perf(rome_cli): replace global allocator with jemalloc #3237

ematipico · 2022-09-16T09:04:57Z

Summary

This PR replaces the default global allocator to use jemallocator on all operative systems, exception for Windows because it's not supported.

Test Plan

Here's some benchmarks from my machine (I use macOS)

parser

group                                 jamalloc                               main
-----                                 --------                               ----
parser/checker.ts                     1.00     98.9±7.70ms    26.3 MB/sec    1.16    115.2±3.25ms    22.6 MB/sec
parser/compiler.js                    1.00     52.8±1.15ms    19.8 MB/sec    1.49     79.0±1.86ms    13.3 MB/sec
parser/d3.min.js                      1.00     34.2±2.76ms     7.7 MB/sec    1.22     41.9±0.96ms     6.3 MB/sec
parser/dojo.js                        1.00      3.1±0.24ms    22.2 MB/sec    1.16      3.6±0.09ms    19.2 MB/sec
parser/ios.d.ts                       1.00     76.2±3.42ms    24.5 MB/sec    1.33    101.1±5.28ms    18.5 MB/sec
parser/jquery.min.js                  1.00      8.4±0.70ms     9.8 MB/sec    1.40     11.8±0.25ms     7.0 MB/sec
parser/math.js                        1.00     68.2±6.90ms     9.5 MB/sec    1.21     82.7±1.39ms     7.8 MB/sec
parser/parser.ts                      1.00      2.2±0.18ms    22.1 MB/sec    1.13      2.5±0.07ms    19.6 MB/sec
parser/pixi.min.js                    1.00     43.2±1.93ms    10.2 MB/sec    1.33     57.4±1.07ms     7.6 MB/sec
parser/react-dom.production.min.js    1.00     11.4±0.31ms    10.1 MB/sec    1.32     15.1±0.43ms     7.6 MB/sec
parser/react.production.min.js        1.00   618.1±75.15µs    10.0 MB/sec    1.31   810.4±21.71µs     7.6 MB/sec
parser/router.ts                      1.00  1744.8±96.27µs    35.2 MB/sec    1.23      2.1±0.08ms    28.6 MB/sec
parser/tex-chtml-full.js              1.00     92.9±5.44ms     9.8 MB/sec    1.21    112.5±6.80ms     8.1 MB/sec
parser/three.min.js                   1.00     51.7±7.25ms    11.4 MB/sec    1.15     59.4±1.29ms     9.9 MB/sec
parser/typescript.js                  1.00   381.1±29.09ms    24.9 MB/sec    1.25    478.1±8.76ms    19.9 MB/sec
parser/vue.global.prod.js             1.00     13.7±0.39ms     8.8 MB/sec    1.39     19.1±0.48ms     6.3 MB/sec

formatter

group                                    formatter-jamalloc                      formatter-main
-----                                    ------------------                      --------------
formatter/checker.ts                     1.00   360.2±11.12ms     7.2 MB/sec     1.99   718.7±88.40ms     3.6 MB/sec
formatter/compiler.js                    1.00   213.1±24.96ms     4.9 MB/sec     1.94   413.6±32.24ms     2.5 MB/sec
formatter/d3.min.js                      1.00    162.9±6.40ms  1647.7 KB/sec     2.14   348.4±24.40ms   770.3 KB/sec
formatter/dojo.js                        1.00     10.1±0.38ms     6.8 MB/sec     1.83     18.6±0.68ms     3.7 MB/sec
formatter/ios.d.ts                       1.00    224.4±7.34ms     8.3 MB/sec     2.09   469.8±52.19ms     4.0 MB/sec
formatter/jquery.min.js                  1.00     46.2±3.25ms  1830.6 KB/sec     1.88     86.9±4.87ms   973.9 KB/sec
formatter/math.js                        1.00   316.2±16.29ms     2.0 MB/sec     2.07   654.2±44.04ms  1013.5 KB/sec
formatter/parser.ts                      1.00      7.6±0.24ms     6.4 MB/sec     1.78     13.5±0.81ms     3.6 MB/sec
formatter/pixi.min.js                    1.00   182.9±20.46ms     2.4 MB/sec     2.04   372.9±26.13ms  1205.1 KB/sec
formatter/react-dom.production.min.js    1.00     53.6±3.05ms     2.1 MB/sec     2.16   115.6±16.34ms  1019.3 KB/sec
formatter/react.production.min.js        1.00      2.5±0.07ms     2.5 MB/sec     1.82      4.5±0.22ms  1385.1 KB/sec
formatter/router.ts                      1.00      6.0±0.37ms    10.2 MB/sec     1.65      9.9±0.51ms     6.2 MB/sec
formatter/tex-chtml-full.js              1.00   405.0±17.74ms     2.3 MB/sec     2.19   886.4±69.69ms  1052.8 KB/sec
formatter/three.min.js                   1.00   210.4±23.22ms     2.8 MB/sec     1.99   418.5±33.31ms  1436.6 KB/sec
formatter/typescript.js                  1.00  1444.9±154.91ms     6.6 MB/sec    1.67       2.4±0.10s     3.9 MB/sec
formatter/vue.global.prod.js             1.00     70.9±5.05ms  1740.0 KB/sec     2.07   146.9±13.28ms   839.9 KB/sec

