New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
encoding/json: optimize field sorting with slices and cmp #67046
Conversation
Refactor field sorting in typeFields to use slices.SortFunc and cmp.Compare, resulting in performance improvements in cache misses and slight increases in hit latency. Memory allocation and operation counts also reduced significantly for cache misses. ``` goos: darwin goarch: arm64 pkg: encoding/json cpu: Apple M2 Max │ old.txt │ new.txt │ │ sec/op │ sec/op vs base │ TypeFieldsCache/MissTypes1-12 5.068µ ± 2% 5.032µ ± 1% ~ (p=0.055 n=10) TypeFieldsCache/MissTypes10-12 15.22µ ± 2% 14.79µ ± 2% -2.84% (p=0.000 n=10) TypeFieldsCache/MissTypes100-12 98.40µ ± 3% 94.72µ ± 1% -3.73% (p=0.002 n=10) TypeFieldsCache/MissTypes1000-12 860.3µ ± 3% 840.7µ ± 1% ~ (p=0.105 n=10) TypeFieldsCache/MissTypes10000-12 8.092m ± 3% 7.987m ± 4% ~ (p=0.280 n=10) TypeFieldsCache/MissTypes100000-12 80.90m ± 9% 80.62m ± 2% ~ (p=0.631 n=10) TypeFieldsCache/MissTypes1000000-12 1009.5m ± 7% 933.8m ± 7% -7.50% (p=0.011 n=10) TypeFieldsCache/HitTypes1-12 1.554n ± 3% 1.617n ± 3% +4.05% (p=0.004 n=10) TypeFieldsCache/HitTypes10-12 1.575n ± 1% 1.624n ± 6% +3.14% (p=0.000 n=10) TypeFieldsCache/HitTypes100-12 1.566n ± 1% 1.627n ± 5% +3.86% (p=0.000 n=10) TypeFieldsCache/HitTypes1000-12 1.526n ± 4% 1.596n ± 3% +4.55% (p=0.001 n=10) TypeFieldsCache/HitTypes10000-12 1.506n ± 0% 1.573n ± 3% +4.41% (p=0.000 n=10) TypeFieldsCache/HitTypes100000-12 1.480n ± 1% 1.529n ± 7% +3.35% (p=0.001 n=10) TypeFieldsCache/HitTypes1000000-12 1.473n ± 2% 1.468n ± 2% ~ (p=0.739 n=10) geomean 1.371µ 1.374µ +0.25% │ old.txt │ new.txt │ │ B/op │ B/op vs base │ TypeFieldsCache/MissTypes1-12 2.023Ki ± 0% 2.000Ki ± 0% -1.16% (p=0.000 n=10) TypeFieldsCache/MissTypes10-12 10.95Ki ± 0% 10.71Ki ± 0% -2.17% (p=0.000 n=10) TypeFieldsCache/MissTypes100-12 101.91Ki ± 0% 99.53Ki ± 0% -2.34% (p=0.000 n=10) TypeFieldsCache/MissTypes1000-12 1044.3Ki ± 0% 1020.8Ki ± 0% -2.25% (p=0.000 n=10) TypeFieldsCache/MissTypes10000-12 9.957Mi ± 0% 9.730Mi ± 0% -2.29% (p=0.000 n=10) TypeFieldsCache/MissTypes100000-12 97.79Mi ± 0% 95.50Mi ± 0% -2.34% (p=0.000 n=10) TypeFieldsCache/MissTypes1000000-12 1018.4Mi ± 0% 995.5Mi ± 0% -2.25% (p=0.000 n=10) TypeFieldsCache/HitTypes1-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ TypeFieldsCache/HitTypes10-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ TypeFieldsCache/HitTypes100-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ TypeFieldsCache/HitTypes1000-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ TypeFieldsCache/HitTypes10000-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ TypeFieldsCache/HitTypes100000-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ TypeFieldsCache/HitTypes1000000-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ geomean ² -1.06% ² ¹ all samples are equal ² summaries must be >0 to compute geomean │ old.txt │ new.txt │ │ allocs/op │ allocs/op vs base │ TypeFieldsCache/MissTypes1-12 