5248b2567b
[ot-font/h-advance] Adjust varStore cache condition
...
This gives best performance for short strings, now that we have a h-advance cache as well.
The en-words benchmark in particular, now ot-font is faster than ft.
Second to last line is of interest:
Before:
-----------------------------------------------------------------------------------------------------
Benchmark Time CPU Iterations
-----------------------------------------------------------------------------------------------------
BM_Shape/en-words.txt/Roboto-Regular.ttf/hb 29.8 ms 29.8 ms 23
BM_Shape/en-words.txt/Roboto-Regular.ttf/ft 30.4 ms 30.4 ms 23
BM_Shape/en-words.txt/SourceSerifVariable-Roman.ttf/hb 16.3 ms 16.3 ms 43
BM_Shape/en-words.txt/SourceSerifVariable-Roman.ttf/ft 16.5 ms 16.5 ms 42
BM_Shape/en-words.txt/SourceSerifVariable-Roman.ttf/var/hb 18.0 ms 18.0 ms 39
BM_Shape/en-words.txt/SourceSerifVariable-Roman.ttf/var/ft 17.8 ms 17.8 ms 39
After:
behdad@Behdads-MacBook-Pro harfbuzz % ninja -Cbuild && build/perf/benchmark-shape --benchmark_filter=en-words
-----------------------------------------------------------------------------------------------------
Benchmark Time CPU Iterations
-----------------------------------------------------------------------------------------------------
BM_Shape/en-words.txt/Roboto-Regular.ttf/hb 30.0 ms 30.0 ms 23
BM_Shape/en-words.txt/Roboto-Regular.ttf/ft 30.3 ms 30.3 ms 23
BM_Shape/en-words.txt/SourceSerifVariable-Roman.ttf/hb 16.3 ms 16.3 ms 43
BM_Shape/en-words.txt/SourceSerifVariable-Roman.ttf/ft 16.4 ms 16.4 ms 43
BM_Shape/en-words.txt/SourceSerifVariable-Roman.ttf/var/hb 17.6 ms 17.6 ms 40
BM_Shape/en-words.txt/SourceSerifVariable-Roman.ttf/var/ft 17.8 ms 17.8 ms 39
2022-05-26 05:02:50 -06:00
12fff976b6
[ot-var] Use atomic int for cached-serial
2022-05-26 05:02:50 -06:00
0919eaa6e8
[ot-font] Remove lock around cache
...
Not needed.
2022-05-26 05:02:50 -06:00
3548b6025f
[ot-font] Cache h-advances for variable fonts
2022-05-26 05:02:50 -06:00
39a07bf3eb
[ot-font] Rename cache to varStore_cache
2022-05-26 05:02:50 -06:00
970e03ecae
[ot-font] Add a hb_ot_font_t struct
2022-05-26 05:02:50 -06:00
80c49933c6
[hb-ft] Adjust serial signature
2022-05-26 05:02:50 -06:00
56e0ff9ea1
[ft] If hb_font changed, update FT_Face
...
Fixes https://github.com/harfbuzz/harfbuzz/issues/2270
Rather untested.
2022-05-26 05:02:50 -06:00
d0de389de8
[font] Fix test
2022-05-26 05:02:50 -06:00
a2015cd300
[font] Add a separate serial_coords
2022-05-26 05:02:50 -06:00
8629df188a
[ft] Discard advance cache if font changed
...
Uses newly added font serial API.
Part of https://github.com/harfbuzz/harfbuzz/issues/2270
But doesn't set new coords on the FT_Face. That's a lot more
work :(.
2022-05-26 05:02:50 -06:00
48db1c9583
[font] Add serial API
...
New API:
+ hb_font_get_serial()
+ hb_font_changed()
Fixes https://github.com/harfbuzz/harfbuzz/issues/2426
Unused internally as of now.
2022-05-26 05:02:50 -06:00
efa2a5796e
[map] Add hb_map_copy()
...
New API:
+ hb_map_copy()
2022-05-26 05:01:57 -06:00
3b28cff9c0
[cff1] Fix null dereference on memory alloc failure
2022-05-26 04:42:17 -06:00
8df9aba997
Actually try to fix null-size undefined behavior
...
Related to:
https://github.com/harfbuzz/harfbuzz/pull/2067
https://bugzilla.mozilla.org/show_bug.cgi?id=1577584
2022-05-26 03:59:21 -06:00
e00c7358a0
[subset] special case table presence check for hb_face_create_from_tables faces.
2022-05-25 22:39:00 +00:00
6a149a09e7
[subset] fix use of lazy static constructor.
2022-05-25 22:22:35 +00:00
d4c7939eb7
[subset] use a list of known tables instead of handled tables.
2022-05-25 22:11:32 +00:00
3472f73b79
[subset] also include no subset tables when guessing which tables are present.
2022-05-25 21:52:48 +00:00
9564d98739
[subset] fix subsetting of faces created via hb_face_create_for_tables.
...
Fixes #3609 .
