add GetAndSet to bitmap

[paulwe]

precomputing the mask seems to be a very minor pessimization but we can skip one call to getSlotAndOffset in (*RTPStatsReceiver).Update by combining get/set into one method.

also some other driveby cleanup

goos: darwin
goarch: arm64
pkg: github.com/livekit/protocol/utils
cpu: Apple M1 Pro
BenchmarkBitmap
BenchmarkBitmap/Set
BenchmarkBitmap/Set-10 584310650 2.055 ns/op
BenchmarkBitmap/Set-10 586064010 2.049 ns/op
BenchmarkBitmap/Set-10 582722875 2.057 ns/op
BenchmarkBitmap/Set-10 585069157 2.063 ns/op
BenchmarkBitmap/Set-10 566389279 2.063 ns/op
PASS
ok github.com/livekit/protocol/utils 15.692s

goos: darwin
goarch: arm64
pkg: github.com/livekit/protocol/utils
cpu: Apple M1 Pro
BenchmarkBitmap
BenchmarkBitmap/Set
BenchmarkBitmap/Set-10 591135066 2.030 ns/op
BenchmarkBitmap/Set-10 592524807 2.055 ns/op
BenchmarkBitmap/Set-10 585467359 2.022 ns/op
BenchmarkBitmap/Set-10 591682300 2.024 ns/op
BenchmarkBitmap/Set-10 593558163 2.027 ns/op

[changeset-bot]

⚠️ No Changeset found

Latest commit: 060682a

Merging this PR will not cause a version bump for any packages. If these changes should not result in a new version, you're good to go. If these changes should result in a version bump, you need to add a changeset.

Click here to learn what changesets are, and how to add one.

Click here if you're a maintainer who wants to add a changeset to this PR

💥 An error occurred when fetching the changed packages and changesets in this PR

Some errors occurred when validating the changesets config:
The package or glob expression "github.com/livekit/protocol" specified in the `fixed` option does not match any package in the project. You may have misspelled the package name or provided an invalid glob expression. Note that glob expressions must be defined according to https://www.npmjs.com/package/micromatch.

[boks1971]

ouch, that was a pessimisation, did not expect that.

[paulwe]

i'm not sure this will have any effect either way. Set has higher cpu use than IsSet which runs unconditionally. maybe the cost is dominated by some cache operation...

in a microbenchmark the extra call has no impact

BenchmarkBitmap
BenchmarkBitmap/IsSet/Set
BenchmarkBitmap/IsSet/Set-10 544463373 2.092 ns/op
BenchmarkBitmap/IsSet/Set-10 583395589 2.140 ns/op
BenchmarkBitmap/IsSet/Set-10 590237834 2.048 ns/op
BenchmarkBitmap/IsSet/Set-10 589507502 2.046 ns/op
BenchmarkBitmap/IsSet/Set-10 587347552 2.043 ns/op
BenchmarkBitmap/GetAndSet
BenchmarkBitmap/GetAndSet-10 589057398 2.048 ns/op
BenchmarkBitmap/GetAndSet-10 592168935 2.044 ns/op
BenchmarkBitmap/GetAndSet-10 591168433 2.042 ns/op
BenchmarkBitmap/GetAndSet-10 592299975 2.034 ns/op
BenchmarkBitmap/GetAndSet-10 592091506 2.040 ns/op

[boks1971]

i'm not sure this will have any effect either way. Set has higher cpu use than IsSet which runs unconditionally. maybe the cost is dominated by some cache operation...

IsSet runs only for out-of-order packets. That's why it must be smaller.

[boks1971]

ClearRangeAndSet might have a bigger impact if it can be optimised as it runs for all in-order packets. ClearRange will be a no-op unless there are gaps - https://github.com/livekit/livekit/blob/28c39efa06c980d9cfa9c110b26b7b721d2503fe/pkg/sfu/rtpstats/rtpstats_receiver.go#L325

[paulwe]

inlining doesn't seem to have an effect either 🫤

[paulwe]

ClearRangeAndSet might have a bigger impact if it can be optimised as it runs for all in-order packets. ClearRange will be a no-op unless there are gaps - https://github.com/livekit/livekit/blob/28c39efa06c980d9cfa9c110b26b7b721d2503fe/pkg/sfu/rtpstats/rtpstats_receiver.go#L325

these don't use the same bitmap indices so i don't see anything to eliminate by merging them