Time bar (total: 16.7s)
| 1× | search |
| True | Other | False | Iter |
|---|---|---|---|
| 0% | 0.6% | 99.4% | 0 |
| 0% | 0.6% | 99.4% | 1 |
| 0% | 0.6% | 99.4% | 2 |
| 0.3% | 0.3% | 99.4% | 3 |
| 0.3% | 0.3% | 99.4% | 4 |
| 0.5% | 0.2% | 99.4% | 5 |
| 0.5% | 0.2% | 99.4% | 6 |
| 0.6% | 0.1% | 99.4% | 7 |
| 0.6% | 0.1% | 99.4% | 8 |
| 0.6% | 0% | 99.4% | 9 |
| 0.6% | 0% | 99.4% | 10 |
| 0.6% | 0% | 99.4% | 11 |
| 0.6% | 0% | 99.4% | 12 |
| 0.6% | 0% | 99.4% | 13 |
| 0.6% | 0% | 99.4% | 14 |
Compiled 41 to 27 computations (34.1% saved)
| 3.2s | 8256× | body | 128 | valid |
Compiled 108 to 70 computations (35.2% saved)
| 1× | egg-herbie |
| 608× | associate-/l*_binary32 |
| 462× | associate-*l*_binary32 |
| 382× | associate-*l/_binary32 |
| 363× | associate-*r*_binary32 |
| 360× | distribute-lft-in_binary32 |
Useful iterations: 1 (0.0ms)
| Iter | Nodes | Cost |
|---|---|---|
| 0 | 14 | 33 |
| 1 | 45 | 31 |
| 2 | 160 | 31 |
| 3 | 661 | 31 |
| 4 | 2791 | 31 |
| 5 | 4866 | 31 |
| 6 | 4978 | 31 |
3 alts after pruning (3 fresh and 0 done)
| Pruned | Kept | Total | |
|---|---|---|---|
| New | 0 | 2 | 2 |
| Fresh | 0 | 1 | 1 |
| Picked | 0 | 0 | 0 |
| Done | 0 | 0 | 0 |
| Total | 0 | 3 | 3 |
| Status | Error | Program |
| 0.5b | (/.f32 (fma.f32 alpha alpha -1) (*.f32 (*.f32 (PI.f32) (log.f32 (*.f32 alpha alpha))) (fma.f32 (fma.f32 alpha alpha -1) (*.f32 cosTheta cosTheta) 1))) | |
| ▶ | 0.5b | (/.f32 (-.f32 (*.f32 alpha alpha) 1) (*.f32 (*.f32 (PI.f32) (log.f32 (*.f32 alpha alpha))) (+.f32 1 (*.f32 (*.f32 (-.f32 (*.f32 alpha alpha) 1) cosTheta) cosTheta)))) |
Compiled 145 to 87 computations (40% saved)
Found 4 expressions with local error:
| New | Error | Program |
| ✓ | 0.1b | (*.f32 (*.f32 (-.f32 (*.f32 alpha alpha) 1) cosTheta) cosTheta) |
| ✓ | 0.2b | (*.f32 (-.f32 (*.f32 alpha alpha) 1) cosTheta) |
| ✓ | 0.3b | (/.f32 (-.f32 (*.f32 alpha alpha) 1) (*.f32 (*.f32 (PI.f32) (log.f32 (*.f32 alpha alpha))) (+.f32 1 (*.f32 (*.f32 (-.f32 (*.f32 alpha alpha) 1) cosTheta) cosTheta)))) |
| ✓ | 0.4b | (*.f32 (PI.f32) (log.f32 (*.f32 alpha alpha))) |
4 calls:
| 149.0ms | (/.f32 (-.f32 (*.f32 alpha alpha) 1) (*.f32 (*.f32 (PI.f32) (log.f32 (*.f32 alpha alpha))) (+.f32 1 (*.f32 (*.f32 (-.f32 (*.f32 alpha alpha) 1) cosTheta) cosTheta)))) |
| 54.0ms | (*.f32 (PI.f32) (log.f32 (*.f32 alpha alpha))) |
| 24.0ms | (*.f32 (*.f32 (-.f32 (*.f32 alpha alpha) 1) cosTheta) cosTheta) |
| 17.0ms | (*.f32 (-.f32 (*.f32 alpha alpha) 1) cosTheta) |
| 2× | batch-egg-rewrite |
| 266× | log1p-udef_binary32 |
| 265× | expm1-udef_binary32 |
| 154× | add-sqr-sqrt_binary32 |
| 146× | log1p-expm1-u_binary32 |
| 146× | expm1-log1p-u_binary32 |
4 calls:
| 138.0ms | (*.f32 (*.f32 (-.f32 (*.f32 alpha alpha) 1) cosTheta) cosTheta) |
| 138.0ms | (*.f32 (-.f32 (*.f32 alpha alpha) 1) cosTheta) |
| 138.0ms | (/.f32 (-.f32 (*.f32 alpha alpha) 1) (*.f32 (*.f32 (PI.f32) (log.f32 (*.f32 alpha alpha))) (+.f32 1 (*.f32 (*.f32 (-.f32 (*.f32 alpha alpha) 1) cosTheta) cosTheta)))) |
