Metrics for fma

analyze7.0ms (0.1%)

Algorithm

1×

Search

Probability	Valid	Unknown	Precondition	Infinite	Domain	Can't	Iter
0%	0%	0%	100%	0%	0%	0%	0
0%	0%	0%	100%	0%	0%	0%	1
0%	0%	0%	100%	0%	0%	0%	2
50%	0%	0%	100%	0%	0%	0%	3
75%	0%	0%	100%	0%	0%	0%	4
87.5%	0%	0%	100%	0%	0%	0%	5
93.7%	0%	0%	100%	0%	0%	0%	6
96.9%	0%	0%	100%	0%	0%	0%	7
98.4%	0%	0%	100%	0%	0%	0%	8
99.2%	0%	0%	100%	0%	0%	0%	9
99.6%	0%	0%	100%	0%	0%	0%	10
99.8%	0%	0%	100%	0%	0%	0%	11
99.9%	0%	0%	100%	0%	0%	0%	12

Compiler

Compiled 27 to 16 computations (40.7% saved)

Precisions

Click to see histograms. Total time spent on operations: 5.0ms

...c/correct-round.rkt:119:19: 2.0ms (43.9% of total)

ival-sub: 1.0ms (21.9% of total)

ival-mult: 1.0ms (21.9% of total)

ival-add: 0.0ms (0% of total)

ival-and: 0.0ms (0% of total)

const: 0.0ms (0% of total)

backward-pass: 0.0ms (0% of total)

ival-<=: 0.0ms (0% of total)

sample2.7s (50.8%)

Results

2.3s

8256×

valid

Precisions

Click to see histograms. Total time spent on operations: 1.9s

...c/correct-round.rkt:119:19: 646.0ms (34.5% of total)

ival-mult: 449.0ms (24% of total)

ival-add: 421.0ms (22.5% of total)

backward-pass: 138.0ms (7.4% of total)

ival-<=: 93.0ms (5% of total)

ival-sub: 89.0ms (4.8% of total)

const: 25.0ms (1.3% of total)

ival-and: 11.0ms (0.6% of total)

Bogosity

preprocess207.0ms (3.9%)

Algorithm

2×	egg-herbie

Rules

1950×	`fnmadd-define`
1912×	`fnmadd-define`
1874×	`fma-define`
1732×	`+-lowering-+.f64`
1732×	`+-lowering-+.f32`

Iterations

Useful iterations: 4 (0.0ms)

Iter	Nodes	Cost
0	19	61
1	69	43
2	312	39
3	2050	33
4	7877	33
0	11	18
1	34	14
2	132	13
3	768	11
4	2239	5
5	5011	5
0	8223	5

Stop Event

1×	iter limit
1×	node limit
1×	node limit

Calls

Call 1

Inputs
(+ (* (+ 1 (* t 1/5000000000000000)) (+ 1 (* t 1/5000000000000000))) (- -1 (* 2 (* t 1/5000000000000000))))

Outputs
(+ (* (+ 1 (* t 1/5000000000000000)) (+ 1 (* t 1/5000000000000000))) (- -1 (* 2 (* t 1/5000000000000000))))
(.f64 (.f64 t t) #s(literal 1/25000000000000000000000000000000 binary64))

Call 2

Inputs

(+ (* (+ 1 (* t 1/5000000000000000)) (+ 1 (* t 1/5000000000000000))) (- -1 (* 2 (* t 1/5000000000000000))))

(+ (* (+ 1 (* (neg t) 1/5000000000000000)) (+ 1 (* (neg t) 1/5000000000000000))) (- -1 (* 2 (* (neg t) 1/5000000000000000))))

(neg (+ (* (+ 1 (* (neg t) 1/5000000000000000)) (+ 1 (* (neg t) 1/5000000000000000))) (- -1 (* 2 (* (neg t) 1/5000000000000000)))))

