. A new ra spilling module (ra/ra-spill-with-renaming) is implemented.
42           This module tries to remove local (i.e. basic block level) redundancies
43           during spilling.
44    
45        b. A new framework for performing region based register allocation.
46           Not yet entirely functional.
47    
48    2. X86
49    
50       a. DefUse for POP was missing the stack pointer [found by Lal]
51       b. Reload for CALL was incorrect in X86Spill [found by John]
52       c. Various fixes in X86Spill so that it can be used correctly for
53          the new spilling module.
54    
55    3. SSA/IR
56    
57       a. New module ir/dj-dataflow.sml implements elimination based
58          data flow analysis.
59    
60    4. MLRiscGen
61    
62       a. Fix for gc type annotation
63    
64    5. MDGen
65    
66       Various fixes for machine description -> ml code translation.  For ssa
67       only.
68    
69    ----------------------------------------------------------------------
70    Name: Allen Leung
71    Date: 2000/05/08 22:17 EDT
72    Tag: leunga-20000508-labexp
73    Description:
74    
75      Fermin has found a few assembly problems with constant expressions
76      generated in LabelExp.  Mostly, the problems involve extra parentheses,
77      which choke on dumb assemblers.  This is his fix.
78    
79    ----------------------------------------------------------------------
80    Name: Dave MacQueen
81    Date: 2000/04/09 14:00 EDT
82    Tag: dbm-20000502-Version_110_28
83    Description:
84    
85    1. Updated src/compiler/TopLevel/main/version.sml to version 110.28
86    
87    2. Updated config/version to 110.28
88    
89    3. Updated config/srcarchiveurl
90    
91    3. New boot files!
92       ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.28/
93    
94    ----------------------------------------------------------------------
95    Name: Matthias Blume
96    Date: 2000/05/01 19:05:00 JST
97    Tag: blume-20000501-noweb
98    Description:
99    
100    A new noweb tool has been added.  The existing system is entirely
101    unaffected by this, but some CM users have asked for renewed noweb
102    support.  Everything is documented in the CM manual.
103    
104    New (plugin) libraries:
105    
106       noweb-tool.cm
107       nw-ext.cm
108    
109    ----------------------------------------------------------------------
110    Name: Dave MacQueen
111    Date: 2000/04/30 12:40PM EDT
112    Tag: dbm-20000430-bug_fixes
113    Description:
114    
115    1. Fix for bug 1498
116       smlnj/src/system/Basis/Implementation/Unsafe/object.sig
117       smlnj/src/system/Basis/Implementation/Unsafe/object.sml
118         added toRealArray function
119       smlnj/src/compiler/MiscUtil/print/ppobj.sml
120         added check for tag Obj.RealArray to array printing case in ppObj
121    
122    2. Fix for bug 1510
123       smlnj/src/compiler/Semant/types/typesutil.sml
124         fixed definition of dummyargs (used by equalTycon) so that
125         dummy args are distinct types
126    
127    ----------------------------------------------------------------------
128    Name: Matthias Blume
129    Date: 2000/04/30 01:00:00 JST
130    Tag: blume-20000430-versions
131    Description:
132    
133    1. CM version numbering added.  This is an implementation of Lal's
134       proposal for adding version numbers and version checking to .cm
135       files.  Lal said that his proposal was just that -- a proposal.
136       For the time being I went ahead and implemented it so that people
137       can comment on it.  Everything is completely backward-compatible
138       (except for the stable library format, i.e., new bootfiles!).
139    
140       As usual, see the CM manual for details.
141    
142    2. An alternative syntax for anchored paths has been implemented.
143       Dave has recently voiced the same concerns that I had when I did
144       this, so there should be some support.  My take is that eventually
145       I will let support for the current syntax (where anchors are
146       "implicit") fade out in favor of the new, explicit syntax.
147       In order to be backward-compatible, both old and new syntax are
148       currently supported.
149    
150       Again, see the CM manual for details.
151    
152    3. Parallel make is trying to be slightly smarter:  When the master
153       process finds a "bottleneck", i.e., when there is only one
154       compilation unit that can be compiled and everybody else is
155       waiting on it, then it will simply compile it directly instead
156       of clumsily telling one of the slaves to do it.
