[levkk]

#1156

[codecov]

Codecov Report

❌ Patch coverage is 97.58713% with 9 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
pgdog-postgres-types/src/bigint.rs	58.33%	5 Missing ⚠️
pgdog-config/src/database.rs	0.00%	2 Missing ⚠️
...eplication/logical/publisher/replication_stream.rs	80.00%	1 Missing ⚠️
pgdog/src/frontend/router/comment.rs	0.00%	1 Missing ⚠️

📢 Thoughts on this report? Let us know!

[github-actions]

Metal bench (arm64) failed. 45 of 152 measures outside bounds.

suite	throughput	cpu/txn	user cycles/txn	user instr/txn	kernel cycles/txn	kernel instr/txn	ctx/txn
pooler		19.97 (-0.8% ± 0.8%)	20,343 (-1.0% ± 0.2%)	35,276 (+0.1% ± 0.05%)	30,988 (-1.0% ± 1.3%)	51,779 (+0.0% ± 0.2%)
lb		20.44 (-0.4% ± 0.6%)	20,842 (-1.0% ± 0.3%)	36,201 (+0.1% ± 0.04%)	31,492 (-0.1% ± 1.1%)	53,077 (+0.1% ± 0.2%)
sharding-w2		38.3 (-0.3% ± 0.7%)	53,028 (+0.1% ± 0.6%)	102,633 (+0.1% ± 0.00%)	47,502 (-0.6% ± 1.0%)	54,323 (-0.0% ± 0.1%)	1.514 (-0.2% ± 0.1%)
sharding-w8	218,345 (+0.0% ± 0.4%)	33.65 (-0.1% ± 0.3%)	51,529 (-0.6% ± 0.5%)	57,956 (+0.1% ± 0.06%)	41,683 (+0.7% ± 0.4%)	42,954 (+0.0% ± 0.07%)	0.1416 (+0.3% ± 1.1%)

Counters per txn

suite	frontend stall	l1i refill	backend stall
pooler	8,374 (-3.3% ± 0.7%)	2,794 (-0.3% ± 0.3%)
lb	8,590 (-3.6% ± 0.8%)	2,840 (-0.4% ± 0.3%)
sharding-w2	24,680 (+0.3% ± 1.8%)	4,930 (-0.0% ± 1.1%)
sharding-w8	26,246 (-0.9% ± 0.6%)	4,726 (-1.7% ± 0.3%)

benchmark	measure	result	baseline	change	bound
extended_cached::aggregate	alloc_bytes	941	917	+2.62% ± 0.00%	+1%
extended_cached::begin	alloc_bytes	240	216	+11.11% ± 0.00%	+1%
extended_cached::cte	alloc_bytes	1,204	1,180	+2.03% ± 0.00%	+1%
extended_cached::delete_by_key	alloc_bytes	636	612	+3.92% ± 0.00%	+1%
extended_cached::insert	alloc_bytes	560	536	+4.48% ± 0.00%	+1%
extended_cached::insert_split	alloc_bytes	6,810	6,530	+4.29% ± 0.00%	+1%
extended_cached::join	alloc_bytes	1,588	1,564	+1.53% ± 0.00%	+1%
extended_cached::order_limit_offset	alloc_bytes	931	907	+2.65% ± 0.00%	+1%
extended_cached::select_all_shards	alloc_bytes	624	600	+4.00% ± 0.00%	+1%
extended_cached::select_by_key	alloc_bytes	936	912	+2.63% ± 0.00%	+1%
extended_cached::select_in_list	alloc_bytes	1,144	1,120	+2.14% ± 0.00%	+1%
extended_cached::select_omni	alloc_bytes	432	408	+5.88% ± 0.00%	+1%
extended_cached::set	alloc_bytes	425	401	+5.99% ± 0.00%	+1%
extended_cached::subquery	alloc_bytes	816	792	+3.03% ± 0.00%	+1%
extended_cached::update_by_key	alloc_bytes	636	612	+3.92% ± 0.00%	+1%
extended_cold::aggregate	alloc_bytes	2,006	1,982	+1.21% ± 0.00%	+1%
extended_cold::begin	alloc_bytes	678	654	+3.67% ± 0.00%	+1%
extended_cold::cte	alloc_bytes	1,755	1,731	+1.39% ± 0.00%	+1%
extended_cold::delete_by_key	alloc_bytes	1,125	1,101	+2.18% ± 0.00%	+1%
extended_cold::insert	alloc_bytes	1,511	1,487	+1.61% ± 0.00%	+1%
extended_cold::insert_split	alloc_bytes	9,386	9,106	+3.07% ± 0.00%	+1%
extended_cold::join	alloc_bytes	2,166	2,142	+1.12% ± 0.00%	+1%
extended_cold::order_limit_offset	alloc_bytes	1,547	1,523	+1.58% ± 0.00%	+1%
extended_cold::select_all_shards	alloc_bytes	1,094	1,070	+2.24% ± 0.00%	+1%
extended_cold::select_by_key	alloc_bytes	1,435	1,411	+1.70% ± 0.00%	+1%
extended_cold::select_in_list	alloc_bytes	1,660	1,636	+1.47% ± 0.00%	+1%
extended_cold::select_omni	alloc_bytes	915	891	+2.69% ± 0.00%	+1%
extended_cold::set	alloc_bytes	911	887	+2.71% ± 0.00%	+1%
extended_cold::subquery	alloc_bytes	1,368	1,344	+1.79% ± 0.00%	+1%
extended_cold::update_by_key	alloc_bytes	1,152	1,128	+2.13% ± 0.00%	+1%
simple_cold::aggregate	alloc_bytes	1,870	1,846	+1.30% ± 0.00%	+1%
simple_cold::begin	alloc_bytes	614	590	+4.07% ± 0.00%	+1%
simple_cold::cte	alloc_bytes	1,691	1,667	+1.44% ± 0.00%	+1%
simple_cold::delete_by_key	alloc_bytes	1,061	1,037	+2.31% ± 0.00%	+1%
simple_cold::insert	alloc_bytes	1,447	1,423	+1.69% ± 0.00%	+1%
simple_cold::insert_split	alloc_bytes	8,058	7,922	+1.72% ± 0.00%	+1%
simple_cold::join	alloc_bytes	2,102	2,078	+1.15% ± 0.00%	+1%
simple_cold::order_limit_offset	alloc_bytes	1,425	1,401	+1.71% ± 0.00%	+1%
simple_cold::select_all_shards	alloc_bytes	1,030	1,006	+2.39% ± 0.00%	+1%
simple_cold::select_by_key	alloc_bytes	1,371	1,347	+1.78% ± 0.00%	+1%

