Commit Graph

16933 Commits (66f500bdd75f96b56a0670ee91854d4aa766d95c)

Author SHA1 Message Date
Slava Pestov 555543faae compiler: tweak generated code 2009-09-04 03:01:18 -05:00
Slava Pestov 1c87486320 math.vectors.simd: allow punning SIMD vectors between types 2009-09-04 02:35:58 -05:00
Slava Pestov 136d6a42f3 math.vectors.simd: slightly faster 'sum' on 256-bit vectors: add the two components then do horizontal add, instead of doing a horizontal add on each one and adding the results 2009-09-04 02:23:25 -05:00
Slava Pestov 8223715a07 compiler.cfg.intrinsics: fix type detection on the alien type for vector accessors 2009-09-04 02:22:54 -05:00
Slava Pestov af14fd78dd math.vectors.simd: docs 2009-09-04 01:22:18 -05:00
Slava Pestov fc195f33f1 functors: support private words with DEFINES-PRIVATE; use this to make some words generated by math.vectors.simd.functor private 2009-09-04 01:21:59 -05:00
Joe Groff 57a40bc703 make struct-mirror an instance of assoc 2009-09-03 22:31:55 -05:00
Slava Pestov 698f32c4a1 math.vectors.simd: define fallbacks for all vector constructors so that code can still work even if SIMD is not available 2009-09-03 21:37:55 -05:00
Slava Pestov 1f5193198b compiler: clean up code generation for alien boxing/unboxing a bit 2009-09-03 21:22:43 -05:00
Joe Groff b1ba82c84f convert comparison branch code in compiler to use locals 2009-09-03 21:19:39 -05:00
Slava Pestov 214bf6372c benchmark.nbody-simd: new benchmark; a version of benchmark.nbody that uses math.vectors.simd:double-4 instead of double-arrays 2009-09-03 20:59:49 -05:00
Slava Pestov 20dfbf7ac8 More SIMD work
- Rename SIMD types and register representations: <type>-<count> rather than <count><type>-array
- Make a functor to define 256-bit vector types, use it to define float-8 type
- Make SIMD instructions pure-insns so that they participate in value numbering
2009-09-03 20:58:56 -05:00
Joe Groff 0da9e989f5 Merge branch 'master' of git://factorcode.org/git/factor 2009-09-03 20:32:27 -05:00
Joe Groff 0b9e5c034a add compiler comparison codes for floating-point unordered comparisons; update x86 backend to generate proper code for all floating-point comparisons 2009-09-03 20:32:05 -05:00
Joe Groff 6aef0184d9 add unit tests for comparisons against nan 2009-09-03 17:27:06 -05:00
Doug Coleman a9b6671de1 WHAMMY! convert io.sockets.unix 2009-09-03 17:08:54 -05:00
Doug Coleman 62261f18fe update network structs. NO WHAMMY, NO WHAMMY, NO WHAMMY...STOP!!! 2009-09-03 16:57:36 -05:00
Joe Groff 7a429a530e forget the old struct class when redefining a struct class so the old accessors get purged 2009-09-03 11:12:58 -05:00
Slava Pestov 80ed4bc918 Merge branch 'master' into simd 2009-09-03 03:45:58 -05:00
Slava Pestov 9cc705f6ba math.vectors.simd: split off intrinsics into a sub-vocabulary, to avoid loading most of the SIMD code on bootstrap 2009-09-03 03:43:43 -05:00
Slava Pestov f811208271 Detect SSE version and enable the correct set of SIMD intrinsics 2009-09-03 03:28:38 -05:00
Slava Pestov 11aadb74af compiler.tree.propagation: type function for clone had an issue, sometimes clone would get optimized out because of incorrect constant folding 2009-09-03 02:40:18 -05:00
Slava Pestov 52b99c050e Initial implementation of SSE vector intrinsics:
- cpu.architecture: add SSE vector representations
- compiler.cfg.intrinsics.alien: remove an attempt at optimization that value numbering handles now
- compiler.cfg.representations: support instructions where the representation is set in the 'rep' slot, and support conversions between single and double floats