157    
158    4. Support for "unsharing" added.  This is necessary in order to be
159       able to have two different versions of the same library running
160       at the same time (e.g., for trying out a new MLRISC while still
161       having the old MLRISC linked into the current compiler, etc.)
162       See the CM manual.
163    
164    5. Simple "makedepend" functionality added for generating Makefile
165       dependency information.  (This is rather crude at the moment.
166       Expect some changes here in the future.)
167    
168    6. ".fun" added as a recognized suffix for ML files. Also documented
169       explicitly in the manual that the fallback behavior (unknown suffix
170       -> ML file) is not an official feature!
171    
172    7. Small changes to the pickler for stable libraries.
173    
174    8. Several internal changes to CM (for cleanup/improvement).
175    
176    
177    !!!! NEW BINFILES !!!!
178    
179    ----------------------------------------------------------------------
180    Name: Matthias Blume
181    Date: 2000/04/28 17:30:00 JST
182    Tag: blume-20000428-pathconfig
183    Description:
184    
185    1. I changed config/install.sh to remove duplicate entries from the
186       lib/pathconfig file at the end.  Moreover, the final version of
187       lib/pathconfig is sorted alphabetically.  The same (sorting) is done
188       in src/system/installml.
189    
190    2. The config/install.sh script now consistently uses relative
191       pathnames in lib/pathconfig whenever the anchor is in the lib
192       directory.  (So far this was true for the libraries that come
193       pre-compiled and bundled as part of the bootfiles but not for
194       libraries that are compiled by the script itself.)
195    
196    ----------------------------------------------------------------------
197    Name: Matthias Blume
198    Date: 2000/04/26 13:10:00 JST
199    Tag: blume-20000426-fun_suffix
200    Description:
201    
202    Added ".fun" as a recognized file name suffix (for ML code).
203    
204    ----------------------------------------------------------------------
205    Name: Allen Leung
206    Date: 2000/04/25 17:00:00 EST
207    Tag: leunga-20000425-alpha-ra
208    Description:
209    
210    1. Alpha
211    
212        PSEUDOARITH was missing in AlphaRewrite.  This causes an endless loop
213    in C--.
214    
215    2. RA
216    
217       Added a flag "ra-dump-size" to print out the size of the flowgraph
218       and the interference graph.
219    
220    ----------------------------------------------------------------------
221    Name: Dave MacQueen
222    Date: 2000/04/25/
223    Tag: dbm-20000425-mlyacc_doc_examples
224    Description:
225      Updated mlyacc.tex sections 5 and 7 for SML '97 and CM.
226      Updated all three examples in src/ml-yacc/examples to run
227      under 110.* using CM.make.
228    
229    ----------------------------------------------------------------------
230    Name: Allen Leung
231    Date: 2000/04/20 23:04:00 EST
232    Tag: leunga-20000420-ssa-c---stuff
233    Description:
234    
235      This update synchronizes my repository with Yale's.  Most of these
236    changes, however, do not affect SML/NJ at all (the RA is an exception).
237    
238    1. Register Allocator
239    
240       a. An improvement in the interference graph construction:
241          Given a copy
242    
243                s <- t
244    
245          no interference edge between s and t is added for this definition of s.
246    
247       b. I've added two new spill heuristic modules that Fermin and I developed
248          (in the new library RA.cm). These are unused in SML/NJ but maybe
249          useful for others (Moby?)
250    
251    2. X86
252    
253       a. Various fixes in the backend provided by Fermin [C--] and Lal.
254    
255    3. Alpha
256    
257       a. Added the BSR instruction and code generation that goes with it [C--]
258       b. Other fixes too numerous to recount provided by Fermin [C--]
259    
260    4. Regmaps
261    
262       a. The regmaps are not initialized with the identity physical bindings
263          at creation time.  This is unneeded.
264    
265    5. MLRISC Optimizations
266    
267       a. The DJ-Graph module can now compute the iterated dominance frontiers
268          intersects with liveness incrementally in linear time! Woohoo!
269          This is now used in my new SSA construction algorithm.
270    
271       b. THe branch reorganization module is now smarter about linear chains of
272          basic blocks.