Outputs
(+ (* (+ 1 (* t 1/5000000000000000)) (+ 1 (* t 1/5000000000000000))) (- -1 (* 2 (* t 1/5000000000000000))))
(fmsub (fnmadd t -1/5000000000000000 2) (* t 1/5000000000000000) (* t 1/2500000000000000))
(+ (* (+ 1 (* (neg t) 1/5000000000000000)) (+ 1 (* (neg t) 1/5000000000000000))) (- -1 (* 2 (* (neg t) 1/5000000000000000))))
(fmsub (fnmadd t 1/5000000000000000 2) (* t -1/5000000000000000) (* t -1/2500000000000000))
(neg (+ (* (+ 1 (* (neg t) 1/5000000000000000)) (+ 1 (* (neg t) 1/5000000000000000))) (- -1 (* 2 (* (neg t) 1/5000000000000000)))))
(fnmsub (fnmadd t 1/5000000000000000 2) (* t -1/5000000000000000) (* t 1/2500000000000000))

explain208.0ms (4%)

FPErrors

Click to see full error table

Ground Truth	Example	Example	Subexpression
256	-	-	`(+.f64 (.f64 (+.f64 #s(literal 1 binary64) (.f64 t #s(literal 1/5000000000000000 binary64))) (+.f64 #s(literal 1 binary64) (.f64 t #s(literal 1/5000000000000000 binary64)))) (-.f64 #s(literal -1 binary64) (.f64 #s(literal 2 binary64) (*.f64 t #s(literal 1/5000000000000000 binary64)))))`
0	-	-	`(*.f64 t #s(literal 1/5000000000000000 binary64))`
0	-	-	`#s(literal -1 binary64)`
0	-	-	`#s(literal 1/5000000000000000 binary64)`
0	-	-	`#s(literal 2 binary64)`
0	-	-	`(+.f64 #s(literal 1 binary64) (*.f64 t #s(literal 1/5000000000000000 binary64)))`
0	-	-	`(.f64 (+.f64 #s(literal 1 binary64) (.f64 t #s(literal 1/5000000000000000 binary64))) (+.f64 #s(literal 1 binary64) (*.f64 t #s(literal 1/5000000000000000 binary64))))`
0	-	-	`(-.f64 #s(literal -1 binary64) (.f64 #s(literal 2 binary64) (.f64 t #s(literal 1/5000000000000000 binary64))))`
0	-	-	`(.f64 #s(literal 2 binary64) (.f64 t #s(literal 1/5000000000000000 binary64)))`
0	-	-	`t`
0	-	-	`#s(literal 1 binary64)`

Results

143.0ms

512×

valid

Compiler

Compiled 134 to 35 computations (73.9% saved)

Precisions

Click to see histograms. Total time spent on operations: 92.0ms

ival-add: 28.0ms (30.4% of total)

ival-mult: 25.0ms (27.2% of total)

...c/correct-round.rkt:119:19: 20.0ms (21.7% of total)

backward-pass: 13.0ms (14.1% of total)

ival-sub: 5.0ms (5.4% of total)

const: 2.0ms (2.2% of total)

eval0.0ms (0%)

Compiler: Compiled 6 to 4 computations (33.3% saved)

prune1.0ms (0%)

Alt Table

Click to see full alt table

Status	Accuracy	Program
▶	99.4%	(.f64 (.f64 t t) #s(literal 1/25000000000000000000000000000000 binary64))

Compiler

Compiled 6 to 4 computations (33.3% saved)

simplify3.0ms (0.1%)

Algorithm

1×	egg-herbie

Localize:

Found 2 expressions of interest:

New	Metric	Score	Program
✓	cost-diff	0	(*.f64 t t)
✓	cost-diff	0	(.f64 (.f64 t t) #s(literal 1/25000000000000000000000000000000 binary64))

Rules

14×	`-lowering-.f32`
14×	`-lowering-.f64`
6×	`*-commutative`
2×	`associate-r`
2×	`associate-l`

Iterations

Useful iterations: 0 (0.0ms)

Iter	Nodes	Cost
0	4	10
1	7	10
2	10	10
0	10	10

Stop Event

1×	iter limit
1×	saturated

Calls

Call 1

Inputs
(* (* t t) 1/25000000000000000000000000000000)
(* t t)
t
1/25000000000000000000000000000000

Outputs
(* (* t t) 1/25000000000000000000000000000000)
(.f64 t (.f64 t #s(literal 1/25000000000000000000000000000000 binary64)))
(* t t)
(*.f64 t t)
t
1/25000000000000000000000000000000
#s(literal 1/25000000000000000000000000000000 binary64)

localize26.0ms (0.5%)