| 137.0ms | (*.f32 (PI.f32) (log.f32 (*.f32 alpha alpha))) |
Useful iterations: 0 (0.0ms)
| Iter | Nodes | Cost |
|---|---|---|
| 0 | 14 | 62 |
| 1 | 310 | 60 |
| 2 | 4242 | 60 |
| 3 | 4972 | 60 |
| 0 | 0 | 0 |
| 1 | 0 | 0 |
| 1× | egg-herbie |
| 335× | associate-/r*_binary32 |
| 303× | cancel-sign-sub-inv_binary32 |
| 267× | associate-*r*_binary32 |
| 258× | fma-neg_binary32 |
| 223× | sub-neg_binary32 |
Useful iterations: 2 (0.0ms)
| Iter | Nodes | Cost |
|---|---|---|
| 0 | 264 | 3015 |
| 1 | 865 | 2707 |
| 2 | 3733 | 2705 |
| 3 | 4989 | 2705 |
11 alts after pruning (11 fresh and 0 done)
| Pruned | Kept | Total | |
|---|---|---|---|
| New | 154 | 11 | 165 |
| Fresh | 1 | 0 | 1 |
| Picked | 1 | 0 | 1 |
| Done | 0 | 0 | 0 |
| Total | 156 | 11 | 167 |
| Status | Error | Program |
| ▶ | 0.4b | (/.f32 (-.f32 (*.f32 alpha alpha) 1) (*.f32 (log.f32 (pow.f32 (*.f32 alpha alpha) (PI.f32))) (+.f32 1 (*.f32 (*.f32 (-.f32 (*.f32 alpha alpha) 1) cosTheta) cosTheta)))) |
| 0.7b | (*.f32 (/.f32 (pow.f32 (cbrt.f32 (fma.f32 alpha alpha -1)) 2) (PI.f32)) (/.f32 (cbrt.f32 (fma.f32 alpha alpha -1)) (*.f32 (log.f32 (*.f32 alpha alpha)) (fma.f32 (fma.f32 alpha alpha -1) (*.f32 cosTheta cosTheta) 1)))) | |
| 0.6b | (*.f32 (+.f32 alpha 1) (*.f32 (+.f32 alpha -1) (/.f32 1 (*.f32 (*.f32 (PI.f32) (log.f32 (*.f32 alpha alpha))) (fma.f32 (fma.f32 alpha alpha -1) (*.f32 cosTheta cosTheta) 1))))) | |
| 0.6b | (cbrt.f32 (/.f32 (pow.f32 (fma.f32 alpha alpha -1) 3) (pow.f32 (*.f32 (*.f32 (PI.f32) (log.f32 (*.f32 alpha alpha))) (fma.f32 (fma.f32 alpha alpha -1) (*.f32 cosTheta cosTheta) 1)) 3))) | |
| 0.8b | (fma.f32 2 (/.f32 (*.f32 (*.f32 cosTheta cosTheta) (*.f32 alpha alpha)) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha)))) (+.f32 (fma.f32 4 (/.f32 (*.f32 (pow.f32 alpha 6) (pow.f32 cosTheta 6)) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha)))) (fma.f32 3 (*.f32 (/.f32 (*.f32 alpha alpha) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha)))) (pow.f32 cosTheta 4)) (fma.f32 4 (/.f32 (*.f32 (*.f32 alpha alpha) (pow.f32 cosTheta 6)) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha)))) (/.f32 (*.f32 (pow.f32 cosTheta 4) (pow.f32 alpha 6)) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha))))))) (fma.f32 (/.f32 alpha (PI.f32)) (/.f32 alpha (*.f32 2 (log.f32 alpha))) (-.f32 (/.f32 (/.f32 -1/2 (log.f32 alpha)) (PI.f32)) (+.f32 (fma.f32 (/.f32 (pow.f32 cosTheta 6) (*.f32 2 (log.f32 alpha))) (/.f32 (pow.f32 alpha 8) (PI.f32)) (fma.f32 6 (/.f32 (*.f32 (pow.f32 alpha 4) (pow.f32 cosTheta 6)) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha)))) (+.f32 (/.f32 (pow.f32 cosTheta 6) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha)))) (fma.f32 3 (/.f32 (*.f32 (pow.f32 alpha 4) (pow.f32 cosTheta 4)) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha)))) (/.f32 (*.f32 cosTheta cosTheta) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha)))))))) (fma.f32 (/.f32 (*.f32 cosTheta cosTheta) (*.f32 2 (log.f32 alpha))) (/.f32 (pow.f32 alpha 4) (PI.f32)) (/.f32 (pow.f32 cosTheta 4) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha)))))))))) | |