273    
274    
275    ----------------------------------------------------------------------
276    Name: Matthias Blume
277    Date: 2000/04/12 13:52:00 JST
278    Tag: blume_main_v110p27_1
279    Description:
280    
281    Changed install.sh script to handle archive files without version number
282    and to use "boot.<arch>-<os>" instead of "sml.boot.<arch>-<os>" for the
283    name of the boot file archive.
284    
285    ----------------------------------------------------------------------
286    Name: Dave MacQueen
287    Date: 2000/04/09 14:00 EDT
288    Tag: dbm-20000410-Version_110_27
289    Description:
290    
291    1. Updated src/compiler/TopLevel/main/version.sml to version 110.27
292    
293    2. Updated src/config/version to 110.27
294    
295    3. New boot files!
296    
297    ----------------------------------------------------------------------
298    Name: Allen Leung
299    Date: 2000/04/09 19:09:00 EST
300    Tag: leunga-20000409-misc
301    Description:
302    
303    1.  Yet another fix for x86 assembly for idivl, imull, mull and friends.
304    
305    2.  Miscellaneous improvements to MLRISC (unused in sml/nj)
306    
307    ----------------------------------------------------------------------
308    Name: Stefan
309    Date: 2000/04/07 10:00:00 EDT
310    Tag: monnier-20000406-branch-handling
311    Description:
312    
313    Improved handling of branches (mostly those generated from
314    polymorphic equality), removed switchoff and changed the
315    default optimization settings (more cpsopt and less flintopt).
316    
317    ----------------------------------------------------------------------
318    Name: Allen Leung
319    Date: 2000/04/06 01:30:00 EST
320    Tag: leunga-20000406-peephole-x86-SSA-2
321    Description:
322    
323       Forgot a few files.
324    
325    ----------------------------------------------------------------------
326    Name: Allen Leung
327    Date: 2000/04/06 00:36:00 EST
328    Tag: leunga-20000406-peephole-x86-SSA
329    Description:
330    
331    1.  New Peephole code
332    
333    2.  Minor improvement to X86 instruction selection
334    
335    3.  Various fixes to SSA and machine description -> code translator
336    
337    ----------------------------------------------------------------------
338    Name: Matthias Blume
339    Date: 2000/04/05 12:30:00 JST
340    Tag: blume_main_v110p26p2_3
341    Description:
342    
343    This update just merges three minor cosmetic updates to CM's sources
344    to get ready for the 110.27 code freeze on Friday.  No functionality
345    has changed.
346    
347    ----------------------------------------------------------------------
348    Name: Allen Leung
349    Date: 2000/04/04 19:39:00 EST
350    Tag: leunga-20000404-x86-asm
351    Description:
352    
353    1.  Fixed a problem in X86 assembly.
354    
355        Things like
356    
357           jmp %eax
358           jmp (%eax)
359    
360        should be output as
361    
362           jmp *%eax
363           jmp *(%eax)
364    
365    2.  Assembly output
366    
367          Added a new flag
368    
369              "asm-indent-copies" (default to false)
370    
371          When this flag is on, parallel copies will be indented an extra level.
372    
373    ----------------------------------------------------------------------
374    Name: Allen Leung
375    Date: 2000/04/04 03:18:00 EST
376    Tag: leunga-20000404-C--Moby
377    Description:
378    
379        All of these fixes are related to C--, Moby, and my own optimization
380        stuff; so they shouldn't affect SML/NJ.
381    
382    1.  X86
383    
384        Various fixes related floating point, and extensions.
385    
386    2.  Alpha
387    
388        Some extra patterns related to loads with signed/zero extension
389        provided by Fermin.
390    
391    3.  Assembly
392    
393        When generating assemby, resolve the value of client defined constants,
394        instead of generating symbolic values.  This is controlled by the
395        new flag "asm-resolve-constants", which is default to true.
396    
397    4.  Machine Descriptions
398    
399        a. The precedence parser was slightly broken when parsing infixr symbols.
400        b. The type generalizing code had the bound variables reversed, resulting
401           in a problem during arity raising.
