benchmarks

[qjustfeelitp]

experiments resulted in

latest net6 and ef6 used
that async enumerable pass through is just for fun
for the best perf without proper indices, the best way to go is at least I think separate queries see the last commit, maybe raw SQL with parameters and separate queries will be fastest, give it a shot

BenchmarkDotNet=v0.13.2, OS=Windows 11 (10.0.22000.856/21H2)
12th Gen Intel Core i9-12900, 1 CPU, 24 logical and 16 physical cores
.NET SDK=7.0.100-rc.1.22431.12
  [Host] : .NET 6.0.9 (6.0.922.41905), X64 RyuJIT AVX2

Job=InProcess  Toolchain=InProcessEmitToolchain  

Method	Mean	Error	StdDev	Median	Ratio	Allocated	Alloc Ratio
GetAuthors	531,522,260.00 ns	10,165,518.398 ns	15,215,269.291 ns	530,266,350.00 ns	1.000	192848136 B	1.000
GetAuthors_Optimized	3,905,332.13 ns	222,979.085 ns	657,458.862 ns	3,654,059.77 ns	0.007	13656 B	0.000
GetAuthors_OptimizedInline	3,585,813.95 ns	134,630.094 ns	377,517.218 ns	3,410,118.75 ns	0.007	57883 B	0.000
GetAuthors_OptimizedInlineWithParameters	3,389,941.35 ns	67,457.295 ns	82,843.616 ns	3,372,915.62 ns	0.006	60217 B	0.000
GetAuthors_OptimizedWithoutExpressions	3,197,558.28 ns	62,037.349 ns	86,967.699 ns	3,199,947.66 ns	0.006	13652 B	0.000
GetAuthors_OptimizedAsync	4,108,718.76 ns	99,986.187 ns	294,811.528 ns	4,133,792.97 ns	0.008	17296 B	0.000
GetAuthors_OptimizedAsyncPassThrough	10.29 ns	0.145 ns	0.135 ns	10.27 ns	0.000	176 B	0.000
GetAuthors_OptimizedParameterized	3,294,977.28 ns	66,837.663 ns	185,206.851 ns	3,226,722.66 ns	0.006	15412 B	0.000
GetAuthors_OptimizedRecordSpan	3,249,605.65 ns	61,928.241 ns	135,933.991 ns	3,219,528.91 ns	0.006	13893 B	0.000
GetAuthors_OptimizedRecordList	3,198,880.39 ns	63,606.253 ns	84,912.521 ns	3,184,392.58 ns	0.006	13981 B	0.000
GetAuthors_OptimizedRecordListInline	4,159,730.20 ns	275,687.471 ns	812,870.730 ns	4,022,810.55 ns	0.008	55783 B	0.000
GetAuthors_OptimizedRecordListInlineWithParameters	4,011,464.06 ns	248,033.087 ns	731,331.154 ns	3,656,656.45 ns	0.008	58651 B	0.000
GetAuthors_OptimizedRecordListWithoutExpressions	3,903,027.41 ns	248,330.099 ns	732,206.900 ns	3,756,379.30 ns	0.007	13986 B	0.000
GetAuthors_OptimizedRecordAsync	4,638,188.51 ns	92,168.553 ns	129,207.439 ns	4,625,794.53 ns	0.009	17274 B	0.000
GetAuthors_OptimizedRecordAsyncPassThrough	10.53 ns	0.080 ns	0.071 ns	10.53 ns	0.000	176 B	0.000
GetAuthors_OptimizedRecordListParameterized	3,986,353.55 ns	232,241.476 ns	684,769.233 ns	3,841,485.55 ns	0.007	15746 B	0.000
GetAuthors_RawSqlArray	3,140,509.43 ns	59,518.686 ns	70,852.761 ns	3,112,501.17 ns	0.006	14118 B	0.000
GetAuthors_RawSqlWithoutParametersArray	4,127,885.30 ns	255,149.075 ns	752,312.804 ns	3,959,378.12 ns	0.008	12173 B	0.000
GetAuthors_OptimizedMultipleQueries	3,494,956.88 ns	67,157.316 ns	65,957.464 ns	3,494,540.23 ns	0.007	90322 B	0.000
GetAuthors_MultipleQuery	2,540,432.96 ns	46,510.594 ns	45,679.622 ns	2,534,841.41 ns	0.005	26285 B	0.000
GetAuthors_MultipleQueryWithoutSelectorExpressions	2,562,001.44 ns	50,546.824 ns	56,182.661 ns	2,569,712.50 ns	0.005	26285 B	0.000

[m-jovanovic]

@qjustfeelitp Are you sure GetAuthors_OptimizedAsyncPassThrough and GetAuthors_OptimizedRecordAsyncPassThrough are executing correctly? It seems like the method completes without enumerating the IAsyncEnumerable and you get such fast times.

[qjustfeelitp]

as i mentioned in description they are there just for fun, they are not meant to be taken seriously 😄