2022-05-25 21:16:37 +00:00
6010feeeb5
[varStore] Rename variable as per review
...
https://github.com/harfbuzz/harfbuzz/pull/3605
2022-05-24 09:01:02 -06:00
3eb7eff487
Remove varstore cache use if HB_NO_VAR
2022-05-21 15:25:53 -06:00
099482a37a
[ot-font] Cache v_advance varstore as well
2022-05-21 15:24:23 -06:00
d9acc045f1
[VarStore] Sprinkle cache_t type around
...
It's available so no need to use void*.
2022-05-21 15:12:14 -06:00
da38312e42
[VarStore] Pepper cache with likely()
2022-05-21 15:12:14 -06:00
cf8f00e354
[VarStore] Don't use NAN
...
Is faster.
With this, I'm seeing 25 to 28% speedup in glyph_h_advances benchmark
of benchmark-font for var/hb tests.
2022-05-21 15:12:14 -06:00
5336ba70f6
[HVAR] Cache VarStore region scalars
2022-05-21 15:12:14 -06:00
880f50f7e4
Refactor varstore cache
2022-05-21 15:12:14 -06:00
f2a2fb91a3
[GPOS] Cache VarStore region scalars
2022-05-21 15:12:14 -06:00
5fbc70c59b
[VarStore] Add cache API
2022-05-21 15:12:14 -06:00
05e82aa12e
[ft] Add missing lock to kerning function
2022-05-20 17:36:59 -06:00
4ea2725704
[set/map] Expose hash API publicly
...
New API:
+ hb_set_hash()
+ hb_map_hash()
2022-05-20 13:19:48 -06:00
2e186d9f24
[buffer] Improve hash function of segment_properties_t
2022-05-20 13:19:48 -06:00
aee123fc83
4.3.0
2022-05-20 21:07:25 +02:00
975a5f9194
[array] Use hb_memcmp instead of memcmp
...
Fixes ubsan error.
2022-05-20 12:34:49 -06:00
55804e8d68
[hb-ft] Minor rearrange of struct members
...
To make clear what members the lock protects.
2022-05-20 11:40:44 -06:00
cbf8f44c9b
[subset-perf] swap instead of copying vertice's when reordering during sort.
2022-05-20 09:41:18 -06:00
b32ca2a292
[subset-perf] remove sort_kahn from repacker.
...
Without an optimized FIFO queue implementation it's nearly as slow as the now optimized sort_shortest_distance.
2022-05-20 09:41:18 -06:00
4266cf3be2
[array] Specialize operator== for bytes_t and ubytes_t
2022-05-19 18:16:54 -06:00
6eaa22e9d7
[serialize] Reduce link_t size from 16 to 12
2022-05-19 18:16:54 -06:00
30ba9a39e2
[vector] Add emplacing push implementation
2022-05-19 18:16:48 -06:00
73b8360dcf
[subset] fix fuzzer found underflow when heap push fails.
...
Fixes https://oss-fuzz.com/testcase-detail/5148625505746944 .
2022-05-19 17:02:34 -06:00
f1bf14ea89
Revert "[set] Cache hash value"
...
This reverts commit 44952bcc25https://github.com/harfbuzz/harfbuzz/issues/3599
2022-05-19 16:49:27 -06:00
b4d1ec310c
[algs] Declare coerce() as constexpr
2022-05-19 16:06:21 -06:00
2fdb7616f5
[map Further adjust hash function
2022-05-19 16:00:43 -06:00
01fc90b68c
[map] Adjust hash function
2022-05-19 16:00:06 -06:00
a47b0aebf5
[vector] Fix remove() implementation
...
test-vector under valgrind happy now.
2022-05-19 15:52:16 -06:00
3bd755c32d
[test-vector] Test remove()
...
Currently buggy. Valgrind confirms.
2022-05-19 15:51:18 -06:00
58f848daa8
[set/map] Adjust hash function return type
2022-05-19 15:43:19 -06:00
6544fc284f
[vector] Add further copy implementation
2022-05-19 15:43:15 -06:00
c19f116952
[meta] Remove non-existing gcc4 trait implementation
2022-05-19 15:43:15 -06:00
679b900e9b
[meta] Fix gcc4 trait implementation
2022-05-19 15:43:15 -06:00
fb77f48ffd
[vector] Optimize vector copy
2022-05-19 15:43:15 -06:00
28b44ac46a
[set] Switch set copy to vector operator =
...
Slows it down currently.
2022-05-19 15:43:15 -06:00
37d3275dec
[test-vector] Enable disabled test
...
This seems to work already.
2022-05-19 15:43:15 -06:00
544ffb913e
[set] Adjust grow_vector condition
2022-05-19 15:43:15 -06:00
0623aa598b
[benchmark-set] Add benchmark for set copy
2022-05-19 15:43:15 -06:00
44952bcc25
[set] Cache hash value
2022-05-19 14:02:48 -06:00
844ac328e4
[set] Fix hb_set_t hash stability
2022-05-19 13:54:31 -06:00
2d0b1248b2
[test-map] Test hb_set_t hash stability
...
Fails currently.
2022-05-19 13:53:53 -06:00
561e02fefb
[map] Make hb_map_t hashable
2022-05-19 13:41:53 -06:00
ad17699089
[map] Add is_equal() / towards making hb_map_t hashable
...