402        c. Various fixes in machine descriptions.
403    
404    ----------------------------------------------------------------------
405    Name: Matthias Blume
406    Date: 2000/04/03 16:05:00 JST
407    Tag: blume_main_v110p26p2_2
408    Description:
409    
410    I eliminated coreEnv from compInfo.  Access to the "Core" structure is
411    now done via the ordinary static environment that is context to each
412    compilation unit.
413    
414    To this end, I arranged that instead of "structure Core" as "structure
415    _Core" is bound in the pervasive environment.  Core access is done via
416    _Core (which can never be accidentially rebound because _Core is not a
417    legal surface-syntax symbol).
418    
419    The current solution is much cleaner because the core environment is
420    now simply part of the pervasive environment which is part of every
421    compilation unit's context anyway.  In particular, this eliminates all
422    special-case handling that was necessary until now in order to deal
423    with dynamic and symbolic parts of the core environment.
424    
425    Remaining hackery (to bind the "magic" symbol _Core) is localized in the
426    compilation mananger's bootstrap compiler (actually: in the "init group"
427    handling).  See the comments in src/system/smlnj/init/init.cmi for
428    more details.
429    
430    I also tried to track down all mentions of "Core" (as string argument
431    to Symbol.strSymbol) in the compiler and replaced them with a
432    reference to the new CoreSym.coreSym.  Seems cleaner since the actual
433    name appears in one place only.
434    
435    Binfile and bootfile format have not changed, but the switchover from
436    the old "init.cmi" to the new one is a bit tricky, so I supplied new
437    bootfiles anyway.
438    
439    ----------------------------------------------------------------------
440    Name: Allen Leung
441    Date: 2000/04/02 21:17:00 EST
442    Tag: leunga-20000402-mltree
443    Description:
444    
445       1. Renamed the constructor CALL in MLTREE by popular demand.
446       2. Added a bunch of files from my repository.  These are currently
447          used by other non-SMLNJ backends.
448    
449    ----------------------------------------------------------------------
450    Name: Allen Leung
451    Date: 2000/03/31 21:15:00 EST
452    Tag: leunga-20000331-aliasing
453    Description:
454    
455    This update contains a rewritten (and hopefully more correct) module
456    for extracting aliasing information from CPS.
457    
458       To turn on this feature:
459    
460            Compiler.Control.CG.memDisambiguate := true
461    
462       To pretty print the region information with assembly
463    
464           Compiler.Control.MLRISC.getFlag "asm-show-region" := true;
465    
466       To control how many levels of aliasing information are printed, use:
467    
468           Compiler.Control.MLRISC.getInt "points-to-show-level" := n
469    
470       The default of n is 3.
471    
472    ----------------------------------------------------------------------
473    Name: David MacQueen
474    Date: 2000/03/31 11:15:00 EST
475    Tag: dbm-20000331-runtime_fix
476    Description:
477    
478    This update contains:
479    
480    1. runtime/c-lib/c-libraries.c
481       includes added in revision 1.2 caused compilation errors on hppa-hpux
482    
483    2. fix for bug 1556
484       system/Basis/Implementation/NJ/internal-signals.sml
485    
486    ----------------------------------------------------------------------
487    Name: Matthias Blume
488    Date: 2000/03/31 18:00:00 JST
489    Tag: blume_main_v110p26p2_1
490    Description:
491    
492    This update contains:
493    
494    1. A small change to CM's handling of stable libraries:
495       CM now maintains one "global" modmap that is used for all stable
496       libraries.  The use of such a global modmap maximizes sharing and
497       minimizes the need for re-traversing parts of environments during
498       modmap construction.  (However, this has minor impact since modmap
499       construction seems to account for just one percent or less of total
500       compile time.)
501    
502    2. I added a "genmap" phase to the statistics.  This is where I got the
503       "one percent" number (see above).
504    
505    3. CM's new tool parameter mechanism just became _even_ better. :)
506       - The parser understands named parameters and recursive options.
507       - The "make" and "shell" tools use these new features.
508         (This makes it a lot easier to cascade these tools.)
