Commit Graph
59 Commits
Author SHA1 Message Date
Nigel Tao e2ce9f1b1d Add a writer1.is_marked method. 2017-11-09 10:38:24 +11:00
Nigel Tao 0bc6f792fb Remove some suspensions from decode_huffman_fast.
name                           old speed      new speed      delta

puffs_adler32_10k/clang        2.41GB/s ± 1%  2.39GB/s ± 2%  -0.88%  (p=0.022 n=9+10)
puffs_adler32_100k/clang       2.41GB/s ± 1%  2.39GB/s ± 2%    ~     (p=0.095 n=10+9)
puffs_flate_decode_1k/clang     136MB/s ± 2%   134MB/s ± 1%  -0.86%  (p=0.036 n=9+8)
puffs_flate_decode_10k/clang    172MB/s ± 0%   175MB/s ± 0%  +1.96%  (p=0.000 n=8+9)
puffs_flate_decode_100k/clang   206MB/s ± 0%   209MB/s ± 1%  +1.38%  (p=0.000 n=8+9)
puffs_zlib_decode_10k/clang     160MB/s ± 1%   163MB/s ± 2%  +1.73%  (p=0.001 n=9+9)
puffs_zlib_decode_100k/clang    189MB/s ± 1%   188MB/s ± 6%    ~     (p=0.143 n=10+10)

puffs_adler32_10k/gcc          3.27GB/s ± 0%  3.20GB/s ± 2%  -2.03%  (p=0.000 n=8+10)
puffs_adler32_100k/gcc         3.27GB/s ± 0%  3.21GB/s ± 1%  -1.89%  (p=0.000 n=8+10)
puffs_flate_decode_1k/gcc       141MB/s ± 1%   139MB/s ± 1%  -1.66%  (p=0.000 n=9+10)
puffs_flate_decode_10k/gcc      200MB/s ± 0%   209MB/s ± 1%  +4.73%  (p=0.000 n=10+10)
puffs_flate_decode_100k/gcc     242MB/s ± 0%   237MB/s ± 0%  -2.19%  (p=0.000 n=9+10)
puffs_zlib_decode_10k/gcc       188MB/s ± 0%   195MB/s ± 1%  +3.71%  (p=0.000 n=10+10)
puffs_zlib_decode_100k/gcc      225MB/s ± 0%   221MB/s ± 0%  -1.84%  (p=0.000 n=8+9)
2017-11-06 16:59:27 +11:00
Nigel Tao 058938006a Remove some suspensions from decode_huffman_fast.
name                           old speed      new speed      delta

puffs_adler32_10k/clang        2.45GB/s ± 0%  2.41GB/s ± 1%  -1.82%  (p=0.000 n=9+9)
puffs_adler32_100k/clang       2.45GB/s ± 0%  2.41GB/s ± 1%  -1.74%  (p=0.000 n=9+10)
puffs_flate_decode_1k/clang     133MB/s ± 0%   136MB/s ± 2%  +1.93%  (p=0.003 n=9+9)
puffs_flate_decode_10k/clang    181MB/s ± 0%   172MB/s ± 0%  -5.18%  (p=0.000 n=9+8)
puffs_flate_decode_100k/clang   204MB/s ± 0%   206MB/s ± 0%  +0.63%  (p=0.000 n=10+8)
puffs_zlib_decode_10k/clang     168MB/s ± 0%   160MB/s ± 1%  -4.82%  (p=0.000 n=10+9)
puffs_zlib_decode_100k/clang    189MB/s ± 0%   189MB/s ± 1%    ~     (p=0.143 n=10+10)

