Diff of /sml/trunk/HISTORY

Parent Directory | Revision Log | Patch

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

---

Colored Diff Long Colored Diff Unidiff Context Diff Side by Side

Legend: Removed from v.580   changed lines   Added in v.635

---

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