analyzer

group                     analyzer-jamalloc                      analyzer-main
-----                     -----------------                      -------------
analyzer/css.js           1.00  1682.7±79.01µs     6.9 MB/sec    2.03      3.4±0.10ms     3.4 MB/sec
analyzer/index.js         1.00      4.5±0.08ms     7.2 MB/sec    2.43     10.9±0.36ms     3.0 MB/sec
analyzer/lint.ts          1.00      2.5±0.20ms    16.8 MB/sec    1.73      4.3±0.12ms     9.7 MB/sec
analyzer/parser.ts        1.00      5.8±0.28ms     8.4 MB/sec    1.84     10.7±0.18ms     4.6 MB/sec
analyzer/router.ts        1.00      4.0±0.14ms    15.4 MB/sec    1.85      7.4±0.10ms     8.3 MB/sec
analyzer/statement.ts     1.00      5.4±0.25ms     6.5 MB/sec    1.99     10.8±0.72ms     3.3 MB/sec
analyzer/typescript.ts    1.00      9.1±0.52ms     6.0 MB/sec    1.67     15.2±0.26ms     3.6 MB/sec

Memory consumption

I used cargo instruments and I am not an expert at using the trace tool provided by XCode. I share here the screenshots and the trace file in case someone has more knowledge than me.

I run the formatter on the prettier repository, their src folder. No --write.

`jemalloc`

Link to Rome shared drive

`main`

Link to Rome shared drive

netlify · 2022-09-16T09:05:03Z

✅ Deploy Preview for rometools canceled.

Name	Link
🔨 Latest commit	`3e557de`
🔍 Latest deploy log	https://app.netlify.com/sites/rometools/deploys/6329b07a02466a00094865a2

github-actions · 2022-09-16T09:10:13Z

Parser conformance results on ubuntu-latest

js/262

Test result	`main` count	This PR count	Difference
Total	45879	45879	0
Passed	44939	44939	0
Failed	940	940	0
Panics	0	0	0
Coverage	97.95%	97.95%	0.00%

jsx/babel

Test result	`main` count	This PR count	Difference
Total	39	39	0
Passed	36	36	0
Failed	3	3	0
Panics	0	0	0
Coverage	92.31%	92.31%	0.00%

symbols/microsoft

Test result	`main` count	This PR count	Difference
Total	5946	5946	0
Passed	1621	1621	0
Failed	4325	4325	0
Panics	0	0	0
Coverage	27.26%	27.26%	0.00%

ts/babel

Test result	`main` count	This PR count	Difference
Total	588	588	0
Passed	519	519	0
Failed	69	69	0
Panics	0	0	0
Coverage	88.27%	88.27%	0.00%

ts/microsoft

Test result	`main` count	This PR count	Difference
Total	16257	16257	0
Passed	12395	12395	0
Failed	3862	3862	0
Panics	0	0	0
Coverage	76.24%	76.24%	0.00%

github-actions · 2022-09-16T09:11:56Z

Playground for commit 7647d7b

MichaReiser · 2022-09-16T09:12:52Z

Have you compared performance with mimalloc? Depending on different allocators for every platform increases the risk of platforms specific issues and requires us to test every change on every platform.

That's why I would prefer to use the default system allocator for all (maybe except windows) platforms or replace the default for all platforms.