puffs_adler32_10k/gcc          3.26GB/s ± 0%  3.27GB/s ± 0%  +0.12%  (p=0.000 n=10+8)
puffs_adler32_100k/gcc         3.27GB/s ± 0%  3.27GB/s ± 0%  +0.18%  (p=0.000 n=10+8)
puffs_flate_decode_1k/gcc       130MB/s ± 0%   141MB/s ± 1%  +8.71%  (p=0.000 n=10+9)
puffs_flate_decode_10k/gcc      203MB/s ± 0%   200MB/s ± 0%  -1.72%  (p=0.000 n=9+10)
puffs_flate_decode_100k/gcc     241MB/s ± 0%   242MB/s ± 0%  +0.54%  (p=0.000 n=8+9)
puffs_zlib_decode_10k/gcc       188MB/s ± 0%   188MB/s ± 0%  +0.26%  (p=0.000 n=9+10)
puffs_zlib_decode_100k/gcc      223MB/s ± 0%   225MB/s ± 0%  +0.93%  (p=0.000 n=9+8)
2017-11-06 15:25:19 +11:00
Nigel Tao 13731f989f Add a reader1.unread_u8 method.
Also re-enable a number of std/flate tests and benchmarks.

It's not an optimization per se, but as a data point, here are some
benchmark numbers between 36a93fa and this commit. 36a93fa is "36a93fa
Add a TODO about impure functions", which is one commit before "f968950
Read bits up front in decode_huffman_fast", and f968950 was the commit
that made decode_huffman_fast read more than one byte at a time.

name                             old speed      new speed       delta

puffs_adler32_10k/clang          2.45GB/s ± 0%   2.45GB/s ± 0%     ~     (p=0.684 n=10+10)
puffs_adler32_100k/clang         2.45GB/s ± 0%   2.46GB/s ± 0%   +0.07%  (p=0.019 n=10+10)
puffs_flate_decode_1k/clang       124MB/s ± 0%    132MB/s ± 0%   +6.06%  (p=0.000 n=10+10)
puffs_flate_decode_10k/clang      145MB/s ± 0%    174MB/s ± 0%  +20.19%  (p=0.000 n=9+10)
puffs_flate_decode_100k/clang     168MB/s ± 0%    208MB/s ± 0%  +23.31%  (p=0.000 n=10+10)
puffs_zlib_decode_10k/clang       136MB/s ± 0%    162MB/s ± 0%  +18.94%  (p=0.000 n=10+9)
puffs_zlib_decode_100k/clang      158MB/s ± 0%    192MB/s ± 0%  +21.58%  (p=0.000 n=10+9)

puffs_adler32_10k/gcc            3.26GB/s ± 0%   3.26GB/s ± 0%     ~     (p=0.631 n=10+10)
puffs_adler32_100k/gcc           3.27GB/s ± 0%   3.27GB/s ± 0%     ~     (p=0.237 n=10+8)
puffs_flate_decode_1k/gcc        95.3MB/s ± 0%  105.5MB/s ± 0%  +10.67%  (p=0.000 n=10+10)
puffs_flate_decode_10k/gcc        110MB/s ± 0%    140MB/s ± 0%  +27.32%  (p=0.000 n=9+9)
puffs_flate_decode_100k/gcc       121MB/s ± 0%    150MB/s ± 0%  +24.02%  (p=0.000 n=8+10)
puffs_zlib_decode_10k/gcc         112MB/s ± 0%    120MB/s ± 0%   +6.96%  (p=0.000 n=10+9)
puffs_zlib_decode_100k/gcc        123MB/s ± 0%    134MB/s ± 0%   +8.93%  (p=0.000 n=9+8)
2017-11-05 15:59:02 +11:00
Nigel Tao 55e0bc513b Don't suspend for read_u8 if data is available.
It's not obvious why the benchmarks get worse when the code looks
better, but the long term goal is for all of decode_huffman_fast to not
suspend, and this is one step along that path.

name                           old speed      new speed      delta

puffs_adler32_10k/clang        2.40GB/s ± 1%  2.40GB/s ± 1%     ~     (p=0.841 n=5+5)
puffs_adler32_100k/clang       2.40GB/s ± 1%  2.39GB/s ± 2%     ~     (p=0.421 n=5+5)
puffs_flate_decode_1k/clang     131MB/s ± 1%   128MB/s ± 0%   -2.51%  (p=0.008 n=5+5)
puffs_flate_decode_10k/clang    171MB/s ± 1%   167MB/s ± 2%   -2.86%  (p=0.008 n=5+5)
puffs_flate_decode_100k/clang   202MB/s ± 0%   193MB/s ± 2%   -4.64%  (p=0.016 n=4+5)
puffs_zlib_decode_10k/clang     159MB/s ± 1%   157MB/s ± 0%   -1.65%  (p=0.008 n=5+5)
puffs_zlib_decode_100k/clang    185MB/s ± 1%   180MB/s ± 1%   -3.11%  (p=0.008 n=5+5)