Localize:

Found 2 expressions of interest:

New	Metric	Score	Program
✓	accuracy	100.0%	(*.f64 t t)
✓	accuracy	99.4%	(.f64 (.f64 t t) #s(literal 1/25000000000000000000000000000000 binary64))

Results

18.0ms

256×

valid

Compiler

Compiled 12 to 5 computations (58.3% saved)

Precisions

Click to see histograms. Total time spent on operations: 9.0ms

ival-mult: 4.0ms (46.4% of total)

...c/correct-round.rkt:119:19: 4.0ms (46.4% of total)

const: 0.0ms (0% of total)

backward-pass: 0.0ms (0% of total)

series2.0ms (0%)

Counts

2 → 24

Calls

Call 1

Inputs
#<alt (* (* t t) 1/25000000000000000000000000000000)>
#<alt (* t t)>

Outputs
#<alt (* 1/25000000000000000000000000000000 (pow t 2))>
#<alt (* 1/25000000000000000000000000000000 (pow t 2))>
#<alt (* 1/25000000000000000000000000000000 (pow t 2))>
#<alt (* 1/25000000000000000000000000000000 (pow t 2))>
#<alt (* 1/25000000000000000000000000000000 (pow t 2))>
#<alt (* 1/25000000000000000000000000000000 (pow t 2))>
#<alt (* 1/25000000000000000000000000000000 (pow t 2))>
#<alt (* 1/25000000000000000000000000000000 (pow t 2))>
#<alt (* 1/25000000000000000000000000000000 (pow t 2))>
#<alt (* 1/25000000000000000000000000000000 (pow t 2))>
#<alt (* 1/25000000000000000000000000000000 (pow t 2))>
#<alt (* 1/25000000000000000000000000000000 (pow t 2))>
#<alt (pow t 2)>
#<alt (pow t 2)>
#<alt (pow t 2)>
#<alt (pow t 2)>
#<alt (pow t 2)>
#<alt (pow t 2)>
#<alt (pow t 2)>
#<alt (pow t 2)>
#<alt (pow t 2)>
#<alt (pow t 2)>
#<alt (pow t 2)>
#<alt (pow t 2)>

Calls

6 calls:

Time	Variable		Point	Expression
1.0ms	t	@	0	(* (* t t) 1/25000000000000000000000000000000)
0.0ms	t	@	inf	(* t t)
0.0ms	t	@	inf	(* (* t t) 1/25000000000000000000000000000000)
0.0ms	t	@	-inf	(* (* t t) 1/25000000000000000000000000000000)
0.0ms	t	@	0	(* t t)

rewrite153.0ms (2.9%)

Algorithm

1×	batch-egg-rewrite

Rules

2200×	`pow-lowering-pow.f64`
2200×	`pow-lowering-pow.f32`
1432×	`prod-exp`
1304×	`-lowering-.f32`
1304×	`-lowering-.f64`

Iterations

Useful iterations: 5 (0.0ms)

Iter	Nodes	Cost
0	4	8
1	9	8
2	19	8
3	35	8
4	91	8
5	285	2
6	1844	2
7	1873	2
8	2003	2
9	3304	2
0	8662	2

Stop Event

1×	iter limit
1×	node limit

Counts

2 → 3

Calls

Call 1

Inputs
(* (* t t) 1/25000000000000000000000000000000)
(* t t)

Outputs
#s(literal 1/25000000000000000000000000000000 binary64)
#s(literal 1 binary64)
t

simplify5.0ms (0.1%)

Algorithm

1×	egg-herbie

Rules

14×	`-lowering-.f32`
14×	`-lowering-.f64`
6×	`*-commutative`
2×	`pow-lowering-pow.f64`
2×	`pow-lowering-pow.f32`

Iterations

Useful iterations: 0 (0.0ms)