| 0.7b | (pow.f32 (sqrt.f32 (/.f32 (fma.f32 alpha alpha -1) (*.f32 (*.f32 (PI.f32) (log.f32 (*.f32 alpha alpha))) (fma.f32 (fma.f32 alpha alpha -1) (*.f32 cosTheta cosTheta) 1)))) 2) | |
| 0.6b | (*.f32 (/.f32 (+.f32 alpha 1) (PI.f32)) (/.f32 (+.f32 alpha -1) (*.f32 (log.f32 (*.f32 alpha alpha)) (fma.f32 (fma.f32 alpha alpha -1) (*.f32 cosTheta cosTheta) 1)))) | |
| 0.5b | (/.f32 (-.f32 (*.f32 alpha alpha) 1) (*.f32 (*.f32 (PI.f32) (log.f32 (*.f32 alpha alpha))) (+.f32 1 (log.f32 (pow.f32 (exp.f32 (fma.f32 alpha alpha -1)) (*.f32 cosTheta cosTheta)))))) | |
| 0.8b | (*.f32 (/.f32 (pow.f32 (cbrt.f32 (fma.f32 alpha alpha -1)) 2) (pow.f32 (cbrt.f32 (*.f32 (*.f32 (PI.f32) (log.f32 (*.f32 alpha alpha))) (fma.f32 (fma.f32 alpha alpha -1) (*.f32 cosTheta cosTheta) 1))) 2)) (cbrt.f32 (/.f32 (fma.f32 alpha alpha -1) (*.f32 (*.f32 (PI.f32) (log.f32 (*.f32 alpha alpha))) (fma.f32 (fma.f32 alpha alpha -1) (*.f32 cosTheta cosTheta) 1))))) | |
| 0.5b | (/.f32 (-.f32 (*.f32 alpha alpha) 1) (*.f32 (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha))) (+.f32 1 (*.f32 (*.f32 (-.f32 (*.f32 alpha alpha) 1) cosTheta) cosTheta)))) | |
| 0.9b | (fma.f32 2 (/.f32 (*.f32 (*.f32 cosTheta cosTheta) (*.f32 alpha alpha)) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha)))) (-.f32 (/.f32 (fma.f32 alpha alpha -1) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha)))) (fma.f32 (/.f32 (*.f32 cosTheta cosTheta) (*.f32 2 (log.f32 alpha))) (/.f32 (pow.f32 alpha 4) (PI.f32)) (/.f32 (*.f32 cosTheta cosTheta) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha))))))) |
Compiled 7061 to 3951 computations (44% saved)
Found 4 expressions with local error:
| New | Error | Program |
| 0.1b | (*.f32 (*.f32 (-.f32 (*.f32 alpha alpha) 1) cosTheta) cosTheta) | |
| 0.2b | (*.f32 (-.f32 (*.f32 alpha alpha) 1) cosTheta) | |
| ✓ | 0.3b | (/.f32 (-.f32 (*.f32 alpha alpha) 1) (*.f32 (log.f32 (pow.f32 (*.f32 alpha alpha) (PI.f32))) (+.f32 1 (*.f32 (*.f32 (-.f32 (*.f32 alpha alpha) 1) cosTheta) cosTheta)))) |
| ✓ | 3.2b | (pow.f32 (*.f32 alpha alpha) (PI.f32)) |
2 calls:
| 152.0ms | (/.f32 (-.f32 (*.f32 alpha alpha) 1) (*.f32 (log.f32 (pow.f32 (*.f32 alpha alpha) (PI.f32))) (+.f32 1 (*.f32 (*.f32 (-.f32 (*.f32 alpha alpha) 1) cosTheta) cosTheta)))) |
| 49.0ms | (pow.f32 (*.f32 alpha alpha) (PI.f32)) |
| 2× | batch-egg-rewrite |
| 379× | fma-neg_binary32 |
| 266× | log1p-udef_binary32 |
| 265× | expm1-udef_binary32 |
| 156× | add-sqr-sqrt_binary32 |
| 149× | log1p-expm1-u_binary32 |
2 calls:
| 155.0ms | (/.f32 (-.f32 (*.f32 alpha alpha) 1) (*.f32 (log.f32 (pow.f32 (*.f32 alpha alpha) (PI.f32))) (+.f32 1 (*.f32 (*.f32 (-.f32 (*.f32 alpha alpha) 1) cosTheta) cosTheta)))) |