puffs_adler32_10k/gcc          3.26GB/s ± 0%  3.26GB/s ± 0%     ~     (p=0.421 n=5+5)
puffs_adler32_100k/gcc         3.27GB/s ± 0%  3.26GB/s ± 0%     ~     (p=0.151 n=5+5)
puffs_flate_decode_1k/gcc       132MB/s ± 0%   129MB/s ± 0%   -2.29%  (p=0.008 n=5+5)
puffs_flate_decode_10k/gcc      170MB/s ± 0%   164MB/s ± 0%   -3.13%  (p=0.008 n=5+5)
puffs_flate_decode_100k/gcc     207MB/s ± 0%   185MB/s ± 0%  -10.45%  (p=0.016 n=5+4)
puffs_zlib_decode_10k/gcc       162MB/s ± 0%   157MB/s ± 0%   -2.95%  (p=0.008 n=5+5)
puffs_zlib_decode_100k/gcc      194MB/s ± 0%   175MB/s ± 0%   -9.78%  (p=0.016 n=5+4)
2017-11-05 08:03:21 +11:00
Nigel Tao aa1f0078db Add decode_huffman_fast.
It's not fast yet, and the immediate impact is indeed a significant
performance loss, but follow-up commits should make it faster.

name                           old speed      new speed      delta

puffs_adler32_10k/clang        2.41GB/s ± 0%  2.40GB/s ± 1%     ~     (p=0.222 n=5+5)
puffs_adler32_100k/clang       2.41GB/s ± 1%  2.39GB/s ± 1%   -0.97%  (p=0.016 n=5+5)
puffs_flate_decode_1k/clang     116MB/s ± 5%   115MB/s ± 0%     ~     (p=0.310 n=5+5)
puffs_flate_decode_10k/clang    145MB/s ± 0%   139MB/s ± 0%   -4.18%  (p=0.016 n=4+5)
puffs_flate_decode_100k/clang   164MB/s ± 5%   158MB/s ± 1%     ~     (p=0.151 n=5+5)
puffs_zlib_decode_10k/clang     135MB/s ± 2%   131MB/s ± 0%   -2.84%  (p=0.008 n=5+5)
puffs_zlib_decode_100k/clang    154MB/s ± 1%   148MB/s ± 2%   -4.09%  (p=0.016 n=4+5)

puffs_adler32_10k/gcc          3.13GB/s ± 4%  3.19GB/s ± 0%   +1.94%  (p=0.032 n=5+5)
puffs_adler32_100k/gcc         3.14GB/s ± 3%  3.20GB/s ± 3%     ~     (p=0.056 n=5+5)
puffs_flate_decode_1k/gcc       122MB/s ± 1%    94MB/s ± 1%  -22.84%  (p=0.008 n=5+5)
puffs_flate_decode_10k/gcc      143MB/s ± 0%   110MB/s ± 0%  -23.24%  (p=0.008 n=5+5)
puffs_flate_decode_100k/gcc     158MB/s ± 0%   118MB/s ± 0%  -24.98%  (p=0.008 n=5+5)
puffs_zlib_decode_10k/gcc       136MB/s ± 0%   119MB/s ± 1%  -12.63%  (p=0.008 n=5+5)
puffs_zlib_decode_100k/gcc      150MB/s ± 1%   124MB/s ± 2%  -17.17%  (p=0.008 n=5+5)
2017-11-04 14:45:11 +11:00
Nigel Tao 4810db3994 Rename decode_huffman to decode_huffman_slow. 2017-11-04 13:38:24 +11:00
Nigel Tao 558d8ddbbd Restrict tPrefix temporary variables's scope. 2017-11-03 14:00:35 +11:00
Nigel Tao acec19e969 Add genFilenameLineComments. 2017-11-03 12:28:15 +11:00
Nigel Tao e3822f9951 Re-introduce puffs_base__limit1.
It's not immediately obvious why this data structure performs differently, but
on net, the benchmark numbers seem better. It also won't require any
PUFFS_USE_NO_OP_PERFORMANCE_HACKS or "redundant but mysteriously performant"
use_limit fields.