Iter	Nodes	Cost
0	5	96
1	7	96
2	11	96
3	12	96
0	12	96

Stop Event

1×	iter limit
1×	saturated

Counts

24 → 24

Calls

Call 1

Inputs
(* 1/25000000000000000000000000000000 (pow t 2))
(* 1/25000000000000000000000000000000 (pow t 2))
(* 1/25000000000000000000000000000000 (pow t 2))
(* 1/25000000000000000000000000000000 (pow t 2))
(* 1/25000000000000000000000000000000 (pow t 2))
(* 1/25000000000000000000000000000000 (pow t 2))
(* 1/25000000000000000000000000000000 (pow t 2))
(* 1/25000000000000000000000000000000 (pow t 2))
(* 1/25000000000000000000000000000000 (pow t 2))
(* 1/25000000000000000000000000000000 (pow t 2))
(* 1/25000000000000000000000000000000 (pow t 2))
(* 1/25000000000000000000000000000000 (pow t 2))
(pow t 2)
(pow t 2)
(pow t 2)
(pow t 2)
(pow t 2)
(pow t 2)
(pow t 2)
(pow t 2)
(pow t 2)
(pow t 2)
(pow t 2)
(pow t 2)

Outputs
(* 1/25000000000000000000000000000000 (pow t 2))
(.f64 #s(literal 1/25000000000000000000000000000000 binary64) (.f64 t t))
(* 1/25000000000000000000000000000000 (pow t 2))
(.f64 #s(literal 1/25000000000000000000000000000000 binary64) (.f64 t t))
(* 1/25000000000000000000000000000000 (pow t 2))
(.f64 #s(literal 1/25000000000000000000000000000000 binary64) (.f64 t t))
(* 1/25000000000000000000000000000000 (pow t 2))
(.f64 #s(literal 1/25000000000000000000000000000000 binary64) (.f64 t t))
(* 1/25000000000000000000000000000000 (pow t 2))
(.f64 #s(literal 1/25000000000000000000000000000000 binary64) (.f64 t t))
(* 1/25000000000000000000000000000000 (pow t 2))
(.f64 #s(literal 1/25000000000000000000000000000000 binary64) (.f64 t t))
(* 1/25000000000000000000000000000000 (pow t 2))
(.f64 #s(literal 1/25000000000000000000000000000000 binary64) (.f64 t t))
(* 1/25000000000000000000000000000000 (pow t 2))
(.f64 #s(literal 1/25000000000000000000000000000000 binary64) (.f64 t t))
(* 1/25000000000000000000000000000000 (pow t 2))
(.f64 #s(literal 1/25000000000000000000000000000000 binary64) (.f64 t t))
(* 1/25000000000000000000000000000000 (pow t 2))
(.f64 #s(literal 1/25000000000000000000000000000000 binary64) (.f64 t t))
(* 1/25000000000000000000000000000000 (pow t 2))
(.f64 #s(literal 1/25000000000000000000000000000000 binary64) (.f64 t t))
(* 1/25000000000000000000000000000000 (pow t 2))
(.f64 #s(literal 1/25000000000000000000000000000000 binary64) (.f64 t t))
(pow t 2)
(*.f64 t t)
(pow t 2)
(*.f64 t t)
(pow t 2)
(*.f64 t t)
(pow t 2)
(*.f64 t t)
(pow t 2)
(*.f64 t t)
(pow t 2)
(*.f64 t t)
(pow t 2)
(*.f64 t t)
(pow t 2)
(*.f64 t t)
(pow t 2)
(*.f64 t t)
(pow t 2)
(*.f64 t t)
(pow t 2)
(*.f64 t t)
(pow t 2)
(*.f64 t t)

eval2.0ms (0%)

Compiler: Compiled 128 to 8 computations (93.8% saved)

prune7.0ms (0.1%)

Pruning

3 alts after pruning (2 fresh and 1 done)

	Pruned	Kept	Total
New	25	2	27
Fresh	0	0	0
Picked	0	1	1
Done	0	0	0
Total	25	3	28

Accuracy

99.4%

Counts

28 → 3

Alt Table

Click to see full alt table

Status	Accuracy	Program
✓	99.4%	(.f64 (.f64 t t) #s(literal 1/25000000000000000000000000000000 binary64))
▶	25.1%	(*.f64 t #s(literal 1/25000000000000000000000000000000 binary64))
▶	23.5%	#s(literal 1/25000000000000000000000000000000 binary64)

Compiler

Compiled 12 to 9 computations (25% saved)

simplify3.0ms (0.1%)

Algorithm

1×	egg-herbie

Localize:

Found 1 expressions of interest:

New	Metric	Score	Program
✓	cost-diff	0	(*.f64 t #s(literal 1/25000000000000000000000000000000 binary64))

Rules

4×	`-lowering-.f32`
4×	`-lowering-.f64`
2×	`*-commutative`

Iterations

Useful iterations: 0 (0.0ms)

Iter	Nodes	Cost
0	3	6
1	4	6
0	4	6

Stop Event

1×	iter limit
1×	saturated

Calls

Call 1

Inputs
1/25000000000000000000000000000000
(* t 1/25000000000000000000000000000000)
t
1/25000000000000000000000000000000

Outputs
1/25000000000000000000000000000000
#s(literal 1/25000000000000000000000000000000 binary64)
(* t 1/25000000000000000000000000000000)
(*.f64 #s(literal 1/25000000000000000000000000000000 binary64) t)
t
1/25000000000000000000000000000000
#s(literal 1/25000000000000000000000000000000 binary64)

localize21.0ms (0.4%)

Localize:

Found 1 expressions of interest:

New	Metric	Score	Program
✓	accuracy	99.3%	(*.f64 t #s(literal 1/25000000000000000000000000000000 binary64))

Results

15.0ms

256×

valid

Compiler

Compiled 8 to 4 computations (50% saved)

Precisions

Click to see histograms. Total time spent on operations: 7.0ms

...c/correct-round.rkt:119:19: 4.0ms (58.7% of total)

ival-mult: 2.0ms (29.3% of total)

const: 0.0ms (0% of total)

backward-pass: 0.0ms (0% of total)

series1.0ms (0%)

Counts

1 → 12

Calls

Call 1

Inputs
#<alt (* t 1/25000000000000000000000000000000)>

Outputs
#<alt (* 1/25000000000000000000000000000000 t)>
#<alt (* 1/25000000000000000000000000000000 t)>
#<alt (* 1/25000000000000000000000000000000 t)>
#<alt (* 1/25000000000000000000000000000000 t)>
#<alt (* 1/25000000000000000000000000000000 t)>
#<alt (* 1/25000000000000000000000000000000 t)>
#<alt (* 1/25000000000000000000000000000000 t)>
#<alt (* 1/25000000000000000000000000000000 t)>
#<alt (* 1/25000000000000000000000000000000 t)>
#<alt (* 1/25000000000000000000000000000000 t)>
#<alt (* 1/25000000000000000000000000000000 t)>
#<alt (* 1/25000000000000000000000000000000 t)>

Calls

3 calls:

Time	Variable		Point	Expression
0.0ms	t	@	0	(* t 1/25000000000000000000000000000000)
0.0ms	t	@	-inf	(* t 1/25000000000000000000000000000000)
0.0ms	t	@	inf	(* t 1/25000000000000000000000000000000)

rewrite8.0ms (0.2%)

Algorithm

1×	batch-egg-rewrite

Rules

4×	`-lowering-.f32`
4×	`-lowering-.f64`
2×	`*-commutative`

Iterations

Useful iterations: 0 (0.0ms)

Iter	Nodes	Cost
0	3	3
1	4	3
0	4	3

Stop Event

1×	iter limit
1×	saturated

Counts

1 → 2

Calls

Call 1

Inputs
(* t 1/25000000000000000000000000000000)

Outputs
(*.f64 t #s(literal 1/25000000000000000000000000000000 binary64))
(*.f64 #s(literal 1/25000000000000000000000000000000 binary64) t)

simplify4.0ms (0.1%)

Algorithm

1×	egg-herbie

Rules

4×	`-lowering-.f32`
4×	`-lowering-.f64`
2×	`*-commutative`

Iterations

Useful iterations: 0 (0.0ms)

Iter	Nodes	Cost
0	3	36
1	4	36
0	4	36

Stop Event

1×	iter limit
1×	saturated

Counts

12 → 12

Calls

Call 1

Inputs
(* 1/25000000000000000000000000000000 t)
(* 1/25000000000000000000000000000000 t)
(* 1/25000000000000000000000000000000 t)
(* 1/25000000000000000000000000000000 t)
(* 1/25000000000000000000000000000000 t)
(* 1/25000000000000000000000000000000 t)
(* 1/25000000000000000000000000000000 t)
(* 1/25000000000000000000000000000000 t)
(* 1/25000000000000000000000000000000 t)
(* 1/25000000000000000000000000000000 t)
(* 1/25000000000000000000000000000000 t)
(* 1/25000000000000000000000000000000 t)