| 155.0ms | (pow.f32 (*.f32 alpha alpha) (PI.f32)) |
Useful iterations: 0 (0.0ms)
| Iter | Nodes | Cost |
|---|---|---|
| 0 | 14 | 41 |
| 1 | 314 | 39 |
| 2 | 4188 | 39 |
| 3 | 4991 | 39 |
| 4 | 5302 | 39 |
| 0 | 0 | 0 |
| 1 | 0 | 0 |
| 1× | egg-herbie |
| 335× | associate-/r*_binary32 |
| 300× | cancel-sign-sub-inv_binary32 |
| 265× | associate-*r*_binary32 |
| 246× | fma-neg_binary32 |
| 220× | sub-neg_binary32 |
Useful iterations: 2 (0.0ms)
| Iter | Nodes | Cost |
|---|---|---|
| 0 | 259 | 2450 |
| 1 | 856 | 2289 |
| 2 | 3741 | 2243 |
| 3 | 5049 | 2243 |
10 alts after pruning (9 fresh and 1 done)
| Pruned | Kept | Total | |
|---|---|---|---|
| New | 177 | 2 | 179 |
| Fresh | 3 | 7 | 10 |
| Picked | 0 | 1 | 1 |
| Done | 0 | 0 | 0 |
| Total | 180 | 10 | 190 |
| Status | Error | Program |
| 0.8b | (fma.f32 2 (/.f32 (*.f32 (*.f32 cosTheta cosTheta) (*.f32 alpha alpha)) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha)))) (+.f32 (fma.f32 4 (/.f32 (*.f32 (pow.f32 alpha 6) (pow.f32 cosTheta 6)) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha)))) (fma.f32 3 (*.f32 (/.f32 (*.f32 alpha alpha) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha)))) (pow.f32 cosTheta 4)) (fma.f32 4 (/.f32 (*.f32 (*.f32 alpha alpha) (pow.f32 cosTheta 6)) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha)))) (/.f32 (*.f32 (pow.f32 cosTheta 4) (pow.f32 alpha 6)) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha))))))) (fma.f32 (/.f32 alpha (PI.f32)) (/.f32 alpha (*.f32 2 (log.f32 alpha))) (-.f32 (/.f32 (/.f32 -1/2 (log.f32 alpha)) (PI.f32)) (+.f32 (fma.f32 (/.f32 (pow.f32 cosTheta 6) (*.f32 2 (log.f32 alpha))) (/.f32 (pow.f32 alpha 8) (PI.f32)) (fma.f32 6 (/.f32 (*.f32 (pow.f32 alpha 4) (pow.f32 cosTheta 6)) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha)))) (+.f32 (/.f32 (pow.f32 cosTheta 6) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha)))) (fma.f32 3 (/.f32 (*.f32 (pow.f32 alpha 4) (pow.f32 cosTheta 4)) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha)))) (/.f32 (*.f32 cosTheta cosTheta) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha)))))))) (fma.f32 (/.f32 (*.f32 cosTheta cosTheta) (*.f32 2 (log.f32 alpha))) (/.f32 (pow.f32 alpha 4) (PI.f32)) (/.f32 (pow.f32 cosTheta 4) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha)))))))))) | |
| 0.6b | (*.f32 (+.f32 alpha 1) (*.f32 (+.f32 alpha -1) (/.f32 1 (*.f32 (log.f32 (pow.f32 (*.f32 alpha alpha) (PI.f32))) (fma.f32 (fma.f32 alpha alpha -1) (*.f32 cosTheta cosTheta) 1))))) | |
| 0.6b | (*.f32 (/.f32 (+.f32 alpha 1) (PI.f32)) (/.f32 (+.f32 alpha -1) (*.f32 (log.f32 (*.f32 alpha alpha)) (fma.f32 (fma.f32 alpha alpha -1) (*.f32 cosTheta cosTheta) 1)))) | |
| ✓ | 0.4b | (/.f32 (-.f32 (*.f32 alpha alpha) 1) (*.f32 (log.f32 (pow.f32 (*.f32 alpha alpha) (PI.f32))) (+.f32 1 (*.f32 (*.f32 (-.f32 (*.f32 alpha alpha) 1) cosTheta) cosTheta)))) |