name                             old speed      new speed      delta

puffs_adler32_10k/clang          2.46GB/s ± 0%  2.43GB/s ± 3%     ~     (p=0.513 n=5+10)
puffs_adler32_100k/clang         2.46GB/s ± 0%  2.44GB/s ± 2%     ~     (p=0.241 n=5+10)
puffs_flate_decode_1k/clang       122MB/s ± 1%   119MB/s ± 2%   -1.86%  (p=0.013 n=5+10)
puffs_flate_decode_10k/clang      147MB/s ± 0%   142MB/s ± 2%   -3.47%  (p=0.001 n=5+10)
puffs_flate_decode_100k/clang     161MB/s ± 0%   161MB/s ± 4%     ~     (p=0.635 n=4+10)
puffs_zlib_decode_10k/clang       139MB/s ± 0%   135MB/s ± 1%   -2.69%  (p=0.001 n=5+10)
puffs_zlib_decode_100k/clang      151MB/s ± 0%   152MB/s ± 3%     ~     (p=1.000 n=5+10)

puffs_adler32_10k/gcc            3.21GB/s ± 1%  3.18GB/s ± 2%     ~     (p=0.310 n=5+10)
puffs_adler32_100k/gcc           3.22GB/s ± 0%  3.19GB/s ± 2%     ~     (p=0.075 n=5+10)
puffs_flate_decode_1k/gcc         121MB/s ± 4%   126MB/s ± 1%   +4.54%  (p=0.002 n=5+8)
puffs_flate_decode_10k/gcc        144MB/s ± 1%   141MB/s ± 1%   -2.52%  (p=0.001 n=5+9)
puffs_flate_decode_100k/gcc       154MB/s ± 1%   158MB/s ± 0%   +2.39%  (p=0.001 n=5+9)
puffs_zlib_decode_10k/gcc         137MB/s ± 2%   135MB/s ± 1%     ~     (p=0.055 n=5+10)
puffs_zlib_decode_100k/gcc        146MB/s ± 2%   150MB/s ± 0%   +2.87%  (p=0.001 n=5+10)

puffs_gif_lzw_decode_20k/clang    132MB/s ± 0%   125MB/s ± 0%   -5.24%  (p=0.001 n=5+10)
puffs_gif_lzw_decode_100k/clang   176MB/s ± 0%   164MB/s ± 0%   -6.82%  (p=0.001 n=5+9)
puffs_gif_decode_1k/clang         363MB/s ± 0%   339MB/s ± 1%   -6.58%  (p=0.002 n=4+10)
puffs_gif_decode_10k/clang        138MB/s ± 0%   133MB/s ± 1%   -3.77%  (p=0.001 n=5+10)
puffs_gif_decode_100k/clang       125MB/s ± 0%   119MB/s ± 0%   -4.79%  (p=0.001 n=5+10)
puffs_gif_decode_1000k/clang      127MB/s ± 0%   121MB/s ± 0%   -5.04%  (p=0.001 n=5+10)