MichaReiser · 2022-09-16T09:17:25Z

You know... this will make it more difficult for me to prove that any of my work reducing allocation is any good :P

IWANABETHATGUY · 2022-09-16T09:20:27Z

Would you mind explaining why replacing an allocator will make it harder to prove reducing allocation ?

Forgive me this stupid question.

MichaReiser · 2022-09-16T09:21:34Z

Would you mind explaining why replacing an allocator will make it harder to prove reducing allocation ?

If every single allocations becomes cheaper than reducing the number of allocations has less effect on the overall performance.

@ematipico Would you mind also comparing the memory consumption? Do you have links that compare the different allocators?

IWANABETHATGUY · 2022-09-16T09:22:29Z

Would you mind explaining why replacing an allocator will make it harder to prove reducing allocation ?

If every single allocations becomes cheaper than reducing the number of allocations has less effect on the overall performance.

@ematipico Would you mind also comparing the memory consumption? Do you have links that compare the different allocators?

Thanks

leops · 2022-09-16T09:56:59Z

Depending on different allocators for every platform increases the risk of platforms specific issues and requires us to test every change on every platform.

Technically we are already depending on different allocators for different platforms, since we use whatever allocator is provided by the OS. I don't know if mimalloc compares favorably to jemalloc in terms of performance, as always with perfs this is something we need to benchmark with our workloads to be sure. In any case I think it would be interesting to use jemalloc on all Unix systems instead of just macOS and mimalloc on Windows, this way we would still smooth out most platform discrepancies by using the same allocator (almost) everywhere

Also, what's the impact of this change on the size of the rome binary ?

MichaReiser · 2022-09-16T10:06:02Z

Depending on different allocators for every platform increases the risk of platforms specific issues and requires us to test every change on every platform.

Technically we are already depending on different allocators for different platforms, since we use whatever allocator is provided by the OS. I don't know if mimalloc compares favorably to jemalloc in terms of performance, as always with perfs this is something we need to benchmark with our workloads to be sure. In any case I think it would be interesting to use jemalloc on all Unix systems instead of just macOS and mimalloc on Windows, this way we would still smooth out most platform discrepancies by using the same allocator (almost) everywhere

Also, what's the impact of this change on the size of the rome binary ?

What was your reasoning why you preferred minmalloc over jemalloc?

ematipico · 2022-09-16T10:18:25Z

The reason why I tried jemalloctor for macOS first was mainly because lighthingcss uses it (former ParcelCSS). And I gave it a try and see what's the result on our benches.

I will give it a whirl in terms of memory usages once I manage to download again all the necessary software. We can close it if you think it's not worth it.

leops · 2022-09-16T11:56:39Z

What was your reasoning why you preferred minmalloc over jemalloc?

For Windows the reason was simply that tikv-jemallocator doesn't support it

MichaReiser · 2022-09-16T12:43:00Z

The reason why I tried jemalloctor for macOS first was mainly because lighthingcss uses it (former ParcelCSS). And I gave it a try and see what's the result on our benches.

I will give it a whirl in terms of memory usages once I manage to download again all the necessary software. We can close it if you think it's not worth it.

The results seem very promising! My main concern is that I want to avoid having one custom allocator for every platform that we support. Using minmalloc for windows and jemallc for UNIX systems seems a reasonable balance between performance and too many configurations (even though I haven't seen any significant improvements on linux)

ematipico · 2022-09-20T07:36:00Z

So, should we just switch to use this allocator for all the OS expect windows?

MichaReiser · 2022-09-20T11:32:22Z

So, should we just switch to use this allocator for all the OS expect windows?

Probably. But it would still be interesting to get some more numbers:

How does it change performance results on linux
Is there a way for us to compare memory consumption before/after? If so, what are the numbers

calibre-analytics · 2022-09-20T12:16:58Z

Comparing perf(rome_cli): replace global allocator with jemalloc for macos Snapshot #3 to median since last deploy of rome.tools.