509       - There is a small syntax change: named parameters use a
510    
511           <name> : ( <option> ... )            or
512           <name> : <string>
513    
514         syntax.  Previously, named parameters were implemented in an
515         ad-hoc fashion by each tool individually (by parsing strings)
516         and had the form
517    
518           <name>=<string>
519    
520       See the CM manual for a full description of these issues.
521    
522    ----------------------------------------------------------------------
523    Name: Matthias Blume
524    Date: 2000/03/30 18:00:00 JST
525    Tag: blume_main_v110p26p2_0
526    Description:
527    
528    !!!!! WARNING !!!!!!
529    !!  New binfiles  !!
530    !!!!!!!!!!!!!!!!!!!!
531    
532    This update contains:
533    
534    1. Moderate changes to CM:
535    
536       - Changes to CM's tools mechanism.  In particular, it is now possible
537       to have tools that accept additional "command line" parameters
538       (specified in the .cm file at each instance where the tool's class is
539       used).
540    
541       This was done to accomodate the new "make" and "shell" tools which
542       facilitate fairly seemless hookup to portions of code managed using
543       Makefiles or Shell scripts.
544    
545       There are no classes "shared" or "private" anymore.  Instead, the
546       sharing annotation is now a parameter to the "sml" class.
547    
548       There is a bit of generic machinery for implementing one's own
549       tools that accept command-line parameters.  However, I am not yet fully
550       satisfied with that part, so expect changes here in the future.
551    
552       All existing tools are described in the CM manual.
553    
554       - Slightly better error handling.  (CM now surpresses many followup
555       error messages that tended to be more annoying than helpful.)
556    
557    2. Major changes to the compiler's static environment data structures.
558    
559       - no CMStaticEnv anymore.
560            - no CMEnv, no "BareEnvironment" (actually, _only_ BareEnvironment,
561              but it is called Environment), no conversions between different
562              kinds of static environments
563    
564       - There is still a notion of a "modmap", but such modmaps are generated
565         on demand at the time when they are needed.  This sounds slow, but I
566         sped up the code that generates modmaps enough for this not to lead to
567         a slowdown of the compiler (at least I didn't detect any).
568    
569       - To facilitate rapid modmap generation, static environments now
570         contain an (optional) "modtree" structure.  Modtree annotations are
571         constructed by the unpickler during unpickling.  (This means that
572         the elaborator does not have to worry about modtrees at all.)
573         Modtrees have the advantage that they are compositional in the same
574         way as the environment data structure itself is compositional.
575         As a result, modtrees never hang on to parts of an environment that
576         has already been rendered "stale" by filtering or rebinding.
577    
578       - I went through many, many trials and errors before arriving at the
579         current solution.  (The initial idea of "linkpaths" did not work.)
580         But the result of all this is that I have touched a lot of files that
581         depend on the "modules" and "types" data structures (most of the
582         elaborator). There were a lot of changes during my "linkpath" trials
583         that could have been reverted to their original state but weren't.
584         Please, don't be too harsh on me for messing with this code a bit more
585         than what was strictly necessary...  (I _did_ resist the tempation
586         of doing any "global reformatting" to avoid an untimely death at
587         Dave's hands. :)
588    
589       - One positive aspect of the previous point:  At least I made sure that
590         all files that I touched now compile without warnings (other than
591         "polyEqual").
592    
593       - compiler now tends to run "leaner" (i.e., ties up less memory in
594         redundant modmaps)
595    
596    ----------------------------------------------------------------------
597    Name: Allen Leung
598    Date: 2000/03/29 18:00:00
599    Tag: leunga-20000327-mlriscGen_hppa_alpha_x86
600    Boot files (optional): ftp://react-ilp.cs.nyu.edu/leunga/110.26.1-sml.boot.x86-unix-20000330.tar.gz
601    Description:
602    
603       This update contains *MAJOR* changes to the way code is generated from CPS
604    in the module mlriscGen, and in various backend modules.
605    
606    CHANGES
607    =======
608    
609    1. MLRiscGen: forward propagation fix.
610    
611       There was a bug in forward propagation introduced at about the same time
612       as the MLRISC x86 backend, which prohibits coalescing to be
613       performed effectively in loops.