46.00 ± 0% 45.00 ± 0% -2.17% (p=0.000 n=10) TypeFieldsCache/MissTypes10-12 215.0 ± 0% 205.0 ± 0% -4.65% (p=0.000 n=10) TypeFieldsCache/MissTypes100-12 1.845k ± 0% 1.745k ± 0% -5.42% (p=0.000 n=10) TypeFieldsCache/MissTypes1000-12 18.08k ± 0% 17.08k ± 0% -5.53% (p=0.000 n=10) TypeFieldsCache/MissTypes10000-12 180.3k ± 0% 170.3k ± 0% -5.55% (p=0.000 n=10) TypeFieldsCache/MissTypes100000-12 1.804M ± 0% 1.704M ± 0% -5.54% (p=0.000 n=10) TypeFieldsCache/MissTypes1000000-12 18.04M ± 0% 17.04M ± 0% -5.54% (p=0.000 n=10) TypeFieldsCache/HitTypes1-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ TypeFieldsCache/HitTypes10-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ TypeFieldsCache/HitTypes100-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ TypeFieldsCache/HitTypes1000-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ TypeFieldsCache/HitTypes10000-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ TypeFieldsCache/HitTypes100000-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ TypeFieldsCache/HitTypes1000000-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ geomean ² -2.49% ² ¹ all samples are equal ² summaries must be >0 to compute geomean ``` Signed-off-by: aimuz <mr.imuz@gmail.com>
This PR (HEAD: d16772d) has been imported to Gerrit for code review. Please visit Gerrit at https://go-review.googlesource.com/c/go/+/581895. Important tips:
|
Message from Gopher Robot: Patch Set 1: (1 comment) Please don’t reply on this GitHub thread. Visit golang.org/cl/581895. |
Message from Zhen Li: Patch Set 1: (1 comment) Please don’t reply on this GitHub thread. Visit golang.org/cl/581895. |
Message from Ian Lance Taylor: Patch Set 1: Commit-Queue+1 (1 comment) Please don’t reply on this GitHub thread. Visit golang.org/cl/581895. |
Message from Go LUCI: Patch Set 1: Dry run: CV is trying the patch. Bot data: {"action":"start","triggered_at":"2024-04-26T03:38:40Z","revision":"66fd832bd2ba89353a6d455cd48f02e2ba5bba10"} Please don’t reply on this GitHub thread. Visit golang.org/cl/581895. |
Message from Zhen Li: Patch Set 1: (1 comment) Please don’t reply on this GitHub thread. Visit golang.org/cl/581895. |
Message from Go LUCI: Patch Set 1: LUCI-TryBot-Result-1 Copied votes on follow-up patch sets have been updated:
Please don’t reply on this GitHub thread. Visit golang.org/cl/581895. |
Message from Go LUCI: Patch Set 2: Dry run: CV is trying the patch. Bot data: {"action":"start","triggered_at":"2024-04-26T03:45:33Z","revision":"d3bd6b580e4897238935abd1c9c811d5c1fcf9a6"} Please don’t reply on this GitHub thread. Visit golang.org/cl/581895. |
Message from Ian Lance Taylor: Patch Set 2: -Commit-Queue Please don’t reply on this GitHub thread. Visit golang.org/cl/581895. |
Message from Go LUCI: Patch Set 2: This CL has passed the run Please don’t reply on this GitHub thread. Visit golang.org/cl/581895. |
Message from Go LUCI: Patch Set 2: LUCI-TryBot-Result+1 Please don’t reply on this GitHub thread. Visit golang.org/cl/581895. |