	LCP^?	CLS^?	TBT^?
Overall ^{Median across all pages and test profiles}	1.95s ^from 559ms	0.0 ^no change	147ms ^from 42ms
Chrome Desktop ^{Chrome Desktop • Cable}	1.95s ^from 586ms	0.0 ^no change	379ms ^from 239ms
iPhone, 4G LTE ^{iPhone 12 • 4G LTE}	1s ^from 218ms	0.0 ^no change	3ms ^from 6ms
Motorola Moto G Power, 3G connection ^{Motorola Moto G Power • Regular 3G}	12.5s ^from 559ms	0.0 ^no change	147ms ^from 42ms

1 page tested

Home

Browser previews

Chrome Desktop	iPhone, 4G LTE	Motorola Moto G Power, 3G connection

Most significant changes

	Value	Budget
JS Parse & Compile Motorola Moto G Power, 3G connection	1.51s ^from 29ms
JS Parse & Compile iPhone, 4G LTE	458ms ^from 9ms
Total JavaScript Size in Bytes Chrome Desktop	4.11 MB ^{from 86.8 KB}
Total JavaScript Size in Bytes iPhone, 4G LTE	4.11 MB ^{from 86.8 KB}
Total JavaScript Size in Bytes Motorola Moto G Power, 3G connection	4.11 MB ^{from 86.8 KB}

_{28 other significant changes: JS Parse & Compile on Chrome Desktop, First Contentful Paint on Motorola Moto G Power, 3G connection, Largest Contentful Paint on Motorola Moto G Power, 3G connection, Speed Index on Motorola Moto G Power, 3G connection, Total Page Size in Bytes on Chrome Desktop, Total Page Size in Bytes on iPhone, 4G LTE, Total Page Size in Bytes on Motorola Moto G Power, 3G connection, Number of Requests on Motorola Moto G Power, 3G connection, Number of Requests on Chrome Desktop, Number of Requests on iPhone, 4G LTE, Time to Interactive on Motorola Moto G Power, 3G connection, First Contentful Paint on Chrome Desktop, First Contentful Paint on iPhone, 4G LTE, Time to Interactive on Chrome Desktop, Speed Index on Chrome Desktop, Speed Index on iPhone, 4G LTE, Largest Contentful Paint on iPhone, 4G LTE, Time to Interactive on iPhone, 4G LTE, Total Blocking Time on Motorola Moto G Power, 3G connection, Largest Contentful Paint on Chrome Desktop, Total Image Size in Bytes on Chrome Desktop, Total Image Size in Bytes on iPhone, 4G LTE, Total Image Size in Bytes on Motorola Moto G Power, 3G connection, Total HTML Size in Bytes on Chrome Desktop, Total HTML Size in Bytes on iPhone, 4G LTE, Total HTML Size in Bytes on Motorola Moto G Power, 3G connection, Lighthouse Performance Score on Motorola Moto G Power, 3G connection, Lighthouse Performance Score on Chrome Desktop}

_{Calibre: Site dashboard | View this PR | Edit settings | View documentation}

ematipico · 2022-09-20T12:22:33Z

!bench_formatter
!bench_parser
!bench_analyzer

github-actions · 2022-09-20T12:35:26Z

Analyzer Benchmark Results

group                     main                                   pr
-----                     ----                                   --
analyzer/css.js           1.24      2.4±0.10ms     4.8 MB/sec    1.00  1950.0±61.42µs     6.0 MB/sec
analyzer/index.js         1.13      6.4±0.25ms     5.1 MB/sec    1.00      5.7±0.25ms     5.7 MB/sec
analyzer/lint.ts          1.14      3.3±0.16ms    12.5 MB/sec    1.00      2.9±0.11ms    14.3 MB/sec
analyzer/parser.ts        1.19      8.1±0.29ms     6.0 MB/sec    1.00      6.8±0.30ms     7.1 MB/sec
analyzer/router.ts        1.14      5.6±0.22ms    10.9 MB/sec    1.00      4.9±0.21ms    12.5 MB/sec
analyzer/statement.ts     1.19      7.7±0.37ms     4.6 MB/sec    1.00      6.4±0.24ms     5.5 MB/sec
analyzer/typescript.ts    1.18     11.5±0.47ms     4.7 MB/sec    1.00      9.7±0.41ms     5.6 MB/sec