614    
615       Effect: speed up of loops in RISC architectures.
616               By itself, this actually slowed down certain benchmarks on the x86.
617    
618    2. MLRiscGen:  forward propagating addresses from consing.
619    
620       I've changed the way consing code is generated.  Basically I separated
621       out the initialization part:
622    
623            store tag,   offset(allocptr)
624            store elem1, offset+4(allocptr)
625            store elem2, offset+8(allocptr)
626            ...
627            store elemn, offset+4n(allocptr)
628    
629       and the address computation part:
630    
631            celladdr <- offset+4+alloctpr
632    
633       and move the address computation part
634    
635       Effect:  register pressure is generally lower as a result.  This
636                makes compilation of certain expressions much faster, such as
637                long lists with non-trivial elements.
638    
639                 [(0,0), (0,0), .... (0,0)]
640    
641    3. MLRiscGen: base pointer elimination.
642    
643        As part of the linkage mechanism, we generate the sequence:
644    
645         L:  ...  <- start of the code fragment
646    
647         L1:
648             base pointer <- linkreg - L1 + L
649    
650         The base pointer was then used for computing relocatable addresses
651       in the code fragment.  Frequently (such as in lots of continuations)
652       this is not needed.  We now eliminate this sequence whenever possible.
653    
654         For compile time efficiency, I'm using a very stupid local heuristic.
655       But in general, this should be done as a control flow analysis.
656    
657       Effect:  Smaller code size.  Speed up of most programs.
658    
659    4. Hppa back end
660    
661         Long jumps in span dependence resolution used to depend on the existence
662      of the base pointer.
663    
664         A jump to a long label L was expanded into the following sequence:
665    
666          LDIL %hi(L-8192), %r29
667          LDO  %lo(L-8192)(%r29), %r29
668          ADD  %r29, baseptr, %r29
669          BV,n %r0(%r29)
670    
671         In the presence of change (3) above, this will not work.  I've changed
672       it so that the following sequence of instructions are generated, which
673       doesn't mention the base pointer at all:
674    
675             BL,n  L', %r29           /* branch and link, L' + 4 -> %r29 */
676        L':  ADDIL L-(L'+4), %r29     /* Compute address of L */
677             BV,n  %r0(%r29)          /* Jump */
678    
679    5. Alpha back end
680    
681          New alpha instructions LDB/LDW have been added, as per Fermin's
682       suggestions.   This is unrelated to all other changes.
683    
684    6. X86 back end
685    
686         I've changed andl to testl in the floating point test sequence
687         whenever appropriate.  The Intel optimization guide states that
688         testl is perferable to andl.
689    
690    7. RA (x86 only)
691    
692         I've improved the spill propagation algorithm, using an approximation
693       of maximal weighted independent sets.   This seems to be necessary to
694       alleviate the negative effect in light of the slow down in (1).
695    
696         I'll write down the algorithm one of these days.
697    
698    8. MLRiscGen: frequencies
699    
700         I've added an annotation that states that all call gc blocks have zero
701       execution frequencies.  This improves register allocation on the x86.
702    
703    BENCHMARKS
704    ==========
705    
706       I've only perform the comparison on 110.25.
707    
708       The platforms are:
709    
710        HPPA  A four processor HP machine (E9000) with 5G of memory.
711        X86   A 300Hhz Pentium II with 128M of memory, and
712        SPARC An Ultra sparc 2 with 512M of memory.
713    
714       I used the following parameters for the SML benchmarks:
715    
716                 @SMLalloc
717         HPPA    256k
718         SPARC   512k
719         X86     256k
720    
721    COMPILATION TIME
722    ----------------
723       Here are the numbers comparing the compilation times of the compilers.
724       I've only compared 110.25 compiling the new sources versus
725       a fixpoint version of the new compiler compiling the same.
726    
727                     110.25                                  New
728               Total  Time in RA  Spill+Reload   Total  Time In RA Spill+Reload
729         HPPA   627s    116s        2684+3584     599s    95s       1003+1879
730         SPARC  892s    173s        2891+3870     708s    116s      1004+1880
731         X86    999s    315s       94006+130691   987s    296s    108877+141957
732    
733                   110.25         New
734                Code Size      Code Size
735         HPPA   8596736         8561421
736         SPARC  8974299         8785143
737         X86    9029180         8716783
738    
739       So in summary, things are at least as good as before.   Dramatic
740       reduction in compilation is obtained on the Sparc; I can't explain it,
741       but it is reproducible.  Perhaps someone should try to reproduce this
742       on their own machines.
