[dependabot]

Bumps github.com/aws/aws-sdk-go-v2/credentials from 1.20.9 to 1.20.10.

Commits

• 4b3b917 Release 2022-09-14
• 3ad6cfd Regenerated Clients
• abfad89 Update SDK's smithy-go dependency to v1.13.3
• b5262aa Update SDK with latest version of API models (#1844)
• a13b7a4 codegen: Upgrade to smithy 1.25.0 (#1835)
• 296bfb3 Update PULL_REQUEST_TEMPLATE.md (#1843)
• 63566f0 Update BuildAuthToken to validate endpoint contains a port (#1837)
• b011f04 Update credentials.go (#1841)
• 018b8c3 Amend typological errors in retryer package comments
• 5b135f8 Release 2022-09-02.2
• Additional commits viewable in compare view

[github-actions]

Engine benchmark (base vs head, with an A/A control)

In-process fake connectors on both pipeline engines (make bench-engine), base and head
run alternately on the same runner. Medians of records/s, higher is better. The noise floor is
measured on the base revision against itself in the same job. This reports whether this runner
could tell the two revisions apart (a change is called only beyond twice the floor); it is not a
throughput claim.

benchmark	base	head	delta	noise floor	read
Engine/1x1-acklatency100us/v1	108,127	110,547	+2.2%	4.9%	within noise
Engine/1x1-acklatency100us/v2	930	932	+0.2%	0.2%	within noise
Engine/1x1-bufferedacks/v1	105,777	104,746	-1.0%	2.0%	within noise
Engine/1x1-bufferedacks/v2	253,579	262,963	+3.7%	3.9%	within noise
Engine/1x1-destbatch100/v1	124,450	123,018	-1.2%	2.8%	within noise
Engine/1x1-destbatch100/v2	820,448	806,353	-1.7%	2.5%	within noise
Engine/1x1-srcbatch100/v1	115,589	117,289	+1.5%	3.9%	within noise
Engine/1x1-srcbatch100/v2	825,619	843,856	+2.2%	10.3%	within noise
Engine/1x1/v1	108,728	106,855	-1.7%	2.7%	within noise
Engine/1x1/v2	376,917	365,882	-2.9%	4.6%	within noise
Engine/2x2/v1	53,566	53,873	+0.6%	3.0%	within noise
Engine/2x2/v2	56,339	57,259	+1.6%	3.2%	within noise
Engine/4x4/v1	35,629	35,664	+0.1%	1.8%	within noise
Engine/4x4/v2	57,716	57,989	+0.5%	2.1%	within noise

Gate: pass — no benchmark regressed more than 10.0% beyond twice the A/A floor.

_{base 3438aaaa31, PR head 2f2378f42d; benchmarked on merge commit 630e10201d - run. Wall-clock shapes (acklatency) depend on runner timer granularity.}