- alien-float, set-alien-float now use the single float representation, and the conversion is implicit; this fixes a long-standing bug where a register could get clobbered because of how %set-alien-float was defined on x86
- math.vectors.specialization: add support for SIMD specialization (where the vector word's body is replaced by another quotation), also specialize the 'sum' word
- math.vectors.simd: 4float-array, 2double-array, 4double-array types, and specializers for the math.vectors words
2009-09-03 02:33:07 -05:00
Slava Pestov 87ea2dfdc5 benchmark.raytracer: replace hard-coded constant with self-documenting code 2009-09-03 02:24:32 -05:00
Slava Pestov 9d749cae67 specialized-arrays: fix clone method 2009-09-03 02:24:03 -05:00
Slava Pestov d47077ef2d struct-arrays: fix new-sequence method 2009-09-03 02:23:37 -05:00
Slava Pestov 9c3214ad98 math.constants: add single float epsilon value 2009-09-03 02:23:22 -05:00
Joe Groff 5d24e48f8c for typed words, put the specialized definition in a gensym, and check the input types and declare the output types in the inlined outer word so the checks can be cleared by the compiler when possible 2009-09-02 18:45:08 -05:00
Joe Groff 35f0c31916 Merge branch 'master' into strong-typing 2009-09-02 15:34:22 -05:00
Joe Groff e9a5ed5931 i suck at reading tech docs--those were m64 instructions, not mm instructions 2009-09-02 12:58:35 -05:00
Joe Groff 52540d67e1 vocab metadata for typed 2009-09-02 12:21:58 -05:00
Joe Groff 53b753bc3a Merge branch 'master' into strong-typing 2009-09-02 12:14:02 -05:00
Joe Groff 05ea0171d4 coerce and check output types on typed words; set "input-classes" and "default-output-classes" props on typed words 2009-09-02 12:13:47 -05:00
Joe Groff 355213e172 separate stack effect typing from hints. put it in a "typed" vocab, and have a TYPED: word that adds the type checking directly to the word 2009-09-02 11:45:30 -05:00
Joe Groff 68279e20bc Merge branch 'master' of git://factorcode.org/git/factor 2009-09-02 11:07:07 -05:00
Joe Groff 0ddf19d033 get rid of useless mm->xmm instructions in cpu.x86.assembler, add MOVHLPS and MOVLHPS 2009-09-02 11:06:08 -05:00
Slava Pestov d8be724092 Fix tree shaker again, gah 2009-09-02 07:28:40 -05:00
Slava Pestov dd1c5b5d68 Merge branch 'master' of git://factorcode.org/git/factor 2009-09-02 06:22:44 -05:00
Slava Pestov 775b9af2f7 compiler: eliminate boilerplate by centralizing info in declarative INSN: syntax 2009-09-02 06:22:37 -05:00
Slava Pestov 9f5577c368 classes.tuple: don't run out of memory inside 'instances' quotation if optimizing compiler is off 2009-09-02 05:19:20 -05:00
Samuel Tardieu 0c76719312 Make "divisors" work with 1 as well 2009-09-02 12:07:23 +02:00
Slava Pestov e140dd3fbb tools.deploy.shaker: fix snafu 2009-09-02 04:30:29 -05:00
Joe Groff 61196024e7 handle the stack effect type as a separate specialization pass, and use coercers when available 2009-09-01 23:13:08 -05:00
Joe Groff 468dcc8d60 Merge branch 'master' into strong-typing 2009-09-01 21:35:29 -05:00
Joe Groff 0c2f6fd63c let's not forget about hex literals 2009-09-01 21:28:23 -05:00
Joe Groff 4c9c1fd963 Merge branch 'master' into strong-typing 2009-09-01 21:18:50 -05:00
Joe Groff 64d2459dac require numeric literals to begin and end with a digit/decimal point so that stuff like "," and "1," don't parse as numbers 2009-09-01 21:14:26 -05:00
Doug Coleman cd90136425 add using to windows.com.wrapper 2009-09-01 18:16:57 -05:00
Doug Coleman 379b56bd9c Merge branch 'master' of git://factorcode.org/git/factor 2009-09-01 18:16:31 -05:00
Joe Groff 11aa46f929 typo in specialize-method 2009-09-01 15:59:59 -05:00