| 0.6b | (cbrt.f32 (/.f32 (pow.f32 (fma.f32 alpha alpha -1) 3) (pow.f32 (*.f32 (*.f32 (PI.f32) (log.f32 (*.f32 alpha alpha))) (fma.f32 (fma.f32 alpha alpha -1) (*.f32 cosTheta cosTheta) 1)) 3))) | |
| ▶ | 0.5b | (/.f32 (-.f32 (*.f32 alpha alpha) 1) (*.f32 (*.f32 (PI.f32) (log.f32 (*.f32 alpha alpha))) (+.f32 1 (log.f32 (pow.f32 (exp.f32 (fma.f32 alpha alpha -1)) (*.f32 cosTheta cosTheta)))))) |
| 0.8b | (*.f32 (/.f32 (pow.f32 (cbrt.f32 (fma.f32 alpha alpha -1)) 2) (pow.f32 (cbrt.f32 (*.f32 (*.f32 (PI.f32) (log.f32 (*.f32 alpha alpha))) (fma.f32 (fma.f32 alpha alpha -1) (*.f32 cosTheta cosTheta) 1))) 2)) (cbrt.f32 (/.f32 (fma.f32 alpha alpha -1) (*.f32 (*.f32 (PI.f32) (log.f32 (*.f32 alpha alpha))) (fma.f32 (fma.f32 alpha alpha -1) (*.f32 cosTheta cosTheta) 1))))) | |
| 0.5b | (/.f32 (-.f32 (*.f32 alpha alpha) 1) (*.f32 (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha))) (+.f32 1 (*.f32 (*.f32 (-.f32 (*.f32 alpha alpha) 1) cosTheta) cosTheta)))) | |
| 0.9b | (fma.f32 2 (/.f32 (*.f32 (*.f32 cosTheta cosTheta) (*.f32 alpha alpha)) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha)))) (-.f32 (/.f32 (fma.f32 alpha alpha -1) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha)))) (fma.f32 (/.f32 (*.f32 cosTheta cosTheta) (*.f32 2 (log.f32 alpha))) (/.f32 (pow.f32 alpha 4) (PI.f32)) (/.f32 (*.f32 cosTheta cosTheta) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha))))))) | |
| 0.7b | (pow.f32 (sqrt.f32 (/.f32 (fma.f32 alpha alpha -1) (*.f32 (log.f32 (pow.f32 (*.f32 alpha alpha) (PI.f32))) (fma.f32 (fma.f32 alpha alpha -1) (*.f32 cosTheta cosTheta) 1)))) 2) |
Compiled 7812 to 4249 computations (45.6% saved)
Found 4 expressions with local error:
| New | Error | Program |
| ✓ | 0.2b | (exp.f32 (fma.f32 alpha alpha -1)) |
| ✓ | 0.3b | (/.f32 (-.f32 (*.f32 alpha alpha) 1) (*.f32 (*.f32 (PI.f32) (log.f32 (*.f32 alpha alpha))) (+.f32 1 (log.f32 (pow.f32 (exp.f32 (fma.f32 alpha alpha -1)) (*.f32 cosTheta cosTheta)))))) |
| 0.4b | (*.f32 (PI.f32) (log.f32 (*.f32 alpha alpha))) | |
| ✓ | 13.3b | (log.f32 (pow.f32 (exp.f32 (fma.f32 alpha alpha -1)) (*.f32 cosTheta cosTheta))) |
3 calls:
| 3.5s | (log.f32 (pow.f32 (exp.f32 (fma.f32 alpha alpha -1)) (*.f32 cosTheta cosTheta))) |
| 1.9s | (/.f32 (-.f32 (*.f32 alpha alpha) 1) (*.f32 (*.f32 (PI.f32) (log.f32 (*.f32 alpha alpha))) (+.f32 1 (log.f32 (pow.f32 (exp.f32 (fma.f32 alpha alpha -1)) (*.f32 cosTheta cosTheta)))))) |
| 4.0ms | (exp.f32 (fma.f32 alpha alpha -1)) |
| 2× | batch-egg-rewrite |
| 304× | log1p-udef_binary32 |
| 176× | add-sqr-sqrt_binary32 |
| 168× | log1p-expm1-u_binary32 |
| 168× | expm1-log1p-u_binary32 |
| 166× | add-log-exp_binary32 |
3 calls:
| 164.0ms | (exp.f32 (fma.f32 alpha alpha -1)) |
| 164.0ms | (/.f32 (-.f32 (*.f32 alpha alpha) 1) (*.f32 (*.f32 (PI.f32) (log.f32 (*.f32 alpha alpha))) (+.f32 1 (log.f32 (pow.f32 (exp.f32 (fma.f32 alpha alpha -1)) (*.f32 cosTheta cosTheta)))))) |
