Thomas Pelletier
618f0181ac
AST Tweaks ( #551 )
...
* Use pointers instead of copying around ast.Node
Node is a 56B struct that is constantly in the hot path. Passing nodes
around by copy had a cost that started to add up. This change replaces
them by pointers. Using unsafe pointer arithmetic and converting
sibling/child indexes to relative offsets, it removes the need to carry
around a pointer to the root of the tree. This saves 8B per Node. This
space will be used to store an extra []byte slice to provide contextual
error handling on all nodes, including the ones whose data is different
than the raw input (for example: strings with escaped characters), while
staying under the size of a cache line.
* Remove conditional
* Add Raw to track range in data for parsed values
* Simplify reference tracking
2021-06-03 21:48:51 -04:00
Thomas Pelletier
250e073408
Stack-based unmarshaler ( #546 )
...
* Benchmark script
* Rewrite unmarshaler using the stack
Instead of tracking the build chain using `target`s, use the stack
instead.
Working and most benchmarks look good, but regression on structs unmarshalling.
~60% slower on ReferenceFile/struct.
* Shortcut to check if last node of iterator
* Remove unecessary pointer allocation
* Skip over unused keys without marking them as seen
* Add some tests
* Fix mktemp on macos
2021-05-31 12:14:13 -04:00
Thomas Pelletier
840df4a229
seen tracker: use array for storage ( #545 )
...
name old time/op new time/op delta
UnmarshalDataset/config-32 81.2ms ± 3% 77.8ms ± 3% -4.25% (p=0.000 n=20+19)
UnmarshalDataset/canada-32 104ms ± 5% 105ms ± 4% ~ (p=0.102 n=20+20)
UnmarshalDataset/citm_catalog-32 57.5ms ± 5% 59.0ms ± 5% +2.54% (p=0.033 n=20+20)
UnmarshalDataset/twitter-32 25.7ms ± 7% 28.1ms ± 5% +9.33% (p=0.000 n=20+20)
UnmarshalDataset/code-32 305ms ± 6% 292ms ± 5% -4.29% (p=0.000 n=20+19)
UnmarshalDataset/example-32 519µs ± 6% 522µs ± 5% ~ (p=0.659 n=20+20)
UnmarshalSimple/struct-32 1.44µs ± 1% 1.17µs ± 6% -18.78% (p=0.000 n=14+20)
UnmarshalSimple/map-32 2.30µs ± 4% 1.99µs ± 4% -13.65% (p=0.000 n=20+19)
ReferenceFile/struct-32 44.1µs ± 4% 38.1µs ± 5% -13.61% (p=0.000 n=18+20)
ReferenceFile/map-32 197µs ± 7% 189µs ± 5% -3.91% (p=0.000 n=20+20)
HugoFrontMatter-32 45.9µs ± 6% 39.3µs ± 6% -14.46% (p=0.000 n=19+20)
name old speed new speed delta
UnmarshalDataset/config-32 12.9MB/s ± 3% 13.5MB/s ± 3% +4.42% (p=0.000 n=20+19)
UnmarshalDataset/canada-32 21.1MB/s ± 5% 20.9MB/s ± 4% ~ (p=0.101 n=20+20)