Microbench: 90 of 135 measures within bounds. See verdict.json for the rest.

Microbench latency: 2 benches changed more than 3%

benchmark	latency	baseline	change	instructions change
simple_cold::set	2,304 ns	2,392 ns	-3.67% ± 0.42%	+0.38% ± 0.00%
simple_cold::begin	1,965 ns	1,901 ns	+3.38% ± 0.70%	+0.55% ± 0.00%

lb flamegraph vs 6c785c0

Open interactive for hover, zoom and search.

Open interactive.

microbench flamegraph vs 6c785c0

Open interactive for hover, zoom and search.

Open interactive.

pooler flamegraph vs 6c785c0

Open interactive for hover, zoom and search.

Open interactive.

sharding-w2 flamegraph vs 6c785c0

Open interactive for hover, zoom and search.

Open interactive.

sharding-w8 flamegraph vs 6c785c0

Open interactive for hover, zoom and search.

Open interactive.

Logs

lb, microbench, pooler, sharding-w2, sharding-w8, verdict.json

Compared 6739d44 with 6c785c0 on m8g.metal-24xl (spot).

[github-actions]

Metal bench (x86) failed. 47 of 152 measures outside bounds.

suite	throughput	cpu/txn	user cycles/txn	user instr/txn	kernel cycles/txn	kernel instr/txn	ctx/txn
pooler		35.15 (-0.5% ± 0.1%)	24,525 (-1.4% ± 0.3%)	36,709 (+0.0% ± 0.01%)	57,605 (-0.3% ± 0.2%)	38,484 (+0.0% ± 0.2%)
lb		36.14 (-0.5% ± 0.2%)	25,599 (-1.5% ± 0.4%)	37,653 (+0.1% ± 0.02%)	58,618 (-0.1% ± 0.3%)	39,177 (+0.1% ± 0.2%)
sharding-w2		57.85 (-0.5% ± 0.3%)	57,108 (-0.3% ± 0.4%)	106,046 (+0.0% ± 0.04%)	73,040 (-0.7% ± 0.8%)	43,007 (+0.0% ± 0.08%)	1.381 (+0.1% ± 0.3%)
sharding-w8	143,706 (-0.3% ± 0.8%)	50.06 (+0.2% ± 0.4%)	62,650 (+0.5% ± 0.4%)	61,446 (+0.0% ± 0.03%)	64,764 (+0.1% ± 0.7%)	34,212 (-0.0% ± 0.06%)	0.1564 (-2.3% ± 0.6%)