puffs_gif_lzw_decode_20k/gcc      141MB/s ± 0%   140MB/s ± 2%     ~     (p=1.000 n=5+10)
puffs_gif_lzw_decode_100k/gcc     182MB/s ± 0%   180MB/s ± 2%   -1.56%  (p=0.001 n=5+10)
puffs_gif_decode_1k/gcc           363MB/s ± 0%   409MB/s ± 3%  +12.71%  (p=0.001 n=5+10)
puffs_gif_decode_10k/gcc          127MB/s ± 0%   161MB/s ± 3%  +26.15%  (p=0.001 n=5+10)
puffs_gif_decode_100k/gcc         112MB/s ± 0%   136MB/s ± 2%  +21.95%  (p=0.001 n=5+10)
puffs_gif_decode_1000k/gcc        114MB/s ± 0%   140MB/s ± 2%  +22.78%  (p=0.001 n=5+10)
2017-11-03 10:09:57 +11:00
Nigel Tao 4b572702f9 Remove the return error "foo" special case. 2017-10-27 14:39:48 +11:00
Nigel Tao 9a8508857f Fix the "limitt" typo. 2017-10-27 11:10:25 +11:00
Nigel Tao cbc8c39375 Delete old limit- and mark-related code. 2017-10-27 10:57:32 +11:00
Nigel Tao 8ff453c752 Re-do how limits work. 2017-10-26 16:59:03 +11:00
Nigel Tao a9ba7e85ac Add reader1.limitt and writer1.limitt methods.
The "limitt" typo will be renamed to "limit" once we remove the keyword
named "limit".
2017-10-26 15:06:59 +11:00
Nigel Tao b886e3a1ea Add reader1.mark and reader1.since_mark methods. 2017-10-26 14:39:38 +11:00
Nigel Tao 5052b16f2d Rename writer1.slice to writer1.since_mark. 2017-10-26 13:03:19 +11:00
Nigel Tao 474a0d2ed7 Make mark a method, not a keyword. 2017-10-26 12:31:27 +11:00
Nigel Tao 36f5ae9bf4 Move the "mark in.dst" into the flate decode loop. 2017-10-26 11:20:42 +11:00
Nigel Tao 21bbbd422f Generate code for "mark in.dst".
The generated code is ugly and the language design is not finalized, but
the flate_backref_crosses_blocks test (testing a correctness bug) now
passes. We can fix the ugliness in subsequent commits.
2017-10-22 14:44:53 +11:00
Nigel Tao 83c8faef4d Reorder some switch cases. 2017-10-22 14:01:26 +11:00
Nigel Tao ac793d33bc Generate placeholder code for "mark in.dst". 2017-10-22 13:57:37 +11:00
Nigel Tao f29c4e278f Follow the pkg__type__method convention for base. 2017-10-21 15:59:08 +11:00
Nigel Tao 6815375de5 Generate double '_'s for pkg__type__method names. 2017-10-21 15:10:25 +11:00
Nigel Tao 2639f8f46a Rename std/flate's decoder to flate_decoder. 2017-10-21 13:05:19 +11:00
Nigel Tao 62935c41d7 Prepare for calculating the zlib checksum. 2017-10-21 11:55:58 +11:00
Nigel Tao 245e3ec7b9 Unroll iterate loops.
name                      old speed      new speed      delta
puffs_adler32_10k/clang   2.12GB/s ± 1%  2.42GB/s ± 1%  +14.24%  (p=0.008 n=5+5)
puffs_adler32_100k/clang  2.12GB/s ± 3%  2.43GB/s ± 0%  +14.29%  (p=0.008 n=5+5)
mimic_adler32_10k/clang   2.93GB/s ± 2%  2.96GB/s ± 0%     ~     (p=0.310 n=5+5)
mimic_adler32_100k/clang  2.89GB/s ± 2%  2.91GB/s ± 0%     ~     (p=1.000 n=5+5)
puffs_adler32_10k/gcc     2.17GB/s ± 1%  3.23GB/s ± 0%  +48.59%  (p=0.008 n=5+5)
puffs_adler32_100k/gcc    2.18GB/s ± 1%  3.23GB/s ± 0%  +48.34%  (p=0.008 n=5+5)
mimic_adler32_10k/gcc     2.97GB/s ± 0%  2.96GB/s ± 1%     ~     (p=1.000 n=5+5)
mimic_adler32_100k/gcc    2.91GB/s ± 1%  2.91GB/s ± 1%     ~     (p=0.548 n=5+5)
2017-10-20 17:28:21 +11:00
Nigel Tao b024cb1d8c Add some support for iterate assertions. 2017-10-20 16:17:20 +11:00
Nigel Tao 2cd6ebed8e Add a "deref" unary operator. 2017-10-20 15:44:30 +11:00
Nigel Tao 51bd52239b Add an "iterate" keyword. 2017-10-20 15:33:31 +11:00
Nigel Tao 50369c782c Add a skeleton of an adler.update method. 2017-10-19 18:58:17 +11:00
Nigel Tao c99fdaf63d Actually copy from the flate decoder's history. 2017-10-13 14:50:34 +11:00
Nigel Tao a33f744da5 Drop the "?" from "writer1.copy_from_slice?". 2017-10-13 11:44:53 +11:00
Nigel Tao 0384c5e4a8 Rename the copy_from_xxx methods for consistency. 2017-10-13 10:59:47 +11:00
Nigel Tao d4372cb5b8 Change license to Apache 2.
Also remove explicit PATENTS file. The new LICENSE file already
discusses patents.

