perf(ts): reuse Reader whilst parsing records #1212

AaronO · 2024-08-12T08:02:23Z

Reduces McapStreamReader heap usage by ~25% and boosts throughput by ~30%.

This is both a refactor and perf boost, with more room for improvement, key changes:

Removes StreamBuffer, hoisted into McapStreamReader
Reuses Reader and DataView class across parse calls, only resetting them when necessary (e.g: append in McapStreamReader)
Splits parseRecord into small scoped parsing functions, this itself is perf neutral (slightly positive) but facilitates future monomorphic fast paths
Moves offsets tracking into Reader which is cleaner and faster

Before

McapStreamReader
        3.48±0.03 op/s  Heap Used: 49.56±12.75 MB/op    Heap Total: 41.47±11.83 MB/op   ArrayBuffers: 112.95±6.87 MB/op
McapIndexedReader
        2.15±0.02 op/s  Heap Used: 70.02±2.84 MB/op     Heap Total: 58.34±3.36 MB/op    ArrayBuffers: 17.86±0.76 MB/op
McapIndexedReader_reverse
        2.18±0.01 op/s  Heap Used: 59.92±2.86 MB/op     Heap Total: 39.81±1.00 MB/op    ArrayBuffers: 14.58±1.42 MB/op

After

McapStreamReader
        4.47±0.08 op/s  Heap Used: 42.35±2.23 MB/op     Heap Total: 32.93±3.76 MB/op    ArrayBuffers: 105.93±12.19 MB/op
McapIndexedReader
        2.38±0.02 op/s  Heap Used: 72.00±1.70 MB/op     Heap Total: 55.12±2.51 MB/op    ArrayBuffers: 17.86±1.85 MB/op
McapIndexedReader_reverse
        2.38±0.02 op/s  Heap Used: 63.41±1.55 MB/op     Heap Total: 39.33±0.53 MB/op    ArrayBuffers: 18.40±1.60 MB/op

Followups

Simplify parsing code further after exploring monomorphic callpaths
Can further improve DataView/Reader churn in indexed readers

This is both a refactor and perf boost, with more room for improvement, key changes: - Removes StreamBuffer, hoisted into McapStreamReader - Reuses Reader and DataView class across parse calls, only resetting them when necessary (e.g: append in McapStreamData) - Splits `parseRecord` into small scoped parsing functions, this itself is perf neutral (slightly positive) but facilitates future monomorphic fast paths - Moves offsets tracking into Reader which is cleaner and faster

jtbandes

Seems like a nice refactor overall! Most of it makes sense to me, I just left a few questions and cleanup suggestions. I do find the duplication of reader.offset = startOffset + recordLength; a bit repetitive and I wonder if there is a way that could be cleaned up...

typescript/core/src/Reader.ts

typescript/core/src/McapIndexedReader.ts

jtbandes · 2024-08-13T21:29:25Z

typescript/core/src/StreamBuffer.test.ts

I think it's a bit of a shame to lose these tests, but I guess the thinking is that the StreamReader is already pretty well tested?

Would it be covered by all the conformance testing? These tests seem really specific to this now deleted class so seems ok?

@defunctzombie I think this is exercised/covered by comformance testing since at least one of those tests reads/streams byte-by-byte. Exercising the appendOrShift.

Now that it's scoped out and #appendOrShift was the only important part, I think it could be fair to split that out into a function and test in isolation.

typescript/core/src/McapStreamReader.ts

typescript/core/src/parse.ts

jtbandes · 2024-08-13T22:24:37Z

typescript/core/src/parse.ts

+  // eslint-disable-next-line @foxglove/no-boolean-parameters
+  validateCrcs: boolean,


same comment about boolean parameters applies here

typescript/core/src/Reader.ts

defunctzombie · 2024-08-14T00:42:19Z

typescript/core/src/StreamBuffer.test.ts

Would it be covered by all the conformance testing? These tests seem really specific to this now deleted class so seems ok?

achim-k · 2024-08-14T10:52:14Z

typescript/core/src/McapIndexedReader.ts

+      const headerReader = new Reader(
+        new DataView(headerRecord.buffer, headerRecord.byteOffset, headerRecord.byteLength),
+      );
+      const headerResult = parseRecord(headerReader, true);


Suggested change

const headerResult = parseRecord(headerReader, true);

const validateCrcs = true;

const headerResult = parseRecord(headerReader, validateCrcs);

Nit, makes it a bit more readable

typescript/core/src/parse.ts

achim-k · 2024-08-14T12:01:50Z

typescript/core/src/parse.ts

+}
+
+function parseSchema(reader: Reader, recordLength: number): TypedMcapRecord {
+  const start = reader.offset;


nit: Rename to startOffset to stay consistent with other parse* functions

typescript/core/src/parse.ts

jtbandes

Looks good, but before merging please add a comment to explain the disabling of the no-boolean-parameters lint rule (see #1212 (comment) and also consider this suggestion: #1212 (comment))

### Changelog Performance improvements (#1212)

This reverts commit 4fcdd33.