743    
744    SML BENCHMARKS
745    --------------
746    
747        On the average, all benchmarks perform at least as well as before.
748    
749          HPPA         Compilation Time     Spill+Reload      Run Time
750                     110.25  New            110.25    New   110.25  New
751    
752          barnesHut  3.158  3.015  4.75%    1+1       0+0   2.980  2.922   2.00%
753              boyer  6.152  5.708  7.77%    0+0       0+0   0.218  0.213   2.34%
754       count-graphs  1.168  1.120  4.32%    0+0       0+0  22.705 23.073  -1.60%
755                fft  0.877  0.792 10.74%    1+3       1+3   0.602  0.587   2.56%
756        knuthBendix  3.180  2.857 11.32%    0+0       0+0   0.675  0.662   2.02%
757             lexgen  6.190  5.290 17.01%    0+0       0+0   0.913  0.788  15.86%
758               life  0.803  0.703 14.22%   25+25      0+0   0.153  0.140   9.52%
759              logic  2.048  2.007  2.08%    6+6       1+1   4.133  4.008   3.12%
760         mandelbrot  0.077  0.080 -4.17%    0+0       0+0   0.765  0.712   7.49%
761             mlyacc 22.932 20.937  9.53%  154+181    32+57  0.468  0.430   8.91%
762            nucleic  5.183  5.060  2.44%    2+2       0+0   0.125  0.120   4.17%
763      ratio-regions  3.357  3.142  6.84%    0+0       0+0  116.225 113.173 2.70%
764                ray  1.283  1.290 -0.52%    0+0       0+0   2.887  2.855   1.11%
765             simple  6.307  6.032  4.56%   28+30      5+7   3.705  3.658   1.28%
766                tsp  0.888  0.862  3.09%    0+0       0+0   7.040  6.893   2.13%
767               vliw 24.378 23.455  3.94%  106+127    25+45  2.758  2.707   1.91%
768      --------------------------------------------------------------------------
769       Average                     6.12%                                   4.09%
770    
771          SPARC        Compilation Time     Spill+Reload      Run Time
772                     110.25  New            110.25    New   110.25  New
773    
774          barnesHut  3.778  3.592  5.20%    2+2       0+0   3.648  3.453    5.65%
775              boyer  6.632  6.110  8.54%    0+0       0+0   0.258  0.242    6.90%
776       count-graphs  1.435  1.325  8.30%    0+0       0+0  33.672 34.737   -3.07%
777                fft  0.980  0.940  4.26%    3+9       2+6   0.838  0.827    1.41%
778        knuthBendix  3.590  3.138 14.39%    0+0       0+0   0.962  0.967   -0.52%
779             lexgen  6.593  6.072  8.59%    1+1       0+0   1.077  1.078   -0.15%
780               life  0.972  0.868 11.90%   26+26      0+0   0.143  0.140    2.38%
781              logic  2.525  2.387  5.80%    7+7       1+1   5.625  5.158    9.05%
782         mandelbrot  0.090  0.093 -3.57%    0+0       0+0   0.855  0.728   17.39%
783             mlyacc 26.732 23.827 12.19%  162+189    32+57  0.550  0.560   -1.79%
784            nucleic  6.233  6.197  0.59%    3+3       0+0   0.163  0.173   -5.77%
785      ratio-regions  3.780  3.507  7.79%    0+0       0+0 133.993 131.035   2.26%
786                ray  1.595  1.550  2.90%    1+1       0+0   3.440  3.418    0.63%
787             simple  6.972  6.487  7.48%   29+32      5+7   3.523  3.525   -0.05%
788                tsp  1.115  1.063  4.86%    0+0       0+0   7.393  7.265    1.77%
789               vliw 27.765 24.818 11.87%  110+135    25+45  2.265  2.135    6.09%
790      ----------------------------------------------------------------------------
791       Average                     6.94%                                    2.64%
792    
793          X86          Compilation Time     Spill+Reload      Run Time
794                     110.25  New            110.25    New   110.25  New
795    