New API:
+ hb_map_is_equal()
2022-05-19 13:36:12 -06:00
14a24d8e3f
[vector] Make hb_vector_t hashable
2022-05-19 13:04:06 -06:00
124f9aeb9b
[set] Make hb_set_t hashable
2022-05-19 12:58:02 -06:00
3ab2c7935f
[subset-perf] Signficiantly speed up ClassDef*::subset.
...
Eliminates the usage of a glyph -> klass hash map and replaces it with a vector storing the mapping. This allows us to use the vector directly as the iterator driving the serialize. Approximately 1% speedup for Noto Nastaliq.
2022-05-19 12:37:28 -06:00
e3e685e5ee
[ot-tags] Fix `min_subtag_len` calculations
2022-05-18 18:30:01 -06:00
0b1c2ff96a
[subset-perf] Remove extra map lookup in ClassDef subset methods.
2022-05-18 18:25:36 -06:00
13ace77f1d
[subset-perf] Use glyph_map instead of set in ClassDefFormat.
2022-05-18 18:25:36 -06:00
adae2f2272
[subset-perf] Cache a glyph map for gsub.
...
This allows us in some cases to avoid using glyph_set_gsub as a filter.
2022-05-18 18:25:36 -06:00
202e6c4699
[subset] Remove unnecessary test
2022-05-18 17:12:43 -06:00
cedf739646
Add some commented-out code
2022-05-18 16:53:26 -06:00
6b62c10f02
[priority-queue] Remove old init/fini
2022-05-18 16:27:54 -06:00
bff8248a9d
[repacker] Pre-alloc vertices
2022-05-18 16:25:03 -06:00
39a424caf0
[priority-queue] Optimize heap access
2022-05-18 16:19:44 -06:00
9308659fd7
[priority-queue] Optimize swap()
2022-05-18 16:14:25 -06:00
c7317ef761
[repacker] Avoid destroying and recreating objects
2022-05-18 16:03:41 -06:00
864e09a0c4
[repacker] Reuse allocated vector
2022-05-18 15:59:29 -06:00
ca77f16470
[repacker] Remove unnecessary vector .fini() calls
2022-05-18 15:55:49 -06:00
4cfc2d668e
[subset] Use a std::move on set_t when feasible
2022-05-18 15:32:19 -06:00
1f578b5a32
[set] Add pre-allocation internal API
2022-05-18 15:25:00 -06:00
48dfbd54a3
[subset] minor cleanup.
2022-05-18 15:24:54 -06:00
482c6e5dc4
[subset-perf] Speed up Coverage::serialize by caching iterator.
2022-05-18 15:24:54 -06:00
14b18725f0
In Coverage::iter_t, assume iterators are from same Coverage object
...
No need to support otherwise.
2022-05-18 15:14:32 -06:00
27141735c3
[subset] Add Coverage::__end__ implementation
2022-05-18 15:12:49 -06:00
c476f58adb
[subset] Write CoverageFormat2::intersects_coverage() as bsearch()
2022-05-18 14:20:23 -06:00
63c6695108
[ot-layout] Cosmetic
...
The implementation of HBUINT16 operator == is slower than just
comparing to ints.
2022-05-18 13:53:52 -06:00
777debd748
[subset] Rewrite CoverageFormat2::intersects as dagger
2022-05-18 13:46:06 -06:00
cf5001fac7
[subset] Optimize CoverageFormat2::intersected_coverage_glyphs
2022-05-18 13:38:29 -06:00
6f37c20798
[subset] Minor rewrite in CoverageFormat2::serialize()
2022-05-18 13:25:42 -06:00
e91863b717
[subset-cff] Pre-size map in subr_remap_t::create()
2022-05-18 12:39:55 -06:00
ce60462173
[subset-plan] Pre-size maps in _create_old_gid_to_new_gid_map()
2022-05-18 12:34:27 -06:00
f82ee17a75
[map] Pre-size map in constructor if we can
2022-05-18 12:19:49 -06:00
b5aa8a27ea
[subset-cff] Cosmetic
2022-05-18 11:58:58 -06:00
0b201623f5
[subset-cff] Fix previous commit
...
Oops!
2022-05-18 11:58:22 -06:00
4792309265
[subset-cff] Access vector directly
2022-05-18 11:54:08 -06:00
7c86f2e763
[subset-cff] Pre-alloc out buffer
2022-05-18 11:45:27 -06:00
0761e7cdfd
[subset-cff] Avoid resetting buffer as encoder does
2022-05-18 11:38:25 -06:00
71aa10a394
[subset-cff] Manually grow vector to avoid memset overhead
2022-05-18 11:37:24 -06:00
f455cc53fd
[subset-cff] Reuse buffer allocation
2022-05-18 11:31:55 -06:00
e24797aeac
[ot-tags] Follow-up to previous commit
...
Part of https://github.com/harfbuzz/harfbuzz/issues/3591
2022-05-18 11:10:10 -06:00
Behdad Esfahbod
Garret Rieger
Khaled Hosny
David Corbett