Message from Ian Lance Taylor: Patch Set 2: (1 comment) Please don’t reply on this GitHub thread. Visit golang.org/cl/581895. |
Message from Zhen Li: Patch Set 2: (1 comment) Please don’t reply on this GitHub thread. Visit golang.org/cl/581895. |
Message from Ian Lance Taylor: Patch Set 3: Auto-Submit+1 Code-Review+2 Commit-Queue+1 (1 comment) Please don’t reply on this GitHub thread. Visit golang.org/cl/581895. |
Message from Go LUCI: Patch Set 3: Dry run: CV is trying the patch. Bot data: {"action":"start","triggered_at":"2024-04-28T17:27:07Z","revision":"d6a90d342ae1ef41c9062c65ab7a914dbceaec96"} Please don’t reply on this GitHub thread. Visit golang.org/cl/581895. |
Message from Ian Lance Taylor: Patch Set 3: -Commit-Queue Please don’t reply on this GitHub thread. Visit golang.org/cl/581895. |
Message from Go LUCI: Patch Set 3: This CL has passed the run Please don’t reply on this GitHub thread. Visit golang.org/cl/581895. |
Message from Go LUCI: Patch Set 3: LUCI-TryBot-Result+1 Please don’t reply on this GitHub thread. Visit golang.org/cl/581895. |
Refactor field sorting in typeFields to use slices.SortFunc and cmp.Compare, resulting in performance improvements in cache misses and slight increases in hit latency. Memory allocation and operation counts also reduced significantly for cache misses. goos: darwin goarch: arm64 pkg: encoding/json cpu: Apple M2 Max │ old.txt │ new.txt │ │ sec/op │ sec/op vs base │ TypeFieldsCache/MissTypes1-12 5.068µ ± 2% 5.032µ ± 1% ~ (p=0.055 n=10) TypeFieldsCache/MissTypes10-12 15.22µ ± 2% 14.79µ ± 2% -2.84% (p=0.000 n=10) TypeFieldsCache/MissTypes100-12 98.40µ ± 3% 94.72µ ± 1% -3.73% (p=0.002 n=10) TypeFieldsCache/MissTypes1000-12 860.3µ ± 3% 840.7µ ± 1% ~ (p=0.105 n=10) TypeFieldsCache/MissTypes10000-12 8.092m ± 3% 7.987m ± 4% ~ (p=0.280 n=10) TypeFieldsCache/MissTypes100000-12 80.90m ± 9% 80.62m ± 2% ~ (p=0.631 n=10) TypeFieldsCache/MissTypes1000000-12 1009.5m ± 7% 933.8m ± 7% -7.50% (p=0.011 n=10) TypeFieldsCache/HitTypes1-12 1.554n ± 3% 1.617n ± 3% +4.05% (p=0.004 n=10) TypeFieldsCache/HitTypes10-12 1.575n ± 1% 1.624n ± 6% +3.14% (p=0.000 n=10) TypeFieldsCache/HitTypes100-12 1.566n ± 1% 1.627n ± 5% +3.86% (p=0.000 n=10) TypeFieldsCache/HitTypes1000-12 1.526n ± 4% 1.596n ± 3% +4.55% (p=0.001 n=10) TypeFieldsCache/HitTypes10000-12 1.506n ± 0% 1.573n ± 3% +4.41% (p=0.000 n=10) TypeFieldsCache/HitTypes100000-12 1.480n ± 1% 1.529n ± 7% +3.35% (p=0.001 n=10) TypeFieldsCache/HitTypes1000000-12 1.473n ± 2% 1.468n ± 2% ~ (p=0.739 n=10) geomean 1.371µ 1.374µ +0.25% │ old.txt │ new.txt │ │ B/op │ B/op vs base │ TypeFieldsCache/MissTypes1-12 2.023Ki ± 0% 2.000Ki ± 0% -1.16% (p=0.000 n=10) TypeFieldsCache/MissTypes10-12 10.95Ki ± 0% 10.71Ki ± 0% -2.17% (p=0.000 n=10) TypeFieldsCache/MissTypes100-12 101.91Ki ± 0% 99.53Ki ± 0% -2.34% (p=0.000 n=10) TypeFieldsCache/MissTypes1000-12 1044.3Ki ± 0% 1020.8Ki ± 0% -2.25% (p=0.000 n=10) TypeFieldsCache/MissTypes10000-12 