796          barnesHut  5.530  5.420  2.03%  593+893   597+915   3.532  3.440   2.66%
797              boyer  8.768  7.747 13.19%  493+199   301+289   0.327  0.297  10.11%
798       count-graphs  2.040  2.010  1.49%  298+394   315+457  26.578 28.660  -7.26%
799                fft  1.327  1.302  1.92%  112+209   115+210   1.055  0.962   9.71%
800        knuthBendix  5.218  5.475 -4.69%  451+598   510+650   0.928  0.932  -0.36%
801             lexgen  9.970  9.623  3.60% 1014+841  1157+885   0.947  0.928   1.97%
802               life  1.183  1.183  0.00%  162+182   145+148   0.127  0.103  22.58%
803              logic  3.285  3.512 -6.45%  514+684   591+836   5.682  5.577   1.88%
804         mandelbrot  0.147  0.143  2.33%   38+41     33+54    0.703  0.690   1.93%
805             mlyacc 35.457 32.763  8.22% 3496+4564 3611+4860  0.552  0.550   0.30%
806            nucleic  7.100  6.888  3.07%  239+168   201+158   0.175  0.173   0.96%
807      ratio-regions  6.388  6.843 -6.65% 1182+257   981+300  120.142 120.345 -0.17%
808                ray  2.332  2.338 -0.29%  346+398   402+494   3.593  3.540   1.51%
809             simple  9.912  9.903  0.08% 1475+941  1579+1168  3.057  3.178  -3.83%
810                tsp  1.623  1.532  5.98%  266+200   250+211   8.045  7.878   2.12%
811               vliw 33.947 35.470 -4.29% 2629+2774 2877+3171  2.072  1.890   9.61%
812      ----------------------------------------------------------------------------
813       Average                     1.22%                                     3.36%
814    
815    ----------------------------------------------------------------------
816    Name: Allen Leung
817    Date: 2000/03/23 16:25:00
818    Tag: leunga-20000323-fix_x86_alpha
819    Description:
820    
821    1. X86 fixes/changes
822    
823       a.  The old code generated for SETcc was completely wrong.
824           The Intel optimization guide is VERY misleading.
825    
826    2. ALPHA fixes/changes
827    
828       a.  Added the instructions LDBU, LDWU, STB, STW as per Fermin's suggestion.
829       b.  Added a new mode byteWordLoadStores to the functor parameter to Alpha()
830       c.  Added reassociation code for address computation.
831    
832    ----------------------------------------------------------------------
833    Name: Allen Leung
834    Date: 2000/03/22 01:23:00
835    Tag: leunga-20000322-fix_x86_hppa_ra
836    Description:
837    
838    1. X86 fixes/changes
839    
840       a.  x86Rewrite bug with MUL3 (found by Lal)
841       b.  Added the instructions FSTS, FSTL
842    
843    2. PA-RISC fixes/changes
844    
845       a.  B label should not be a delay slot candidate!  Why did this work?
846       b.  ADDT(32, REG(32, r), LI n) now generates one instruction instead of two,
847           as it should be.
848       c.  The assembly syntax for fstds and fstdd was wrong.
849       d.  Added the composite instruction COMICLR/LDO, which is the immediate
850           operand variant of COMCLR/LDO.
851    
852    3. Generic MLRISC
853    
854       a.  shuffle.sml rewritten to be slightly more efficient
855       b.  DIV bug in mltree-simplify fixed (found by Fermin)
856    
857    4. Register Allocator
858    
859       a.  I now release the interference graph earlier during spilling.
860           May improve memory usage.
861    
862    ----------------------------------------------------------------------
863    Name: Matthias Blume
864  Date: 2000/03/14 14:15:32  Date: 2000/03/14 14:15:32
865  Tag: blume_main_v110p26p1_2  Tag: blume_main_v110p26p1_2
866  Description:  Description:

Legend:
Removed from v.578  
changed lines
  Added in v.642

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