| 164.0ms | (log.f32 (pow.f32 (exp.f32 (fma.f32 alpha alpha -1)) (*.f32 cosTheta cosTheta))) |
Useful iterations: 0 (0.0ms)
| Iter | Nodes | Cost |
|---|---|---|
| 0 | 18 | 51 |
| 1 | 361 | 49 |
| 2 | 4726 | 49 |
| 3 | 5171 | 49 |
| 0 | 0 | 0 |
| 1 | 0 | 0 |
| 1× | egg-herbie |
| 398× | fma-neg_binary32 |
| 264× | cancel-sign-sub-inv_binary32 |
| 247× | associate-/l/_binary32 |
| 222× | times-frac_binary32 |
| 217× | associate-/r*_binary32 |
Useful iterations: 3 (0.0ms)
| Iter | Nodes | Cost |
|---|---|---|
| 0 | 176 | 2258 |
| 1 | 590 | 2042 |
| 2 | 2684 | 1939 |
| 3 | 4752 | 1933 |
| 4 | 5000 | 1933 |
| 5 | 4975 | 1933 |
10 alts after pruning (8 fresh and 2 done)
| Pruned | Kept | Total | |
|---|---|---|---|
| New | 175 | 1 | 176 |
| Fresh | 1 | 7 | 8 |
| Picked | 0 | 1 | 1 |
| Done | 0 | 1 | 1 |
| Total | 176 | 10 | 186 |
| Status | Error | Program |
| 0.8b | (fma.f32 2 (/.f32 (*.f32 (*.f32 cosTheta cosTheta) (*.f32 alpha alpha)) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha)))) (+.f32 (fma.f32 4 (/.f32 (*.f32 (pow.f32 alpha 6) (pow.f32 cosTheta 6)) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha)))) (fma.f32 3 (*.f32 (/.f32 (*.f32 alpha alpha) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha)))) (pow.f32 cosTheta 4)) (fma.f32 4 (/.f32 (*.f32 (*.f32 alpha alpha) (pow.f32 cosTheta 6)) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha)))) (/.f32 (*.f32 (pow.f32 cosTheta 4) (pow.f32 alpha 6)) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha))))))) (fma.f32 (/.f32 alpha (PI.f32)) (/.f32 alpha (*.f32 2 (log.f32 alpha))) (-.f32 (/.f32 (/.f32 -1/2 (log.f32 alpha)) (PI.f32)) (+.f32 (fma.f32 (/.f32 (pow.f32 cosTheta 6) (*.f32 2 (log.f32 alpha))) (/.f32 (pow.f32 alpha 8) (PI.f32)) (fma.f32 6 (/.f32 (*.f32 (pow.f32 alpha 4) (pow.f32 cosTheta 6)) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha)))) (+.f32 (/.f32 (pow.f32 cosTheta 6) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha)))) (fma.f32 3 (/.f32 (*.f32 (pow.f32 alpha 4) (pow.f32 cosTheta 4)) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha)))) (/.f32 (*.f32 cosTheta cosTheta) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha)))))))) (fma.f32 (/.f32 (*.f32 cosTheta cosTheta) (*.f32 2 (log.f32 alpha))) (/.f32 (pow.f32 alpha 4) (PI.f32)) (/.f32 (pow.f32 cosTheta 4) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha)))))))))) | |
| 0.6b | (*.f32 (+.f32 alpha 1) (*.f32 (+.f32 alpha -1) (/.f32 1 (*.f32 (log.f32 (pow.f32 (*.f32 alpha alpha) (PI.f32))) (fma.f32 (fma.f32 alpha alpha -1) (*.f32 cosTheta cosTheta) 1))))) | |
| 0.6b | (*.f32 (/.f32 (+.f32 alpha 1) (PI.f32)) (/.f32 (+.f32 alpha -1) (*.f32 (log.f32 (*.f32 alpha alpha)) (fma.f32 (fma.f32 alpha alpha -1) (*.f32 cosTheta cosTheta) 1)))) | |
| ✓ | 0.4b | (/.f32 (-.f32 (*.f32 alpha alpha) 1) (*.f32 (log.f32 (pow.f32 (*.f32 alpha alpha) (PI.f32))) (+.f32 1 (*.f32 (*.f32 (-.f32 (*.f32 alpha alpha) 1) cosTheta) cosTheta)))) |