9.957Mi ± 0% 9.730Mi ± 0% -2.29% (p=0.000 n=10) TypeFieldsCache/MissTypes100000-12 97.79Mi ± 0% 95.50Mi ± 0% -2.34% (p=0.000 n=10) TypeFieldsCache/MissTypes1000000-12 1018.4Mi ± 0% 995.5Mi ± 0% -2.25% (p=0.000 n=10) TypeFieldsCache/HitTypes1-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ TypeFieldsCache/HitTypes10-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ TypeFieldsCache/HitTypes100-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ TypeFieldsCache/HitTypes1000-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ TypeFieldsCache/HitTypes10000-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ TypeFieldsCache/HitTypes100000-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ TypeFieldsCache/HitTypes1000000-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ geomean ² -1.06% ² ¹ all samples are equal ² summaries must be >0 to compute geomean │ old.txt │ new.txt │ │ allocs/op │ allocs/op vs base │ TypeFieldsCache/MissTypes1-12 46.00 ± 0% 45.00 ± 0% -2.17% (p=0.000 n=10) TypeFieldsCache/MissTypes10-12 215.0 ± 0% 205.0 ± 0% -4.65% (p=0.000 n=10) TypeFieldsCache/MissTypes100-12 1.845k ± 0% 1.745k ± 0% -5.42% (p=0.000 n=10) TypeFieldsCache/MissTypes1000-12 18.08k ± 0% 17.08k ± 0% -5.53% (p=0.000 n=10) TypeFieldsCache/MissTypes10000-12 180.3k ± 0% 170.3k ± 0% -5.55% (p=0.000 n=10) TypeFieldsCache/MissTypes100000-12 1.804M ± 0% 1.704M ± 0% -5.54% (p=0.000 n=10) TypeFieldsCache/MissTypes1000000-12 18.04M ± 0% 17.04M ± 0% -5.54% (p=0.000 n=10) TypeFieldsCache/HitTypes1-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ TypeFieldsCache/HitTypes10-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ TypeFieldsCache/HitTypes100-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ TypeFieldsCache/HitTypes1000-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ TypeFieldsCache/HitTypes10000-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ TypeFieldsCache/HitTypes100000-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ TypeFieldsCache/HitTypes1000000-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹ geomean ² -2.49% ² ¹ all samples are equal ² summaries must be >0 to compute geomean Change-Id: I62b8f524f5747f4f1b9241a1d849efeef1851049 GitHub-Last-Rev: d16772d GitHub-Pull-Request: #67046 Reviewed-on: https://go-review.googlesource.com/c/go/+/581895 LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com> Auto-Submit: Ian Lance Taylor <iant@google.com> Reviewed-by: Ian Lance Taylor <iant@google.com> Reviewed-by: Joedian Reid <joedian@google.com>
Message from Joedian Reid: Patch Set 3: Code-Review+1 Please don’t reply on this GitHub thread. Visit golang.org/cl/581895. |
This PR is being closed because golang.org/cl/581895 has been merged. |
Refactor field sorting in typeFields to use slices.SortFunc and cmp.Compare,
resulting in performance improvements in cache misses and slight increases in
hit latency. Memory allocation and operation counts also reduced significantly
for cache misses.