UnmarshalDataset/citm_catalog-32 9.72MB/s ± 6% 9.47MB/s ± 5% -2.53% (p=0.031 n=20+20)
UnmarshalDataset/twitter-32 17.2MB/s ± 7% 15.8MB/s ± 5% -8.57% (p=0.000 n=20+20)
UnmarshalDataset/code-32 8.81MB/s ± 7% 9.20MB/s ± 5% +4.47% (p=0.000 n=20+19)
UnmarshalDataset/example-32 15.6MB/s ± 6% 15.5MB/s ± 5% ~ (p=0.644 n=20+20)
UnmarshalSimple/struct-32 7.61MB/s ± 1% 9.39MB/s ± 7% +23.33% (p=0.000 n=15+20)
UnmarshalSimple/map-32 4.78MB/s ± 4% 5.54MB/s ± 5% +15.85% (p=0.000 n=20+19)
ReferenceFile/struct-32 119MB/s ± 4% 138MB/s ± 5% +15.79% (p=0.000 n=18+20)
ReferenceFile/map-32 26.6MB/s ± 7% 27.7MB/s ± 5% +4.06% (p=0.000 n=20+20)
HugoFrontMatter-32 11.9MB/s ± 6% 13.9MB/s ± 6% +16.91% (p=0.000 n=19+20)
name old alloc/op new alloc/op delta
UnmarshalDataset/config-32 16.9MB ± 0% 13.9MB ± 0% -17.48% (p=0.000 n=18+18)
UnmarshalDataset/canada-32 74.3MB ± 0% 74.3MB ± 0% -0.00% (p=0.000 n=20+20)
UnmarshalDataset/citm_catalog-32 37.3MB ± 0% 37.3MB ± 0% +0.11% (p=0.000 n=20+20)
UnmarshalDataset/twitter-32 15.6MB ± 0% 15.6MB ± 0% ~ (p=0.211 n=19+20)
UnmarshalDataset/code-32 59.5MB ± 0% 52.4MB ± 0% -11.96% (p=0.000 n=19+20)
UnmarshalDataset/example-32 238kB ± 0% 239kB ± 0% +0.02% (p=0.000 n=18+20)
UnmarshalSimple/struct-32 981B ± 0% 709B ± 0% -27.73% (p=0.000 n=20+20)
UnmarshalSimple/map-32 1.45kB ± 0% 1.17kB ± 0% -18.82% (p=0.000 n=20+20)
ReferenceFile/struct-32 11.8kB ± 0% 9.7kB ± 0% -17.64% (p=0.000 n=20+20)
ReferenceFile/map-32 51.5kB ± 0% 52.2kB ± 0% +1.30% (p=0.000 n=20+17)
HugoFrontMatter-32 12.1kB ± 0% 11.1kB ± 0% -7.97% (p=0.000 n=20+19)
name old allocs/op new allocs/op delta
UnmarshalDataset/config-32 645k ± 0% 557k ± 0% -13.76% (p=0.000 n=20+16)
UnmarshalDataset/canada-32 896k ± 0% 896k ± 0% -0.00% (p=0.000 n=20+20)
UnmarshalDataset/citm_catalog-32 380k ± 0% 377k ± 0% -0.75% (p=0.000 n=19+20)
UnmarshalDataset/twitter-32 158k ± 0% 158k ± 0% -0.01% (p=0.000 n=18+18)
UnmarshalDataset/code-32 2.92M ± 0% 2.57M ± 0% -11.87% (p=0.000 n=20+20)
UnmarshalDataset/example-32 3.66k ± 0% 3.64k ± 0% -0.63% (p=0.000 n=20+20)
UnmarshalSimple/struct-32 13.0 ± 0% 10.0 ± 0% -23.08% (p=0.000 n=20+20)
UnmarshalSimple/map-32 22.0 ± 0% 19.0 ± 0% -13.64% (p=0.000 n=20+20)
ReferenceFile/struct-32 253 ± 0% 155 ± 0% -38.74% (p=0.000 n=20+20)
ReferenceFile/map-32 1.67k ± 0% 1.44k ± 0% -14.23% (p=0.000 n=20+20)
HugoFrontMatter-32 357 ± 0% 313 ± 0% -12.32% (p=0.000 n=20+20)
2021-05-26 18:47:00 -04:00
Thomas Pelletier
ea225df3ed
v2: errors ( #534 )
...