| 0.6b | (cbrt.f32 (/.f32 (pow.f32 (fma.f32 alpha alpha -1) 3) (pow.f32 (*.f32 (*.f32 (PI.f32) (log.f32 (*.f32 alpha alpha))) (fma.f32 (fma.f32 alpha alpha -1) (*.f32 cosTheta cosTheta) 1)) 3))) | |
| ✓ | 0.5b | (/.f32 (-.f32 (*.f32 alpha alpha) 1) (*.f32 (*.f32 (PI.f32) (log.f32 (*.f32 alpha alpha))) (+.f32 1 (log.f32 (pow.f32 (exp.f32 (fma.f32 alpha alpha -1)) (*.f32 cosTheta cosTheta)))))) |
| 0.8b | (*.f32 (/.f32 (pow.f32 (cbrt.f32 (fma.f32 alpha alpha -1)) 2) (pow.f32 (cbrt.f32 (*.f32 (*.f32 (PI.f32) (log.f32 (*.f32 alpha alpha))) (fma.f32 (fma.f32 alpha alpha -1) (*.f32 cosTheta cosTheta) 1))) 2)) (cbrt.f32 (/.f32 (fma.f32 alpha alpha -1) (*.f32 (*.f32 (PI.f32) (log.f32 (*.f32 alpha alpha))) (fma.f32 (fma.f32 alpha alpha -1) (*.f32 cosTheta cosTheta) 1))))) | |
| 0.5b | (/.f32 (-.f32 (*.f32 alpha alpha) 1) (*.f32 (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha))) (+.f32 1 (*.f32 (*.f32 (-.f32 (*.f32 alpha alpha) 1) cosTheta) cosTheta)))) | |
| ▶ | 0.5b | (/.f32 (fma.f32 alpha alpha -1) (*.f32 (fma.f32 (*.f32 cosTheta cosTheta) (fma.f32 alpha alpha -1) 1) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha))))) |
| 0.7b | (pow.f32 (sqrt.f32 (/.f32 (fma.f32 alpha alpha -1) (*.f32 (log.f32 (pow.f32 (*.f32 alpha alpha) (PI.f32))) (fma.f32 (fma.f32 alpha alpha -1) (*.f32 cosTheta cosTheta) 1)))) 2) |
Compiled 6801 to 3631 computations (46.6% saved)
Found 4 expressions with local error:
| New | Error | Program |
| ✓ | 0.0b | (*.f32 (fma.f32 (*.f32 cosTheta cosTheta) (fma.f32 alpha alpha -1) 1) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha)))) |
| ✓ | 0.0b | (log.f32 alpha) |
| ✓ | 0.3b | (/.f32 (fma.f32 alpha alpha -1) (*.f32 (fma.f32 (*.f32 cosTheta cosTheta) (fma.f32 alpha alpha -1) 1) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha))))) |
| ✓ | 0.4b | (*.f32 (PI.f32) (log.f32 alpha)) |
4 calls:
| 384.0ms | (/.f32 (fma.f32 alpha alpha -1) (*.f32 (fma.f32 (*.f32 cosTheta cosTheta) (fma.f32 alpha alpha -1) 1) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha))))) |
| 191.0ms | (*.f32 (fma.f32 (*.f32 cosTheta cosTheta) (fma.f32 alpha alpha -1) 1) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha)))) |
| 47.0ms | (*.f32 (PI.f32) (log.f32 alpha)) |
| 40.0ms | (log.f32 alpha) |
| 2× | batch-egg-rewrite |
| 751× | log-prod_binary32 |
| 250× | expm1-udef_binary32 |
| 250× | log1p-udef_binary32 |
| 224× | log-pow_binary32 |
| 142× | add-sqr-sqrt_binary32 |
4 calls:
| 121.0ms | (*.f32 (fma.f32 (*.f32 cosTheta cosTheta) (fma.f32 alpha alpha -1) 1) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha)))) |
| 121.0ms | (log.f32 alpha) |
| 121.0ms | (/.f32 (fma.f32 alpha alpha -1) (*.f32 (fma.f32 (*.f32 cosTheta cosTheta) (fma.f32 alpha alpha -1) 1) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha))))) |
| 121.0ms | (*.f32 (PI.f32) (log.f32 alpha)) |