goos: darwin
goarch: arm64
pkg: encoding/json
cpu: Apple M2 Max
│ old.txt │ new.txt │
│ sec/op │ sec/op vs base │
TypeFieldsCache/MissTypes1-12 5.068µ ± 2% 5.032µ ± 1% ~ (p=0.055 n=10)
TypeFieldsCache/MissTypes10-12 15.22µ ± 2% 14.79µ ± 2% -2.84% (p=0.000 n=10)
TypeFieldsCache/MissTypes100-12 98.40µ ± 3% 94.72µ ± 1% -3.73% (p=0.002 n=10)
TypeFieldsCache/MissTypes1000-12 860.3µ ± 3% 840.7µ ± 1% ~ (p=0.105 n=10)
TypeFieldsCache/MissTypes10000-12 8.092m ± 3% 7.987m ± 4% ~ (p=0.280 n=10)
TypeFieldsCache/MissTypes100000-12 80.90m ± 9% 80.62m ± 2% ~ (p=0.631 n=10)
TypeFieldsCache/MissTypes1000000-12 1009.5m ± 7% 933.8m ± 7% -7.50% (p=0.011 n=10)
TypeFieldsCache/HitTypes1-12 1.554n ± 3% 1.617n ± 3% +4.05% (p=0.004 n=10)
TypeFieldsCache/HitTypes10-12 1.575n ± 1% 1.624n ± 6% +3.14% (p=0.000 n=10)
TypeFieldsCache/HitTypes100-12 1.566n ± 1% 1.627n ± 5% +3.86% (p=0.000 n=10)
TypeFieldsCache/HitTypes1000-12 1.526n ± 4% 1.596n ± 3% +4.55% (p=0.001 n=10)
TypeFieldsCache/HitTypes10000-12 1.506n ± 0% 1.573n ± 3% +4.41% (p=0.000 n=10)
TypeFieldsCache/HitTypes100000-12 1.480n ± 1% 1.529n ± 7% +3.35% (p=0.001 n=10)
TypeFieldsCache/HitTypes1000000-12 1.473n ± 2% 1.468n ± 2% ~ (p=0.739 n=10)
geomean 1.371µ 1.374µ +0.25%
TypeFieldsCache/MissTypes1-12 2.023Ki ± 0% 2.000Ki ± 0% -1.16% (p=0.000 n=10)
TypeFieldsCache/MissTypes10-12 10.95Ki ± 0% 10.71Ki ± 0% -2.17% (p=0.000 n=10)
TypeFieldsCache/MissTypes100-12 101.91Ki ± 0% 99.53Ki ± 0% -2.34% (p=0.000 n=10)
TypeFieldsCache/MissTypes1000-12 1044.3Ki ± 0% 1020.8Ki ± 0% -2.25% (p=0.000 n=10)
TypeFieldsCache/MissTypes10000-12 9.957Mi ± 0% 9.730Mi ± 0% -2.29% (p=0.000 n=10)
TypeFieldsCache/MissTypes100000-12 97.79Mi ± 0% 95.50Mi ± 0% -2.34% (p=0.000 n=10)
TypeFieldsCache/MissTypes1000000-12 1018.4Mi ± 0% 995.5Mi ± 0% -2.25% (p=0.000 n=10)
TypeFieldsCache/HitTypes1-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹
TypeFieldsCache/HitTypes10-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹
TypeFieldsCache/HitTypes100-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹
TypeFieldsCache/HitTypes1000-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹
TypeFieldsCache/HitTypes10000-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹
TypeFieldsCache/HitTypes100000-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹
TypeFieldsCache/HitTypes1000000-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹
geomean ² -1.06% ²
¹ all samples are equal
² summaries must be >0 to compute geomean
TypeFieldsCache/MissTypes1-12 46.00 ± 0% 45.00 ± 0% -2.17% (p=0.000 n=10)
TypeFieldsCache/MissTypes10-12 215.0 ± 0% 205.0 ± 0% -4.65% (p=0.000 n=10)
TypeFieldsCache/MissTypes100-12 1.845k ± 0% 1.745k ± 0% -5.42% (p=0.000 n=10)
TypeFieldsCache/MissTypes1000-12 18.08k ± 0% 17.08k ± 0% -5.53% (p=0.000 n=10)
TypeFieldsCache/MissTypes10000-12 180.3k ± 0% 170.3k ± 0% -5.55% (p=0.000 n=10)
TypeFieldsCache/MissTypes100000-12 1.804M ± 0% 1.704M ± 0% -5.54% (p=0.000 n=10)
TypeFieldsCache/MissTypes1000000-12 18.04M ± 0% 17.04M ± 0% -5.54% (p=0.000 n=10)
TypeFieldsCache/HitTypes1-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹
TypeFieldsCache/HitTypes10-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹
TypeFieldsCache/HitTypes100-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹
TypeFieldsCache/HitTypes1000-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹
TypeFieldsCache/HitTypes10000-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹
TypeFieldsCache/HitTypes100000-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹
TypeFieldsCache/HitTypes1000000-12 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹
geomean ² -2.49% ²
¹ all samples are equal
² summaries must be >0 to compute geomean