github-actions · 2022-09-20T12:42:13Z

Parser Benchmark Results

group                                 main                                   pr
-----                                 ----                                   --
parser/checker.ts                     1.00    138.5±7.97ms    18.8 MB/sec    1.07    148.7±6.96ms    17.5 MB/sec
parser/compiler.js                    1.01     86.3±4.56ms    12.1 MB/sec    1.00     85.3±3.41ms    12.3 MB/sec
parser/d3.min.js                      1.01     51.8±2.63ms     5.1 MB/sec    1.00     51.4±2.24ms     5.1 MB/sec
parser/dojo.js                        1.09      4.7±0.14ms    14.6 MB/sec    1.00      4.3±0.22ms    16.0 MB/sec
parser/ios.d.ts                       1.00    121.0±5.88ms    15.4 MB/sec    1.09    131.6±7.02ms    14.2 MB/sec
parser/jquery.min.js                  1.00     13.6±0.34ms     6.1 MB/sec    1.02     13.8±0.83ms     6.0 MB/sec
parser/math.js                        1.00     97.7±4.53ms     6.6 MB/sec    1.08    105.7±4.76ms     6.1 MB/sec
parser/parser.ts                      1.05      3.2±0.07ms    15.3 MB/sec    1.00      3.0±0.12ms    16.0 MB/sec
parser/pixi.min.js                    1.00     59.6±1.83ms     7.4 MB/sec    1.08     64.3±2.66ms     6.8 MB/sec
parser/react-dom.production.min.js    1.00     18.9±0.82ms     6.1 MB/sec    1.00     18.9±1.07ms     6.1 MB/sec
parser/react.production.min.js        1.09  1043.4±39.05µs     5.9 MB/sec    1.00   955.3±31.87µs     6.4 MB/sec
parser/router.ts                      1.09      2.8±0.11ms    22.2 MB/sec    1.00      2.5±0.07ms    24.2 MB/sec
parser/tex-chtml-full.js              1.00    133.1±5.81ms     6.8 MB/sec    1.08    143.5±5.55ms     6.4 MB/sec
parser/three.min.js                   1.00     69.6±3.73ms     8.4 MB/sec    1.02     70.7±3.03ms     8.3 MB/sec
parser/vue.global.prod.js             1.03     23.4±0.78ms     5.1 MB/sec    1.00     22.8±0.97ms     5.3 MB/sec

github-actions · 2022-09-20T12:57:58Z

Formatter Benchmark Results

group                                    main                                   pr
-----                                    ----                                   --
formatter/checker.ts                     1.18   600.8±12.45ms     4.3 MB/sec    1.00   509.4±14.16ms     5.1 MB/sec
formatter/compiler.js                    1.14   319.0±10.86ms     3.3 MB/sec    1.00    279.2±7.64ms     3.8 MB/sec
formatter/d3.min.js                      1.21   270.4±12.77ms   992.6 KB/sec    1.00    223.4±7.91ms  1201.2 KB/sec
formatter/dojo.js                        1.13     16.2±0.56ms     4.2 MB/sec    1.00     14.3±0.38ms     4.8 MB/sec
formatter/ios.d.ts                       1.10    342.2±8.24ms     5.5 MB/sec    1.00    310.8±8.04ms     6.0 MB/sec
formatter/jquery.min.js                  1.02     65.8±2.34ms  1285.4 KB/sec    1.00     64.3±2.20ms  1316.4 KB/sec
formatter/math.js                        1.18   532.1±12.71ms  1246.2 KB/sec    1.00    451.2±9.18ms  1469.7 KB/sec
formatter/parser.ts                      1.16     11.0±0.32ms     4.4 MB/sec    1.00      9.4±0.30ms     5.2 MB/sec
formatter/pixi.min.js                    1.17    299.9±9.00ms  1498.5 KB/sec    1.00    256.9±8.78ms  1749.6 KB/sec
formatter/react-dom.production.min.js    1.12     84.8±3.76ms  1390.5 KB/sec    1.00     75.5±2.53ms  1560.4 KB/sec
formatter/react.production.min.js        1.16      4.0±0.14ms  1568.5 KB/sec    1.00      3.5±0.09ms  1817.1 KB/sec
formatter/router.ts                      1.10      8.4±0.27ms     7.3 MB/sec    1.00      7.6±0.26ms     8.1 MB/sec
formatter/tex-chtml-full.js              1.19   689.1±20.88ms  1354.2 KB/sec    1.00   580.1±10.92ms  1608.6 KB/sec
formatter/three.min.js                   1.20   341.4±10.96ms  1760.9 KB/sec    1.00    284.7±8.25ms     2.1 MB/sec
formatter/typescript.js                  1.00       2.1±0.03s     4.5 MB/sec  
formatter/vue.global.prod.js             1.10    111.4±4.73ms  1107.3 KB/sec    1.00    101.7±5.72ms  1213.4 KB/sec

MichaReiser · 2022-09-21T06:55:29Z

formatter/typescript.js 1.00 2.1±0.03s 4.5 MB/sec

Uhm, what happened with typescript?

ematipico · 2022-09-21T09:18:21Z

formatter/typescript.js 1.00 2.1±0.03s 4.5 MB/sec

Uhm, what happened with typescript?

