trace: optimize id parsing and string functions#7157
Closed
bboreham wants to merge 4 commits intoopen-telemetry:mainfrom
Closed
trace: optimize id parsing and string functions#7157bboreham wants to merge 4 commits intoopen-telemetry:mainfrom
bboreham wants to merge 4 commits intoopen-telemetry:mainfrom
Conversation
With specialized routines, we can avoid the allocation of hex.DecodeString since we know the structure of the IDs. We can use `==` instead of bytes.Equal for arrays. From the Go [spec]: > Array types are comparable if their array element types are comparable. Two > array values are equal if their corresponding element values are equal. The > elements are compared in ascending index order, and comparison stops as soon > as two element values differ (or all elements have been compared). [spec]: https://go.dev/ref/spec#Comparison_operators To generate: ```sh mkdir private cd sdk go test -run=xxxxMatchNothingxxxx -bench=. -count=10 go.opentelemetry.io/otel/sdk/trace -timeout=30m | tee ../private/base.txt go test -run=xxxxMatchNothingxxxx -bench=. -count=10 go.opentelemetry.io/otel/sdk/trace -timeout=30m | tee ../private/new.txt benchstat ../private/base.txt ../private/new.txt ``` Results: ``` goos: darwin goarch: arm64 pkg: go.opentelemetry.io/otel/sdk/trace cpu: Apple M2 Max │ ../private/base.txt │ ../private/new.txt │ │ sec/op │ sec/op vs base │ Truncate/Unlimited-12 0.2086n ± 1% 0.2140n ± 3% +2.59% (p=0.017 n=10) Truncate/Zero-12 0.3048n ± 0% 0.3070n ± 1% +0.71% (p=0.014 n=10) Truncate/Short-12 0.2083n ± 2% 0.2097n ± 1% ~ (p=0.148 n=10) Truncate/ASCII-12 0.6870n ± 0% 0.6855n ± 0% ~ (p=0.493 n=10) Truncate/ValidUTF-8-12 1.298n ± 0% 1.302n ± 1% +0.31% (p=0.003 n=10) Truncate/InvalidUTF-8-12 9.457n ± 0% 9.420n ± 1% ~ (p=0.529 n=10) Truncate/MixedUTF-8-12 17.30n ± 1% 17.29n ± 0% ~ (p=0.359 n=10) RecordingSpanSetAttributes/WithLimit/false-12 2.055µ ± 1% 2.082µ ± 9% +1.29% (p=0.014 n=10) RecordingSpanSetAttributes/WithLimit/true-12 4.368µ ± 0% 4.364µ ± 0% -0.08% (p=0.049 n=10) SpanEnd-12 72.57n ± 17% 73.75n ± 16% ~ (p=0.853 n=10) TraceStart/with_a_simple_span-12 320.1n ± 10% 314.6n ± 10% ~ (p=0.165 n=10) TraceStart/with_several_links-12 432.7n ± 2% 429.4n ± 1% ~ (p=0.063 n=10) TraceStart/with_attributes-12 477.3n ± 1% 468.1n ± 6% -1.94% (p=0.005 n=10) SpanLimits/AttributeValueLengthLimit-12 4.401µ ± 1% 4.439µ ± 2% ~ (p=0.089 n=10) SpanLimits/AttributeCountLimit-12 4.125µ ± 1% 4.151µ ± 1% +0.62% (p=0.014 n=10) SpanLimits/EventCountLimit-12 3.900µ ± 2% 3.935µ ± 1% +0.88% (p=0.023 n=10) SpanLimits/LinkCountLimit-12 3.870µ ± 2% 3.901µ ± 1% ~ (p=0.148 n=10) SpanLimits/AttributePerEventCountLimit-12 4.212µ ± 1% 4.243µ ± 1% +0.75% (p=0.008 n=10) SpanLimits/AttributePerLinkCountLimit-12 4.200µ ± 1% 4.224µ ± 0% +0.57% (p=0.041 n=10) SpanSetAttributesOverCapacity-12 1.661µ ± 1% 1.653µ ± 0% -0.48% (p=0.049 