Reverts #1212 This broke streaming in Foxglove app. Until we can investigate we need to put the library back to a working state.

Reduces `McapStreamReader` heap usage by ~25% and boosts throughput by ~30%. This is both a refactor and perf boost, with more room for improvement, key changes: - Removes StreamBuffer, hoisted into McapStreamReader - Reuses Reader and DataView class across parse calls, only resetting them when necessary (e.g: append in McapStreamReader) - Splits `parseRecord` into small scoped parsing functions, this itself is perf neutral (slightly positive) but facilitates future monomorphic fast paths - Moves offsets tracking into Reader which is cleaner and faster ### Before ``` McapStreamReader 3.48±0.03 op/s Heap Used: 49.56±12.75 MB/op Heap Total: 41.47±11.83 MB/op ArrayBuffers: 112.95±6.87 MB/op McapIndexedReader 2.15±0.02 op/s Heap Used: 70.02±2.84 MB/op Heap Total: 58.34±3.36 MB/op ArrayBuffers: 17.86±0.76 MB/op McapIndexedReader_reverse 2.18±0.01 op/s Heap Used: 59.92±2.86 MB/op Heap Total: 39.81±1.00 MB/op ArrayBuffers: 14.58±1.42 MB/op ``` ### After ``` McapStreamReader 4.47±0.08 op/s Heap Used: 42.35±2.23 MB/op Heap Total: 32.93±3.76 MB/op ArrayBuffers: 105.93±12.19 MB/op McapIndexedReader 2.38±0.02 op/s Heap Used: 72.00±1.70 MB/op Heap Total: 55.12±2.51 MB/op ArrayBuffers: 17.86±1.85 MB/op McapIndexedReader_reverse 2.38±0.02 op/s Heap Used: 63.41±1.55 MB/op Heap Total: 39.33±0.53 MB/op ArrayBuffers: 18.40±1.60 MB/op ``` ### Followups - Simplify parsing code further after exploring monomorphic callpaths - Can further improve DataView/Reader churn in indexed readers

### Changelog Typescript: Performance improvements for readers ### Docs None ### Description Reintroduces the changes made in #1212, which have been reverted in #1227, with an additional fix & test (separate commit) for the buffer append logic. The original code of #1212 removed the `streamBuffer` object and implemented the buffer append ing logic directly in the `McapStreamReader` class. The append logic had a bug which caused existing data to be partially overridden by new data. From #1212: > Reduces `McapStreamReader` heap usage by ~25% and boosts throughput by ~30%. > > This is both a refactor and perf boost, with more room for improvement, key changes: > - Removes StreamBuffer, hoisted into McapStreamReader > - Reuses Reader and DataView class across parse calls, only resetting them when necessary (e.g: append in McapStreamReader) > - Splits `parseRecord` into small scoped parsing functions, this itself is perf neutral (slightly positive) but facilitates future monomorphic fast paths > - Moves offsets tracking into Reader which is cleaner and faster > > ### Before > > ``` > McapStreamReader > 3.48±0.03 op/s Heap Used: 49.56±12.75 MB/op Heap Total: 41.47±11.83 MB/op ArrayBuffers: 112.95±6.87 MB/op > McapIndexedReader > 2.15±0.02 op/s Heap Used: 70.02±2.84 MB/op Heap Total: 58.34±3.36 MB/op ArrayBuffers: 17.86±0.76 MB/op > McapIndexedReader_reverse > 2.18±0.01 op/s Heap Used: 59.92±2.86 MB/op Heap Total: 39.81±1.00 MB/op ArrayBuffers: 14.58±1.42 MB/op > ``` > > ### After > > ``` > McapStreamReader > 4.47±0.08 op/s Heap Used: 42.35±2.23 MB/op Heap Total: 32.93±3.76 MB/op ArrayBuffers: 105.93±12.19 MB/op > McapIndexedReader > 2.38±0.02 op/s Heap Used: 72.00±1.70 MB/op Heap Total: 55.12±2.51 MB/op ArrayBuffers: 17.86±1.85 MB/op > McapIndexedReader_reverse > 2.38±0.02 op/s Heap Used: 63.41±1.55 MB/op Heap Total: 39.33±0.53 MB/op ArrayBuffers: 18.40±1.60 MB/op > ``` --------- Co-authored-by: Aaron O'Mullan <[email protected]>