Counters per txn

suite	frontend stall	l1i refill	backend stall
pooler	77,158 (-2.9% ± 0.4%)	3,748 (-3.1% ± 0.1%)
lb	80,165 (-2.5% ± 0.6%)	3,880 (-3.0% ± 0.09%)
sharding-w2	170,123 (-2.5% ± 0.8%)	6,400 (-2.7% ± 0.3%)
sharding-w8	133,241 (-0.8% ± 1.5%)	5,101 (-0.2% ± 0.3%)

benchmark	measure	result	baseline	change	bound
extended_cached::aggregate	alloc_bytes	941	917	+2.62% ± 0.00%	+1%
extended_cached::begin	alloc_bytes	240	216	+11.11% ± 0.00%	+1%
extended_cached::cte	alloc_bytes	1,228	1,204	+1.99% ± 0.00%	+1%
extended_cached::delete_by_key	alloc_bytes	644	620	+3.87% ± 0.00%	+1%
extended_cached::insert	alloc_bytes	560	536	+4.48% ± 0.00%	+1%
extended_cached::insert_split	alloc_bytes	6,810	6,530	+4.29% ± 0.00%	+1%
extended_cached::join	alloc_bytes	1,612	1,588	+1.51% ± 0.00%	+1%
extended_cached::order_limit_offset	alloc_bytes	931	907	+2.65% ± 0.00%	+1%
extended_cached::order_limit_offset	instructions	18,060	17,666	+2.23% ± 0.00%	+1%
extended_cached::select_all_shards	alloc_bytes	624	600	+4.00% ± 0.00%	+1%
extended_cached::select_by_key	alloc_bytes	952	928	+2.59% ± 0.00%	+1%
extended_cached::select_in_list	alloc_bytes	1,160	1,136	+2.11% ± 0.00%	+1%
extended_cached::select_omni	alloc_bytes	432	408	+5.88% ± 0.00%	+1%
extended_cached::set	alloc_bytes	425	401	+5.99% ± 0.00%	+1%
extended_cached::set	instructions	8,213	7,845	+4.69% ± 0.00%	+1%
extended_cached::subquery	alloc_bytes	816	792	+3.03% ± 0.00%	+1%
extended_cached::update_by_key	alloc_bytes	644	620	+3.87% ± 0.00%	+1%
extended_cold::aggregate	alloc_bytes	2,006	1,982	+1.21% ± 0.00%	+1%
extended_cold::begin	alloc_bytes	678	654	+3.67% ± 0.00%	+1%
extended_cold::cte	alloc_bytes	1,779	1,755	+1.37% ± 0.00%	+1%
extended_cold::delete_by_key	alloc_bytes	1,133	1,109	+2.16% ± 0.00%	+1%
extended_cold::insert	alloc_bytes	1,511	1,487	+1.61% ± 0.00%	+1%
extended_cold::insert_split	alloc_bytes	9,386	9,106	+3.07% ± 0.00%	+1%
extended_cold::join	alloc_bytes	2,190	2,166	+1.11% ± 0.00%	+1%
extended_cold::order_limit_offset	alloc_bytes	1,547	1,523	+1.58% ± 0.00%	+1%
extended_cold::select_all_shards	alloc_bytes	1,094	1,070	+2.24% ± 0.00%	+1%
extended_cold::select_by_key	alloc_bytes	1,451	1,427	+1.68% ± 0.00%	+1%
extended_cold::select_in_list	alloc_bytes	1,676	1,652	+1.45% ± 0.00%	+1%
extended_cold::select_omni	alloc_bytes	915	891	+2.69% ± 0.00%	+1%
extended_cold::set	alloc_bytes	911	887	+2.71% ± 0.00%	+1%
extended_cold::subquery	alloc_bytes	1,368	1,344	+1.79% ± 0.00%	+1%
extended_cold::update_by_key	alloc_bytes	1,160	1,136	+2.11% ± 0.00%	+1%
simple_cold::aggregate	alloc_bytes	1,870	1,846	+1.30% ± 0.00%	+1%
simple_cold::begin	alloc_bytes	614	590	+4.07% ± 0.00%	+1%
simple_cold::cte	alloc_bytes	1,715	1,691	+1.42% ± 0.00%	+1%
simple_cold::delete_by_key	alloc_bytes	1,069	1,045	+2.30% ± 0.00%	+1%
simple_cold::insert	alloc_bytes	1,447	1,423	+1.69% ± 0.00%	+1%
simple_cold::insert_split	alloc_bytes	8,058	7,922	+1.72% ± 0.00%	+1%
simple_cold::join	alloc_bytes	2,126	2,102	+1.14% ± 0.00%	+1%
simple_cold::order_limit_offset	alloc_bytes	1,425	1,401	+1.71% ± 0.00%	+1%

Microbench: 88 of 135 measures within bounds. See verdict.json for the rest.

lb flamegraph vs 6c785c0

Open interactive for hover, zoom and search.

Open interactive.

microbench flamegraph vs 6c785c0

Open interactive for hover, zoom and search.

Open interactive.

pooler flamegraph vs 6c785c0

Open interactive for hover, zoom and search.

Open interactive.

sharding-w2 flamegraph vs 6c785c0

Open interactive for hover, zoom and search.

Open interactive.

sharding-w8 flamegraph vs 6c785c0

Open interactive for hover, zoom and search.

Open interactive.

Logs

lb, microbench, pooler, sharding-w2, sharding-w8, verdict.json

Compared 6739d44 with 6c785c0 on m7a.metal-48xl (spot).