Something went wrong during the download of the file:

 "Status(520, Response[status: 520, status_text: , url: https://www.unpkg.com/[email protected]/lib/typescript.js])"

ematipico · 2022-09-21T09:40:05Z

@rome/staff I updated the description of the issue with some findings around memory consumption.

MichaReiser

I'm not sure if the allocation numbers are representative because they're only measuring how much memory was requested and freed by the system allocator which is expected to be less if you use a custom allocator that re-uses allocations before requesting new pages from the system allocator.

However, you shared some numbers over slack which are promising.

I checked with activity monitory. With jemalloc the memory usage is around 20MB top. main instead is 45MB top

I think the results overall are very promising and I verified them on my local machine. There are a few cases where performance regresses or does not improve, but the numbers greatly improve for all benchmarks traversing the CST.

group                                    jemalloc                               main
-----                                    --------                               ----
analyzer/css.js                          1.00  1026.2±11.93µs    11.3 MB/sec    1.24   1268.7±5.37µs     9.2 MB/sec
analyzer/index.js                        1.00      2.9±0.05ms    11.1 MB/sec    1.16      3.4±0.01ms     9.6 MB/sec
analyzer/lint.ts                         1.00   1541.7±7.90µs    27.0 MB/sec    1.16  1782.6±239.92µs    23.3 MB/sec
analyzer/parser.ts                       1.00      3.6±0.05ms    13.5 MB/sec    1.15      4.2±0.01ms    11.7 MB/sec
analyzer/router.ts                       1.00      2.4±0.02ms    25.4 MB/sec    1.23      3.0±0.01ms    20.7 MB/sec
analyzer/statement.ts                    1.00      3.3±0.06ms    10.9 MB/sec    1.30      4.2±0.01ms     8.4 MB/sec
analyzer/typescript.ts                   1.00      5.1±0.05ms    10.6 MB/sec    1.25      6.4±0.11ms     8.5 MB/sec
formatter/checker.ts                     1.00    220.9±1.81ms    11.8 MB/sec    1.15    253.8±4.23ms    10.2 MB/sec
formatter/compiler.js                    1.00    123.4±3.05ms     8.5 MB/sec    1.13    139.6±0.90ms     7.5 MB/sec
formatter/d3.min.js                      1.00    102.9±0.91ms     2.5 MB/sec    1.10    112.7±1.26ms     2.3 MB/sec
formatter/dojo.js                        1.00      6.7±0.02ms    10.2 MB/sec    1.16      7.8±0.03ms     8.8 MB/sec
formatter/ios.d.ts                       1.00    140.1±0.76ms    13.3 MB/sec    1.17    164.1±2.29ms    11.4 MB/sec
formatter/jquery.min.js                  1.00     28.9±0.34ms     2.9 MB/sec    1.10     31.8±0.85ms     2.6 MB/sec
formatter/math.js                        1.00    207.7±1.46ms     3.1 MB/sec    1.17    243.5±4.20ms     2.7 MB/sec