AaronO added 2 commits August 12, 2024 17:00

fix: isKnownOpcode unused import

e3eccc0

github-actions bot deployed to mcap (Preview) August 12, 2024 08:11 View deployment

lint

7897fc3

github-actions bot deployed to mcap (Preview) August 12, 2024 17:36 View deployment

jtbandes self-requested a review August 12, 2024 20:12

AaronO added 4 commits August 13, 2024 12:07

cleanup: #buffer.consume()

b2e5fe9

simpler Reader.string()

3a8f09f

fix parseRecord ensure advanced by record length accounting for padding

e471521

Remove Reader.rewind() given single call site

68ca507

github-actions bot deployed to mcap (Preview) August 13, 2024 04:23 View deployment

github-actions bot deployed to mcap (Preview) August 13, 2024 04:25 View deployment

parse fix attachment crc

4aadbc5

github-actions bot deployed to mcap (Preview) August 13, 2024 04:29 View deployment

jtbandes reviewed Aug 13, 2024

View reviewed changes

defunctzombie requested a review from achim-k August 14, 2024 00:26

defunctzombie reviewed Aug 14, 2024

View reviewed changes

comments and nits

6e54927

github-actions bot deployed to mcap (Preview) August 14, 2024 03:30 View deployment

achim-k reviewed Aug 14, 2024

View reviewed changes

jtbandes approved these changes Aug 14, 2024

View reviewed changes

add bool arg notes

4689528

github-actions bot deployed to mcap (Preview) August 15, 2024 00:47 View deployment

AaronO merged commit 4fcdd33 into main Aug 15, 2024
23 checks passed

AaronO deleted the perf/parse-record-reader-reuse branch August 15, 2024 00:49

achim-k mentioned this pull request Aug 30, 2024

Bump @mcap/core version #1226

Merged

achim-k added a commit that referenced this pull request Aug 30, 2024

Bump @mcap/core version (#1226)

462981d

### Changelog Performance improvements (#1212)

defunctzombie added a commit that referenced this pull request Aug 30, 2024

Revert "perf(ts): reuse Reader whilst parsing records (#1212)"

825aadb

This reverts commit 4fcdd33.

defunctzombie mentioned this pull request Aug 30, 2024

Revert "perf(ts): reuse Reader whilst parsing records" #1227

Merged

defunctzombie added a commit that referenced this pull request Aug 30, 2024

Revert "perf(ts): reuse Reader whilst parsing records" (#1227)

ed894e7

Reverts #1212 This broke streaming in Foxglove app. Until we can investigate we need to put the library back to a working state.

achim-k mentioned this pull request Sep 12, 2024

TypeScript: Reuse Reader for stream reader #1236

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

perf(ts): reuse Reader whilst parsing records #1212

perf(ts): reuse Reader whilst parsing records #1212

AaronO commented Aug 12, 2024 •

edited

Loading

jtbandes left a comment

jtbandes Aug 13, 2024

defunctzombie Aug 14, 2024

AaronO Aug 14, 2024

jtbandes Aug 13, 2024

defunctzombie Aug 14, 2024

achim-k Aug 14, 2024

achim-k Aug 14, 2024

jtbandes left a comment

		// eslint-disable-next-line @foxglove/no-boolean-parameters
		validateCrcs: boolean,

	const headerResult = parseRecord(headerReader, true);
	const validateCrcs = true;
	const headerResult = parseRecord(headerReader, validateCrcs);

perf(ts): reuse Reader whilst parsing records #1212

perf(ts): reuse Reader whilst parsing records #1212

Conversation

AaronO commented Aug 12, 2024 • edited Loading

Before

After

Followups

jtbandes left a comment

Choose a reason for hiding this comment

jtbandes Aug 13, 2024

Choose a reason for hiding this comment

defunctzombie Aug 14, 2024

Choose a reason for hiding this comment

AaronO Aug 14, 2024

Choose a reason for hiding this comment

jtbandes Aug 13, 2024

Choose a reason for hiding this comment

defunctzombie Aug 14, 2024

Choose a reason for hiding this comment

achim-k Aug 14, 2024

Choose a reason for hiding this comment

achim-k Aug 14, 2024

Choose a reason for hiding this comment

jtbandes left a comment

Choose a reason for hiding this comment

AaronO commented Aug 12, 2024 •

edited

Loading