n=10) StartEndSpan/AlwaysSample-12 317.9n ± 0% 316.5n ± 0% -0.44% (p=0.007 n=10) StartEndSpan/NeverSample-12 152.3n ± 0% 152.0n ± 0% -0.23% (p=0.005 n=10) SpanWithAttributes_4/AlwaysSample-12 527.2n ± 0% 532.4n ± 1% +1.00% (p=0.000 n=10) SpanWithAttributes_4/NeverSample-12 240.6n ± 0% 241.6n ± 0% +0.46% (p=0.000 n=10) SpanWithAttributes_8/AlwaysSample-12 704.5n ± 0% 718.3n ± 1% +1.97% (p=0.000 n=10) SpanWithAttributes_8/NeverSample-12 325.0n ± 0% 327.2n ± 1% +0.68% (p=0.000 n=10) SpanWithAttributes_all/AlwaysSample-12 576.8n ± 0% 584.7n ± 1% +1.37% (p=0.000 n=10) SpanWithAttributes_all/NeverSample-12 264.6n ± 1% 263.3n ± 0% -0.47% (p=0.045 n=10) SpanWithAttributes_all_2x/AlwaysSample-12 818.6n ± 1% 834.9n ± 0% +1.98% (p=0.000 n=10) SpanWithAttributes_all_2x/NeverSample-12 378.3n ± 0% 382.9n ± 1% +1.23% (p=0.000 n=10) SpanWithEvents_4/AlwaysSample-12 715.1n ± 1% 721.1n ± 0% +0.83% (p=0.003 n=10) SpanWithEvents_4/NeverSample-12 156.1n ± 1% 155.1n ± 1% -0.64% (p=0.002 n=10) SpanWithEvents_8/AlwaysSample-12 1.098µ ± 0% 1.104µ ± 0% +0.55% (p=0.000 n=10) SpanWithEvents_8/NeverSample-12 158.8n ± 0% 158.6n ± 1% ~ (p=0.288 n=10) SpanWithEvents_WithStackTrace/AlwaysSample-12 438.8n ± 0% 438.5n ± 0% ~ (p=0.868 n=10) SpanWithEvents_WithStackTrace/NeverSample-12 168.2n ± 1% 167.4n ± 1% -0.48% (p=0.014 n=10) SpanWithEvents_WithTimestamp/AlwaysSample-12 430.6n ± 0% 432.9n ± 0% +0.53% (p=0.001 n=10) SpanWithEvents_WithTimestamp/NeverSample-12 193.7n ± 0% 190.1n ± 1% -1.91% (p=0.000 n=10) TraceID_DotString-12 42.37n ± 0% 24.80n ± 0% -41.45% (p=0.000 n=10) SpanID_DotString-12 31.30n ± 0% 17.22n ± 0% -44.98% (p=0.000 n=10) SpanProcessorOnEnd/batch:_10,_spans:_10-12 163.3n ± 0% 163.4n ± 0% ~ (p=0.120 n=10) SpanProcessorOnEnd/batch:_10,_spans:_100-12 1.639µ ± 0% 1.635µ ± 0% -0.27% (p=0.000 n=10) SpanProcessorOnEnd/batch:_100,_spans:_10-12 163.3n ± 0% 163.2n ± 0% ~ (p=0.115 n=10) SpanProcessorOnEnd/batch:_100,_spans:_100-12 1.636µ ± 0% 1.635µ ± 1% ~ (p=0.509 n=10) SpanProcessorVerboseLogging-12 6.769µ ± 2% 6.600µ ± 2% -2.49% (p=0.030 n=10) geomean 221.4n 216.4n -2.29% │ ../private/base.txt │ ../private/new.txt │ │ B/op │ B/op vs base │ Truncate/Unlimited-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ Truncate/Zero-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ Truncate/Short-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ Truncate/ASCII-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ Truncate/ValidUTF-8-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ Truncate/InvalidUTF-8-12 16.00 ± 0% 16.00 ± 0% ~ (p=1.000 n=10) ¹ Truncate/MixedUTF-8-12 32.00 ± 0% 32.00 ± 0% ~ (p=1.000 n=10) ¹ RecordingSpanSetAttributes/WithLimit/false-12 6.891Ki ± 0% 6.891Ki ± 0% ~ (p=1.000 n=10) ¹ RecordingSpanSetAttributes/WithLimit/true-12 7.023Ki ± 0% 7.023Ki ± 0% ~ (p=1.000 n=10) ¹ SpanEnd-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ TraceStart/with_a_simple_span-12 528.0 ± 0% 528.0 ± 0% ~ (p=1.000 n=10) ¹ TraceStart/with_several_links-12 704.0 ± 0% 704.0 ± 0% ~ (p=1.000 n=10) ¹ TraceStart/with_attributes-12 784.0 ± 0% 784.0 ± 0% ~ (p=1.000 n=10) ¹ SpanLimits/AttributeValueLengthLimit-12 10.56Ki ± 0% 10.56Ki ± 0% ~ (p=1.000 n=10) ¹ SpanLimits/AttributeCountLimit-12 9.844Ki ± 0% 9.844Ki ± 0% ~ (p=1.000 n=10) ¹ SpanLimits/EventCountLimit-12 9.422Ki ± 0% 9.422Ki ± 0% ~ (p=1.000 n=10) ¹ SpanLimits/LinkCountLimit-12 9.031Ki ± 0% 9.031Ki ± 0% ~ (p=1.000 n=10) ¹ SpanLimits/AttributePerEventCountLimit-12 10.47Ki ± 0% 10.47Ki ± 0% ~ (p=1.000 n=10) ¹ SpanLimits/AttributePerLinkCountLimit-12 10.47Ki ± 0% 10.47Ki ± 0% ~ (p=1.000 n=10) ¹ SpanSetAttributesOverCapacity-12 592.0 ± 0% 592.0 ± 0% ~ (p=1.000 n=10) ¹ StartEndSpan/AlwaysSample-12 528.0 ± 0% 528.0 ± 0% ~ (p=1.000 n=10) ¹ StartEndSpan/NeverSample-12 144.0 ± 0% 144.0 ± 0% ~ (p=1.000 n=10) ¹ SpanWithAttributes_4/AlwaysSample-12 1.016Ki ± 0% 1.016Ki ± 0% ~ (p=1.000 n=10) ¹ SpanWithAttributes_4/NeverSample-12 400.0 ± 0% 400.0 ± 0% ~ (p=1.000 n=10) ¹ SpanWithAttributes_8/AlwaysSample-12 1.516Ki ± 0% 1.516Ki ± 0% ~ (p=1.000 n=10) ¹ SpanWithAttributes_8/NeverSample-12 656.0 ± 0% 656.0 ± 0% ~ (p=1.000 n=10) ¹ SpanWithAttributes_all/AlwaysSample-12 1.141Ki ± 0% 1.141Ki ± 0% ~ (p=1.000 n=10) ¹ SpanWithAttributes_all/NeverSample-12 464.0 ± 0% 464.0 ± 0% ~ (p=1.000 n=10) ¹ SpanWithAttributes_all_2x/AlwaysSample-12 1.891Ki ± 0% 1.891Ki ± 0% ~ (p=1.000 n=10) ¹ SpanWithAttributes_all_2x/NeverSample-12 848.0 ± 0% 848.0 ± 0% ~ (p=1.000 n=10) ¹ SpanWithEvents_4/AlwaysSample-12 1.016Ki ± 0% 1.016Ki ± 0% ~ (p=1.000 n=10) ¹ SpanWithEvents_4/NeverSample-12 144.0 ± 0% 144.0 ± 0% ~ (p=1.000 n=10) ¹ SpanWithEvents_8/AlwaysSample-12 1.641Ki ± 0% 1.641Ki ± 0% ~ (p=1.000 n=10) ¹ SpanWithEvents_8/NeverSample-12 144.0 ± 0% 144.0 ± 0% ~ (p=1.000 n=10) ¹ SpanWithEvents_WithStackTrace/AlwaysSample-12 624.0 ± 0% 624.0 ± 0% ~ (p=1.000 n=10) ¹ SpanWithEvents_WithStackTrace/NeverSample-12 160.0 ± 0% 160.0 ± 0% ~ (p=1.000 n=10) ¹ SpanWithEvents_WithTimestamp/AlwaysSample-12 648.0 ± 0% 648.0 ± 0% ~ (p=1.000 n=10) ¹ SpanWithEvents_WithTimestamp/NeverSample-12 184.0 ± 0% 184.0 ± 0% ~ (p=1.000 n=10) ¹ SpanProcessorOnEnd/batch:_10,_spans:_10-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ SpanProcessorOnEnd/batch:_10,_spans:_100-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ SpanProcessorOnEnd/batch:_100,_spans:_10-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ SpanProcessorOnEnd/batch:_100,_spans:_100-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ SpanProcessorVerboseLogging-12 9.547Ki ± 0% 9.547Ki ± 0% ~ (p=1.000 n=10) ¹ geomean ² +0.00% ² ¹ all samples are equal ² summaries must be >0 to compute geomean │ ../private/base.txt │ ../private/new.txt │ │ allocs/op │ allocs/op vs base │ Truncate/Unlimited-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ Truncate/Zero-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ Truncate/Short-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ Truncate/ASCII-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ Truncate/ValidUTF-8-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ Truncate/InvalidUTF-8-12 1.000 ± 0% 1.000 ± 0% ~ (p=1.000 n=10) ¹ Truncate/MixedUTF-8-12 1.000 ± 0% 1.000 ± 0% ~ (p=1.000 n=10) ¹ RecordingSpanSetAttributes/WithLimit/false-12 3.000 ± 0% 3.000 ± 0% ~ (p=1.000 n=10) ¹ RecordingSpanSetAttributes/WithLimit/true-12 10.00 ± 0% 10.00 ± 0% ~ (p=1.000 n=10) ¹ SpanEnd-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ TraceStart/with_a_simple_span-12 2.000 ± 0% 2.000 ± 0% ~ (p=1.000 n=10) ¹ TraceStart/with_several_links-12 3.000 ± 0% 3.000 ± 0% ~ (p=1.000 n=10) ¹ TraceStart/with_attributes-12 4.000 ± 0% 4.000 ± 0% ~ (p=1.000 n=10) ¹ SpanLimits/AttributeValueLengthLimit-12 41.00 ± 0% 41.00 ± 0% ~ (p=1.000 n=10) ¹ SpanLimits/AttributeCountLimit-12 38.00 ± 0% 38.00 ± 0% ~ (p=1.000 n=10) ¹ SpanLimits/EventCountLimit-12 35.00 ± 0% 35.00 ± 0% ~ (p=1.000 n=10) ¹ SpanLimits/LinkCountLimit-12 35.00 ± 0% 35.00 ± 0% ~ (p=1.000 n=10) ¹ SpanLimits/AttributePerEventCountLimit-12 38.00 ± 0% 38.00 ± 0% ~ (p=1.000 n=10) ¹ SpanLimits/AttributePerLinkCountLimit-12 38.00 ± 0% 38.00 ± 0% ~ (p=1.000 n=10) ¹ SpanSetAttributesOverCapacity-12 3.000 ± 0% 3.000 ± 0% ~ (p=1.000 n=10) ¹ StartEndSpan/AlwaysSample-12 2.000 ± 0% 2.000 ± 0% ~ (p=1.000 n=10) ¹ StartEndSpan/NeverSample-12 2.000 ± 0% 2.000 ± 0% ~ (p=1.000 n=10) ¹ SpanWithAttributes_4/AlwaysSample-12 4.000 ± 0% 4.000 ± 0% ~ (p=1.000 n=10) ¹ SpanWithAttributes_4/NeverSample-12 3.000 ± 0% 3.000 ± 0% ~ (p=1.000 n=10) ¹ SpanWithAttributes_8/AlwaysSample-12 4.000 ± 0% 4.000 ± 0% ~ (p=1.000 n=10) ¹ SpanWithAttributes_8/NeverSample-12 3.000 ± 0% 3.000 ± 0% ~ (p=1.000 n=10) ¹ SpanWithAttributes_all/AlwaysSample-12 4.000 ± 0% 4.000 ± 0% ~ (p=1.000 n=10) ¹ SpanWithAttributes_all/NeverSample-12 3.000 ± 0% 3.000 ± 0% ~ (p=1.000 n=10) ¹ SpanWithAttributes_all_2x/AlwaysSample-12 4.000 ± 0% 4.000 ± 0% ~ (p=1.000 n=10) ¹ SpanWithAttributes_all_2x/NeverSample-12 3.000 ± 0% 3.000 ± 0% ~ (p=1.000 n=10) ¹ SpanWithEvents_4/AlwaysSample-12 5.000 ± 0% 5.000 ± 0% ~ (p=1.000 n=10) ¹ SpanWithEvents_4/NeverSample-12 2.000 ± 0% 2.000 ± 0% ~ (p=1.000 n=10) ¹ SpanWithEvents_8/AlwaysSample-12 6.000 ± 0% 6.000 ± 0% ~ (p=1.000 n=10) ¹ SpanWithEvents_8/NeverSample-12 2.000 ± 0% 2.000 ± 0% ~ (p=1.000 n=10) ¹ SpanWithEvents_WithStackTrace/AlwaysSample-12 4.000 ± 0% 4.000 ± 0% ~ (p=1.000 n=10) ¹ SpanWithEvents_WithStackTrace/NeverSample-12 3.000 ± 0% 3.000 ± 0% ~ (p=1.000 n=10) ¹ SpanWithEvents_WithTimestamp/AlwaysSample-12 5.000 ± 0% 5.000 ± 0% ~ (p=1.000 n=10) ¹ SpanWithEvents_WithTimestamp/NeverSample-12 4.000 ± 0% 4.000 ± 0% ~ (p=1.000 