Commit Graph
15493 Commits
Author SHA1 Message Date
Behdad Esfahbod d1f58e5762 Tweak an unlikely
This doesn't seem to slowdown benchmarks much, but speeds up Duployan
by ~8%.
2025-12-31 16:28:56 -07:00
Behdad Esfahbod 1c3633e121 [GDEF] Use set-digest, not bit-set, for mark coverages again
This seems to be slightly faster, more memory efficient, and
matches HarfRust.
2025-12-31 15:56:43 -07:00
Behdad Esfahbod d58bc8c94d Remove a couple inlines
Speeds up things a bit.
2025-12-31 15:45:58 -07:00
Behdad Esfahbod a789943958 [binary-cache] Enlarge to 256 entries
2% speedup in NotoNastaliqUrdu benchmark.
2025-12-30 12:15:12 -07:00
Behdad Esfahbod a4757aefc2 Remove unused default value
In the HarfRust port, compiler warns about unused value. lol
2025-12-27 17:22:46 -07:00
Behdad Esfahbod aaf22616ba [contextual] Fix typo
This involves a very edge case, where there are no rules in the ruleset.
One copy of the code was correct, other not. Fix.
2025-12-27 17:20:25 -07:00
Khaled Hosny 5a31dd02f0 12.3.0 2025-12-27 21:58:17 +02:00
Behdad Esfahbod 635cd0e854 [calculate-scalar] Another tiny speedup
~1% in varc-hanzi benchmark.
2025-12-24 15:15:06 -07:00
Behdad Esfahbod 17092a67cc [calculate-scalar] A bit more speedup
5% speedup in varc-hanzi drawing.
No slowdown on GoogleSansFlex.
2025-12-24 14:17:38 -07:00
Behdad Esfahbod 21d15507fc [calculate-scalar] Another path we can use the cache
If the scalar=1.0 even if we have interims, we can
still use the cache.
2025-12-24 12:13:54 -07:00
Behdad Esfahbod 1a94172282 [calculate-scalar] Use cache sometimes even if has_interm
More involved logic.
2025-12-24 01:41:40 -07:00
Behdad Esfahbod 0f6fecfaf4 [calculate_scalar] Only jumpskip if num-coords >= 16
Otherwise, it was slowing down GoogleSansFlex rendering.
2025-12-24 01:41:36 -07:00
Behdad Esfahbod 22c876331a [calculate-scalar] Jump-skip 4 zeros at a time 2025-12-23 17:49:37 -07:00
Behdad Esfahbod e7796b5c2a [open-type] Add 64bit integer types 2025-12-23 17:49:37 -07:00
Behdad Esfahbod dcc3e6d106 [algs] Add 64bit int types fastpath 2025-12-23 15:38:18 -07:00
Behdad Esfahbod d960ee32ac [TupleValues] Disable a "fastpath" 2025-12-21 22:06:06 -07:00
Behdad Esfahbod f6ba182e65 Revert "[VARC] Use non-thread-safe scalar-cache"
This reverts commit 2c0dbf1f92.
2025-12-21 20:57:42 -07:00
Behdad Esfahbod 2c0dbf1f92 [VARC] Use non-thread-safe scalar-cache
Since the cache is local to the draw() call.

However, this seems to slow down thing for reasons I don't understand.
Going to revert.
2025-12-21 20:56:52 -07:00
Behdad Esfahbod 19ab59f2f6 [scalar-cache] Unroll a loop 2025-12-21 20:32:21 -07:00
Behdad Esfahbod 82a0d4bd03 [TupleValues] Inline ensure_run fastpath 2025-12-21 20:19:17 -07:00
Behdad Esfahbod c48fa2fa4d [ligature] Adjust fast path to be in sync with contextuals
Shouldn't have any observable difference.
2025-12-20 18:29:28 -07:00
Behdad Esfahbod d15bc8d4e9 [contextual] Use a binary coverage cache (in format2)
Reduces cache size by half.

Since we discard the actual coverage number anyway.
2025-12-20 17:08:48 -07:00
Behdad Esfahbod f9971108eb [contextual] Add coverage cache for format2
5% speedup in NotoNastaliq benchmark.
2025-12-20 16:53:20 -07:00
Behdad Esfahbod f629ecdc5c [contextual] Implement rule-skipping for ChainContext as well
Speeds up NotoNastaliq benchmark another 5%.
2025-12-20 16:29:07 -07:00
Behdad Esfahbod 291e0b474a [matching] Tweak how we choose slow path if ZWJ/ZWNJ around
The new if() blocks are a bug fix.