formatter/parser.ts                      1.00      4.7±0.02ms    10.4 MB/sec    1.13      5.3±0.07ms     9.3 MB/sec
formatter/pixi.min.js                    1.00    115.9±0.86ms     3.8 MB/sec    1.12    130.0±1.78ms     3.4 MB/sec
formatter/react-dom.production.min.js    1.00     34.8±0.13ms     3.3 MB/sec    1.12     39.1±0.38ms     2.9 MB/sec
formatter/react.production.min.js        1.00   1710.6±5.75µs     3.6 MB/sec    1.13   1940.7±8.91µs     3.2 MB/sec
formatter/router.ts                      1.00      3.7±0.02ms    16.8 MB/sec    1.13      4.1±0.05ms    14.9 MB/sec
formatter/tex-chtml-full.js              1.00    267.5±1.42ms     3.4 MB/sec    1.10    295.0±3.07ms     3.1 MB/sec
formatter/three.min.js                   1.00    129.2±0.59ms     4.5 MB/sec    1.11    143.5±0.84ms     4.1 MB/sec
formatter/typescript.js                  1.00   842.4±13.77ms    11.3 MB/sec    1.15   972.7±15.36ms     9.8 MB/sec
formatter/vue.global.prod.js             1.00     44.5±0.48ms     2.7 MB/sec    1.12     49.6±0.84ms     2.4 MB/sec
parser/checker.ts                        1.09     61.2±1.92ms    42.5 MB/sec    1.00     56.1±0.49ms    46.3 MB/sec
parser/compiler.js                       1.00     33.3±0.39ms    31.5 MB/sec    1.06     35.1±0.19ms    29.8 MB/sec
parser/d3.min.js                         1.00     21.6±0.75ms    12.1 MB/sec    1.01     21.7±0.07ms    12.1 MB/sec
parser/dojo.js                           1.00  1879.6±25.53µs    36.5 MB/sec    1.08      2.0±0.01ms    33.8 MB/sec
parser/ios.d.ts                          1.10     51.4±0.91ms    36.3 MB/sec    1.00     46.9±0.72ms    39.8 MB/sec
parser/jquery.min.js                     1.00      5.7±0.09ms    14.5 MB/sec    1.04      5.9±0.01ms    13.9 MB/sec
parser/math.js                           1.05     42.4±1.74ms    15.3 MB/sec    1.00     40.5±1.97ms    16.0 MB/sec
parser/parser.ts                         1.00  1311.6±11.85µs    37.1 MB/sec    1.10  1444.8±12.49µs    33.7 MB/sec
parser/pixi.min.js                       1.05     26.6±0.52ms    16.5 MB/sec    1.00     25.4±0.09ms    17.2 MB/sec
parser/react-dom.production.min.js       1.00      7.9±0.04ms    14.6 MB/sec    1.00      7.9±0.07ms    14.6 MB/sec
parser/react.production.min.js           1.00    386.0±5.42µs    15.9 MB/sec    1.12   433.3±16.39µs    14.2 MB/sec
parser/router.ts                         1.00  1108.6±12.05µs    55.4 MB/sec    1.10  1214.3±23.91µs    50.6 MB/sec
parser/tex-chtml-full.js                 1.10     58.7±1.90ms    15.5 MB/sec    1.00     53.4±0.21ms    17.1 MB/sec
parser/three.min.js                      1.06     30.9±1.11ms    19.0 MB/sec    1.00     29.1±0.06ms    20.2 MB/sec
parser/typescript.js                     1.03    240.5±2.73ms    39.5 MB/sec    1.00   234.5±17.22ms    40.5 MB/sec
parser/vue.global.prod.js                1.00      9.9±0.16ms    12.2 MB/sec    1.00      9.9±0.03ms    12.2 MB/sec