n=10) ¹ SpanProcessorOnEnd/batch:_10,_spans:_10-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ SpanProcessorOnEnd/batch:_10,_spans:_100-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ SpanProcessorOnEnd/batch:_100,_spans:_10-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ SpanProcessorOnEnd/batch:_100,_spans:_100-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ SpanProcessorVerboseLogging-12 35.00 ± 0% 35.00 ± 0% ~ (p=1.000 n=10) ¹ geomean ² +0.00% ² ¹ all samples are equal ² summaries must be >0 to compute geomean ``` ``` goos: darwin goarch: arm64 pkg: go.opentelemetry.io/otel/sdk/trace cpu: Apple M2 Max │ ../private/base_hex.txt │ ../private/new_hex.txt │ │ sec/op │ sec/op vs base │ TraceIDFromHex-12 56.47n ± 0% 15.96n ± 0% -71.74% (p=0.000 n=10) SpanIDFromHex-12 34.680n ± 0% 8.742n ± 1% -74.79% (p=0.000 n=10) geomean 44.26n 11.81n -73.31% │ ../private/base_hex.txt │ ../private/new_hex.txt │ │ B/op │ B/op vs base │ TraceIDFromHex-12 16.00 ± 0% 0.00 ± 0% -100.00% (p=0.000 n=10) SpanIDFromHex-12 8.000 ± 0% 0.000 ± 0% -100.00% (p=0.000 n=10) geomean 11.31 ? ¹ ² ¹ summaries must be >0 to compute geomean ² ratios must be >0 to compute geomean │ ../private/base_hex.txt │ ../private/new_hex.txt │ │ allocs/op │ allocs/op vs base │ TraceIDFromHex-12 1.000 ± 0% 0.000 ± 0% -100.00% (p=0.000 n=10) SpanIDFromHex-12 1.000 ± 0% 0.000 ± 0% -100.00% (p=0.000 n=10) geomean 1.000 ? ¹ ² ¹ summaries must be >0 to compute geomean ² ratios must be >0 to compute geomean ``` Issue: open-telemetry#6721
Signed-off-by: Bryan Boreham <bjboreham@gmail.com>
Signed-off-by: Bryan Boreham <bjboreham@gmail.com>
This version runs ~5% slower. Signed-off-by: Bryan Boreham <bjboreham@gmail.com>
bboreham
requested review from
MrAlias,
XSAM,
dashpole,
dmathieu,
flc1125 and
pellared
as code owners
Contributor
Author
|
Changed to draft in view of the test failures. |
Contributor
|
I think this PR can be closed, see After reviewing the current code/main Branch, the core optimizations described in this PR as addition to #6791 and #7155 (fast hex parsing and allocation-free formatting for TraceID / SpanID, Changeog...) are already present. Fast hex parsing (reverse lookup table)
These loops already implement the optimized hex decoding logic with the reverse lookup table and invalid character checks. Allocation-free formatting for TraceID / SpanID Formatting is already implemented using fixed-size byte arrays and lookup-table encoding:
What is not present from this PR
|
Contributor
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Update of #6791, resolving merge conflicts and adding changelog.
I also proposed a couple more commits rearranging things to avoid duplication between 8-byte and 16-byte versions of the same code; I felt the shorter version is easier to review.
See also #7155, which adds/extends tests to add confidence.