The removed if() block and change to using iter_context instead,
enables the fast path for Arabic shaping now, and speeds up
NotoNastaliqUrdu benchmark by 10%.
2025-12-20 14:45:32 -07:00
Behdad Esfahbod bde924f275 [RuleSet] Add rule-skipping
Shows ~5% speedup in NotoNastaliqUrdu benchmark.
2025-12-20 14:07:43 -07:00
Behdad Esfahbod 761cd5a5c0 [gsubgpos] Fast-track matching of 0 items 2025-12-20 14:05:49 -07:00
Qunxin Liu babdb46be3 [repacker] bug fix for splitting shared subtables (#5686)
And fix splitting for ligatureSubst table with shared coverage table
2025-12-19 21:42:20 -07:00
Qunxin Liu cf31e7ac4e [instancer] support full instancing for avar2 (#5676) 2025-12-11 10:54:49 -07:00
Khaled Hosny a68565099d [kbts] Update to version 2.1 (#5670) 2025-12-06 13:02:03 -07:00
Behdad Esfahbod e9c463cef2 [aat-machine] Add a fast-path (#5667)
Speeds up LucidaGrande benchmark by 12%, Menlo by 6%.
2025-12-02 23:34:14 -07:00
Behdad Esfahbod 6593d25441 [aat-machine] Remove redundant check for buffer successful (#5666)
The loop does it already.
2025-12-02 23:14:44 -07:00
Behdad Esfahbod 15eea57f7b [kbts] Macro 2025-11-29 16:54:40 -07:00
Behdad Esfahbod 31eac71703 [kbts] Fix buffer access (#5660) 2025-11-29 16:14:30 -07:00
Khaled Hosny 46bc4e95f9 [kbts] Update to version 2.0 2025-11-28 12:56:36 +02:00
Behdad Esfahbod d9e3b4efed Fix a compiler warning / error 2025-11-25 18:00:39 -07:00
Behdad Esfahbod 4bfa44407d [TupleValues] Inline decompile_deltas (#5651)
8% speedup in GoogleSansFlex heavy variations test.

$ b/perf/benchmark-font GSF.ttf --benchmark_filter=draw.*var/ot --variations=ROND=40,wght=500,wdth=50,opsz=54,slnt=-10,GRAD=50
2025-11-23 21:27:47 -07:00
Behdad Esfahbod c5a5334aea [TupleValues] Micro-optimize
The compiler was using a memset here, but I think it was also
emitting one conditional (checking i <= stop to begin with).
We know that condition is always true. This shows a measurable
1 or 2 percent speedup for me in a mega benchmark of Google Sans
Flex.
2025-11-23 20:59:48 -07:00
Behdad Esfahbod 80e4fb6a69 [tuple-var] Micro-micro-optimized
0.5% speedup drawing var GoogleSansFlex.
2025-11-22 18:08:27 -07:00
Behdad Esfahbod 90f5cf8858 [scalar-cache] Micro-optimize (#5647)
Value is an atomic. Read it once.

4% speedup drawing GoogleSansFlex at variation.
2025-11-21 23:12:42 -07:00
Behdad Esfahbod ecb8320b74 [atomic] Remove unused operator |= (#5648)
Not sure why it was added.
2025-11-21 23:12:29 -07:00
Behdad Esfahbod f632633900 [array] Fix build 2025-11-21 17:47:14 -07:00
Behdad Esfahbod 072299a25c [gvar] More speedup of drawing mega-var-fonts
Another 4% in GoogleSansFlex.
2025-11-21 17:37:00 -07:00
Behdad Esfahbod 9d7b229c7f [scalar-cache] Speed up
Another 7% speed up var drawing of GoogleSansFlex
2025-11-21 17:23:13 -07:00
Behdad Esfahbod f32abad817 [gvar] Speed up tupleHeader size calculation
25% speedup in GoogleSansFlex mega-var-font var drawing

$ b/perf/benchmark-font GSF.ttf --benchmark_filter=draw.*var/ot --variations=ROND=100
2025-11-21 17:11:54 -07:00
Behdad Esfahbod 716ae43fca Typpo 2025-11-21 16:58:43 -07:00
Behdad Esfahbod 52b38616f3 [rust] Fix check-symbols, again for rust update 2025-11-20 19:29:46 -07:00
Behdad Esfahbod 8dcd621739 [morx] Add an unlikely 2025-11-20 19:16:51 -07:00
Behdad Esfahbod 27c1b4b3fc [aat] Another unlikely 2025-11-20 13:24:44 -07:00
Behdad Esfahbod bf86590db6 [aat] Add an unlikely
Measurable speedup.
2025-11-20 04:34:54 -07:00