ematipico · 2022-09-21T14:28:37Z

@leops and @xunilrj what do you think? The numbers seem good and so far I could not see any issue locally. Should we make the switch to jemalloc for all operative systems? (exception for windows)

perf: jamalloc

bb39ada

ematipico temporarily deployed to netlify-playground September 16, 2022 09:05 Inactive

ematipico marked this pull request as ready for review September 16, 2022 09:44

ematipico requested a review from leops as a code owner September 16, 2022 09:44

ematipico requested a review from a team September 16, 2022 09:44

ematipico had a problem deploying to netlify-playground September 20, 2022 12:14 Failure

chore: enable allocator in linux

3e557de

ematipico force-pushed the perf/jamalloc branch from 7806668 to 3e557de Compare September 20, 2022 12:22

ematipico temporarily deployed to netlify-playground September 20, 2022 12:22 Inactive

MichaReiser approved these changes Sep 21, 2022

View reviewed changes

ematipico added this to the 0.10.0 milestone Sep 21, 2022

ematipico added the A-Core Area: core label Sep 21, 2022

ematipico changed the title ~~perf(rome_cli): replace global allocator with jemalloc for macos~~ perf(rome_cli): replace global allocator with jemalloc Sep 21, 2022

ematipico added the A-CLI Area: CLI label Sep 21, 2022

leops approved these changes Sep 21, 2022

View reviewed changes

ematipico merged commit 806de56 into main Sep 21, 2022

ematipico deleted the perf/jamalloc branch September 21, 2022 14:48

MichaReiser mentioned this pull request Feb 11, 2023

perf: Use custom allocator astral-sh/ruff#2768

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

perf(rome_cli): replace global allocator with jemalloc #3237

perf(rome_cli): replace global allocator with jemalloc #3237

ematipico commented Sep 16, 2022 •

edited

Loading

netlify bot commented Sep 16, 2022 •

edited

Loading

github-actions bot commented Sep 16, 2022

github-actions bot commented Sep 16, 2022 •

edited

Loading

MichaReiser commented Sep 16, 2022

MichaReiser commented Sep 16, 2022

IWANABETHATGUY commented Sep 16, 2022 •

edited

Loading

MichaReiser commented Sep 16, 2022 •

edited

Loading

IWANABETHATGUY commented Sep 16, 2022

leops commented Sep 16, 2022

MichaReiser commented Sep 16, 2022

ematipico commented Sep 16, 2022 •

edited

Loading

leops commented Sep 16, 2022

MichaReiser commented Sep 16, 2022

ematipico commented Sep 20, 2022

MichaReiser commented Sep 20, 2022

calibre-analytics bot commented Sep 20, 2022 •

edited

Loading

Browser previews

Most significant changes

ematipico commented Sep 20, 2022

github-actions bot commented Sep 20, 2022

github-actions bot commented Sep 20, 2022

github-actions bot commented Sep 20, 2022

MichaReiser commented Sep 21, 2022

ematipico commented Sep 21, 2022 •

edited

Loading

ematipico commented Sep 21, 2022

MichaReiser left a comment •

edited

Loading

ematipico commented Sep 21, 2022 •

edited

Loading

perf(rome_cli): replace global allocator with jemalloc #3237

perf(rome_cli): replace global allocator with jemalloc #3237

Conversation

ematipico commented Sep 16, 2022 • edited Loading

Summary

Test Plan

parser

formatter

analyzer

Memory consumption

jemalloc

main

netlify bot commented Sep 16, 2022 • edited Loading

✅ Deploy Preview for rometools canceled.

github-actions bot commented Sep 16, 2022

Parser conformance results on ubuntu-latest

js/262

jsx/babel

symbols/microsoft

ts/babel

ts/microsoft

github-actions bot commented Sep 16, 2022 • edited Loading

MichaReiser commented Sep 16, 2022

MichaReiser commented Sep 16, 2022

IWANABETHATGUY commented Sep 16, 2022 • edited Loading

MichaReiser commented Sep 16, 2022 • edited Loading

IWANABETHATGUY commented Sep 16, 2022

leops commented Sep 16, 2022

MichaReiser commented Sep 16, 2022

ematipico commented Sep 16, 2022 • edited Loading

leops commented Sep 16, 2022

MichaReiser commented Sep 16, 2022

ematipico commented Sep 20, 2022

MichaReiser commented Sep 20, 2022

calibre-analytics bot commented Sep 20, 2022 • edited Loading

1 page tested

Browser previews

Most significant changes

ematipico commented Sep 20, 2022

github-actions bot commented Sep 20, 2022

Analyzer Benchmark Results

github-actions bot commented Sep 20, 2022

Parser Benchmark Results

github-actions bot commented Sep 20, 2022

Formatter Benchmark Results

MichaReiser commented Sep 21, 2022

ematipico commented Sep 21, 2022 • edited Loading

ematipico commented Sep 21, 2022

MichaReiser left a comment • edited Loading

Choose a reason for hiding this comment

ematipico commented Sep 21, 2022 • edited Loading

ematipico commented Sep 16, 2022 •

edited

Loading

`jemalloc`

`main`

netlify bot commented Sep 16, 2022 •

edited

Loading

github-actions bot commented Sep 16, 2022 •

edited

Loading

IWANABETHATGUY commented Sep 16, 2022 •

edited

Loading

MichaReiser commented Sep 16, 2022 •

edited

Loading

ematipico commented Sep 16, 2022 •

edited

Loading

calibre-analytics bot commented Sep 20, 2022 •

edited

Loading

ematipico commented Sep 21, 2022 •

edited

Loading

MichaReiser left a comment •

edited

Loading

ematipico commented Sep 21, 2022 •

edited

Loading