```
name old time/op new time/op delta
UnmarshalDataset/config-32 86.7ms ± 2% 87.5ms ± 2% ~ (p=0.113 n=9+10)
UnmarshalDataset/canada-32 129ms ± 4% 106ms ± 3% -17.94% (p=0.000 n=10+10)
UnmarshalDataset/citm_catalog-32 59.4ms ± 5% 58.7ms ± 5% ~ (p=0.393 n=10+10)
UnmarshalDataset/twitter-32 27.0ms ± 7% 26.9ms ± 6% ~ (p=0.720 n=10+9)
UnmarshalDataset/code-32 326ms ± 4% 322ms ± 7% ~ (p=0.661 n=9+10)
UnmarshalDataset/example-32 510µs ±11% 526µs ± 7% ~ (p=0.182 n=10+9)
UnmarshalSimple-32 1.41µs ± 6% 1.41µs ± 4% ~ (p=0.736 n=10+9)
ReferenceFile-32 45.6µs ± 3% 43.9µs ±10% ~ (p=0.089 n=10+10)
name old speed new speed delta
UnmarshalDataset/config-32 12.1MB/s ± 2% 12.0MB/s ± 2% ~ (p=0.108 n=9+10)
UnmarshalDataset/canada-32 17.1MB/s ± 4% 20.9MB/s ± 3% +21.86% (p=0.000 n=10+10)
UnmarshalDataset/citm_catalog-32 9.41MB/s ± 5% 9.51MB/s ± 5% ~ (p=0.362 n=10+10)
UnmarshalDataset/twitter-32 16.4MB/s ± 8% 16.5MB/s ± 6% ~ (p=0.704 n=10+9)
UnmarshalDataset/code-32 8.24MB/s ± 4% 8.34MB/s ± 7% ~ (p=0.675 n=9+10)
UnmarshalDataset/example-32 15.9MB/s ±11% 15.4MB/s ± 7% ~ (p=0.182 n=10+9)
ReferenceFile-32 115MB/s ± 4% 120MB/s ±10% ~ (p=0.085 n=10+10)
name old alloc/op new alloc/op delta
UnmarshalDataset/config-32 16.9MB ± 0% 16.9MB ± 0% -0.02% (p=0.000 n=10+10)
UnmarshalDataset/canada-32 76.8MB ± 0% 74.3MB ± 0% -3.31% (p=0.000 n=10+10)
UnmarshalDataset/citm_catalog-32 37.3MB ± 0% 37.1MB ± 0% -0.60% (p=0.000 n=9+10)
UnmarshalDataset/twitter-32 15.6MB ± 0% 15.6MB ± 0% -0.09% (p=0.000 n=10+10)
UnmarshalDataset/code-32 60.2MB ± 0% 59.3MB ± 0% -1.51% (p=0.000 n=10+9)
UnmarshalDataset/example-32 238kB ± 0% 238kB ± 0% -0.18% (p=0.000 n=10+10)
ReferenceFile-32 11.8kB ± 0% 11.8kB ± 0% ~ (all equal)
name old allocs/op new allocs/op delta
UnmarshalDataset/config-32 653k ± 0% 645k ± 0% -1.20% (p=0.000 n=10+6)
UnmarshalDataset/canada-32 1.01M ± 0% 0.90M ± 0% -11.04% (p=0.000 n=9+10)
UnmarshalDataset/citm_catalog-32 384k ± 0% 370k ± 0% -3.75% (p=0.000 n=10+10)
UnmarshalDataset/twitter-32 160k ± 0% 157k ± 0% -1.32% (p=0.000 n=10+10)
UnmarshalDataset/code-32 2.97M ± 0% 2.91M ± 0% -2.15% (p=0.000 n=10+7)
UnmarshalDataset/example-32 3.69k ± 0% 3.63k ± 0% -1.52% (p=0.000 n=10+10)
ReferenceFile-32 253 ± 0% 253 ± 0% ~ (all equal)
```
2021-05-08 16:04:25 -04:00
Thomas Pelletier
9b67e40640
decoder: strict mode ( #512 )
2021-04-20 21:26:22 -04:00