Note that the Puffs project has not yet been publicly released.
2017-10-12 11:17:41 +11:00
Nigel Tao 5a7f9e6cde Remove some duplicated short_read_src code. 2017-10-07 15:36:16 +11:00
Nigel Tao d275b379a3 Remove the idea of limited reads and writes. 2017-10-07 15:32:00 +11:00
Nigel Tao a12292d094 Have "return ok" set coro_susp_point = 0. 2017-10-07 14:43:25 +11:00
Nigel Tao 585ea2024d Drop the "?" from "copy_from32?".
Also add a PUFFS_COROUTINE_SUSPENSION_POINT_MAYBE_SUSPEND macro.

Also add a loop to decoder.decode, as the code generator was previously
(incorrectly) not generating a coro_susp_point at the end of the
function.
2017-10-07 14:34:31 +11:00
Nigel Tao 2efdf5fab6 Add a lang/builtin package. 2017-10-07 11:37:10 +11:00
Nigel Tao 010c4db54f Allow explicit returns of built-in status codes. 2017-10-06 17:34:58 +11:00
Nigel Tao 2e883fcc23 Drop the "?" from "copy_history32?".
This makes the caller, not the implementation, responsible for detecting
short reads and writes. This (the lack of "?") is consistent with the
slice.copy_from method.

name                           old speed     new speed     delta
puffs_flate_decode_1k/clang    117MB/s ± 6%  120MB/s ± 0%   +2.31%  (p=0.016 n=5+5)
puffs_flate_decode_10k/clang   153MB/s ± 1%  155MB/s ± 0%   +1.16%  (p=0.008 n=5+5)
puffs_flate_decode_100k/clang  168MB/s ± 1%  177MB/s ± 0%   +5.41%  (p=0.008 n=5+5)
puffs_flate_decode_1k/gcc      127MB/s ± 1%  143MB/s ± 0%  +12.15%  (p=0.008 n=5+5)
puffs_flate_decode_10k/gcc     145MB/s ± 0%  180MB/s ± 1%  +24.00%  (p=0.008 n=5+5)
puffs_flate_decode_100k/gcc    163MB/s ± 0%  212MB/s ± 1%  +29.90%  (p=0.008 n=5+5)
2017-10-06 17:02:09 +11:00
Nigel Tao 96a9ab4126 Have copy_history32 compare to the wstart pointer. 2017-10-02 16:08:03 +11:00
Nigel Tao 6f132a74e2 Remove the 'mark' concept, no longer necessary. 2017-10-02 14:02:13 +11:00
Nigel Tao ead5e45ae7 Refactor cgen of expressions-as-statements. 2017-09-28 17:32:05 +10:00
Nigel Tao f1643be221 Have copy_from return the number of bytes copied. 2017-09-28 17:27:34 +10:00
Nigel Tao 7b6f7384d1 Take more care with a NULL reader1 or writer1 buf. 2017-09-23 15:32:42 +10:00
Nigel Tao fd5d44b3fc Add a mark method and type. 2017-09-23 13:17:37 +10:00
Nigel Tao a9f12c23a4 Generate C code for an_array[:]. 2017-09-23 10:51:04 +10:00
Nigel Tao f107ec155d Rename puffs_u16be to puffs_base_load_u16be. 2017-09-23 09:29:45 +10:00