Outputs
(* 1/25000000000000000000000000000000 t)
(*.f64 #s(literal 1/25000000000000000000000000000000 binary64) t)
(* 1/25000000000000000000000000000000 t)
(*.f64 #s(literal 1/25000000000000000000000000000000 binary64) t)
(* 1/25000000000000000000000000000000 t)
(*.f64 #s(literal 1/25000000000000000000000000000000 binary64) t)
(* 1/25000000000000000000000000000000 t)
(*.f64 #s(literal 1/25000000000000000000000000000000 binary64) t)
(* 1/25000000000000000000000000000000 t)
(*.f64 #s(literal 1/25000000000000000000000000000000 binary64) t)
(* 1/25000000000000000000000000000000 t)
(*.f64 #s(literal 1/25000000000000000000000000000000 binary64) t)
(* 1/25000000000000000000000000000000 t)
(*.f64 #s(literal 1/25000000000000000000000000000000 binary64) t)
(* 1/25000000000000000000000000000000 t)
(*.f64 #s(literal 1/25000000000000000000000000000000 binary64) t)
(* 1/25000000000000000000000000000000 t)
(*.f64 #s(literal 1/25000000000000000000000000000000 binary64) t)
(* 1/25000000000000000000000000000000 t)
(*.f64 #s(literal 1/25000000000000000000000000000000 binary64) t)
(* 1/25000000000000000000000000000000 t)
(*.f64 #s(literal 1/25000000000000000000000000000000 binary64) t)
(* 1/25000000000000000000000000000000 t)
(*.f64 #s(literal 1/25000000000000000000000000000000 binary64) t)

eval1.0ms (0%)

Compiler: Compiled 43 to 4 computations (90.7% saved)

prune4.0ms (0.1%)

Pruning

3 alts after pruning (0 fresh and 3 done)

	Pruned	Kept	Total
New	14	0	14
Fresh	0	0	0
Picked	0	2	2
Done	0	1	1
Total	14	3	17

Accuracy

99.4%

Counts

17 → 3

Alt Table

Click to see full alt table

Status	Accuracy	Program
✓	99.4%	(.f64 (.f64 t t) #s(literal 1/25000000000000000000000000000000 binary64))
✓	25.1%	(*.f64 t #s(literal 1/25000000000000000000000000000000 binary64))
✓	23.5%	#s(literal 1/25000000000000000000000000000000 binary64)

Compiler

Compiled 41 to 24 computations (41.5% saved)

regimes6.0ms (0.1%)

Counts

4 → 1

Calls

Call 1

Inputs
#s(literal 1/25000000000000000000000000000000 binary64)
(*.f64 t #s(literal 1/25000000000000000000000000000000 binary64))
(.f64 (.f64 t t) #s(literal 1/25000000000000000000000000000000 binary64))
(+.f64 (.f64 (+.f64 #s(literal 1 binary64) (.f64 t #s(literal 1/5000000000000000 binary64))) (+.f64 #s(literal 1 binary64) (.f64 t #s(literal 1/5000000000000000 binary64)))) (-.f64 #s(literal -1 binary64) (.f64 #s(literal 2 binary64) (*.f64 t #s(literal 1/5000000000000000 binary64)))))

Outputs
(.f64 (.f64 t t) #s(literal 1/25000000000000000000000000000000 binary64))

Calls

3 calls:

2.0ms

(*.f64 t #s(literal 1/5000000000000000 binary64))

2.0ms

1.0ms

(+.f64 (*.f64 (+.f64 #s(literal 1 binary64) (*.f64 t #s(literal 1/5000000000000000 binary64))) (+.f64 #s(literal 1 binary64) (*.f64 t #s(literal 1/5000000000000000 binary64)))) (-.f64 #s(literal -1 binary64) (*.f64 #s(literal 2 binary64) (*.f64 t #s(literal 1/5000000000000000 binary64)))))

Results

Accuracy	Segments	Branch
99.4%	1	`t`
99.4%	1	`(+.f64 (.f64 (+.f64 #s(literal 1 binary64) (.f64 t #s(literal 1/5000000000000000 binary64))) (+.f64 #s(literal 1 binary64) (.f64 t #s(literal 1/5000000000000000 binary64)))) (-.f64 #s(literal -1 binary64) (.f64 #s(literal 2 binary64) (*.f64 t #s(literal 1/5000000000000000 binary64)))))`
99.4%	1	`(*.f64 t #s(literal 1/5000000000000000 binary64))`

Compiler

Compiled 26 to 15 computations (42.3% saved)

regimes2.0ms (0%)

Counts

2 → 1

Calls

Call 1

Inputs
#s(literal 1/25000000000000000000000000000000 binary64)
(*.f64 t #s(literal 1/25000000000000000000000000000000 binary64))

Outputs
(*.f64 t #s(literal 1/25000000000000000000000000000000 binary64))

Calls

1 calls:

2.0ms

Results

Accuracy	Segments	Branch
25.1%	1	`t`

Compiler

Compiled 2 to 1 computations (50% saved)

regimes1.0ms (0%)

Accuracy

Total 0.0b remaining (0%)

Threshold costs 0b (0%)

Counts

1 → 1

Calls

Call 1

Inputs
#s(literal 1/25000000000000000000000000000000 binary64)

Outputs
#s(literal 1/25000000000000000000000000000000 binary64)

Calls

1 calls:

1.0ms

Results

Accuracy	Segments	Branch
23.5%	1	`t`

Compiler

Compiled 2 to 1 computations (50% saved)

simplify7.0ms (0.1%)

Algorithm

1×	egg-herbie

Rules

4×	`*-commutative_binary64`

Iterations

Useful iterations: 0 (0.0ms)

Iter	Nodes	Cost
0	5	9
1	7	9

Stop Event

1×

saturated

Calls

Call 1

Inputs
(.f64 (.f64 t t) #s(literal 1/25000000000000000000000000000000 binary64))
(*.f64 t #s(literal 1/25000000000000000000000000000000 binary64))
#s(literal 1/25000000000000000000000000000000 binary64)

Outputs
(.f64 (.f64 t t) #s(literal 1/25000000000000000000000000000000 binary64))
(*.f64 t #s(literal 1/25000000000000000000000000000000 binary64))
#s(literal 1/25000000000000000000000000000000 binary64)

soundness1.8s (34.2%)

Rules

2200×	`pow-lowering-pow.f64`
2200×	`pow-lowering-pow.f32`
1950×	`fnmadd-define`
1874×	`fma-define`
1732×	`+-lowering-+.f64`

Iterations

Useful iterations: 5 (0.0ms)

Iter	Nodes	Cost
0	4	8
1	9	8
2	19	8
3	35	8
4	91	8
5	285	2
6	1844	2
7	1873	2
8	2003	2
9	3304	2
0	8662	2
0	11	18
1	34	14
2	132	13
3	768	11
4	2239	5
5	5011	5
0	8223	5

Stop Event

1×	done
1×	iter limit
1×	node limit
1×	iter limit
1×	node limit

Compiler

Compiled 17 to 11 computations (35.3% saved)

preprocess109.0ms (2.1%)

Compiler: Compiled 98 to 52 computations (46.9% saved)

fma_test1

analyze7.0ms (0.1%)

sample2.7s (50.8%)

preprocess207.0ms (3.9%)

explain208.0ms (4%)

eval0.0ms (0%)

prune1.0ms (0%)

simplify3.0ms (0.1%)

localize26.0ms (0.5%)

series2.0ms (0%)

rewrite153.0ms (2.9%)

simplify5.0ms (0.1%)

eval2.0ms (0%)

prune7.0ms (0.1%)

simplify3.0ms (0.1%)

localize21.0ms (0.4%)

series1.0ms (0%)

rewrite8.0ms (0.2%)

simplify4.0ms (0.1%)

eval1.0ms (0%)

prune4.0ms (0.1%)

regimes6.0ms (0.1%)

regimes2.0ms (0%)

regimes1.0ms (0%)

simplify7.0ms (0.1%)

soundness1.8s (34.2%)

preprocess109.0ms (2.1%)

end0.0ms (0%)

Profiling