Useful iterations: 0 (0.0ms)
| Iter | Nodes | Cost |
|---|---|---|
| 0 | 15 | 65 |
| 1 | 294 | 65 |
| 2 | 3470 | 65 |
| 3 | 6042 | 65 |
| 0 | 0 | 0 |
| 1 | 0 | 0 |
| 1× | egg-herbie |
| 728× | associate-/r*_binary32 |
| 527× | times-frac_binary32 |
| 463× | fma-def_binary32 |
| 330× | associate-/l*_binary32 |
| 196× | *-commutative_binary32 |
Useful iterations: 1 (0.0ms)
| Iter | Nodes | Cost |
|---|---|---|
| 0 | 473 | 6077 |
| 1 | 1633 | 5808 |
| 2 | 5142 | 5808 |
10 alts after pruning (8 fresh and 2 done)
| Pruned | Kept | Total | |
|---|---|---|---|
| New | 173 | 2 | 175 |
| Fresh | 1 | 6 | 7 |
| Picked | 1 | 0 | 1 |
| Done | 0 | 2 | 2 |
| Total | 175 | 10 | 185 |
| Status | Error | Program |
| ✓ | 0.4b | (/.f32 (-.f32 (*.f32 alpha alpha) 1) (*.f32 (log.f32 (pow.f32 (*.f32 alpha alpha) (PI.f32))) (+.f32 1 (*.f32 (*.f32 (-.f32 (*.f32 alpha alpha) 1) cosTheta) cosTheta)))) |
| 0.5b | (expm1.f32 (log1p.f32 (/.f32 (fma.f32 alpha alpha -1) (*.f32 (fma.f32 (fma.f32 alpha alpha -1) (*.f32 cosTheta cosTheta) 1) (*.f32 (*.f32 (PI.f32) (log.f32 alpha)) 2))))) | |
| 0.8b | (fma.f32 2 (/.f32 (*.f32 (*.f32 cosTheta cosTheta) (*.f32 alpha alpha)) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha)))) (+.f32 (fma.f32 4 (/.f32 (*.f32 (pow.f32 alpha 6) (pow.f32 cosTheta 6)) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha)))) (fma.f32 3 (*.f32 (/.f32 (*.f32 alpha alpha) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha)))) (pow.f32 cosTheta 4)) (fma.f32 4 (/.f32 (*.f32 (*.f32 alpha alpha) (pow.f32 cosTheta 6)) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha)))) (/.f32 (*.f32 (pow.f32 cosTheta 4) (pow.f32 alpha 6)) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha))))))) (fma.f32 (/.f32 alpha (PI.f32)) (/.f32 alpha (*.f32 2 (log.f32 alpha))) (-.f32 (/.f32 (/.f32 -1/2 (log.f32 alpha)) (PI.f32)) (+.f32 (fma.f32 (/.f32 (pow.f32 cosTheta 6) (*.f32 2 (log.f32 alpha))) (/.f32 (pow.f32 alpha 8) (PI.f32)) (fma.f32 6 (/.f32 (*.f32 (pow.f32 alpha 4) (pow.f32 cosTheta 6)) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha)))) (+.f32 (/.f32 (pow.f32 cosTheta 6) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha)))) (fma.f32 3 (/.f32 (*.f32 (pow.f32 alpha 4) (pow.f32 cosTheta 4)) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha)))) (/.f32 (*.f32 cosTheta cosTheta) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha)))))))) (fma.f32 (/.f32 (*.f32 cosTheta cosTheta) (*.f32 2 (log.f32 alpha))) (/.f32 (pow.f32 alpha 4) (PI.f32)) (/.f32 (pow.f32 cosTheta 4) (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha)))))))))) | |
| 0.6b | (*.f32 (+.f32 alpha 1) (*.f32 (+.f32 alpha -1) (/.f32 1 (*.f32 (log.f32 (pow.f32 (*.f32 alpha alpha) (PI.f32))) (fma.f32 (fma.f32 alpha alpha -1) (*.f32 cosTheta cosTheta) 1))))) | |
| 0.6b | (/.f32 (fma.f32 alpha alpha -1) (*.f32 (fma.f32 (*.f32 cosTheta cosTheta) (fma.f32 alpha alpha -1) 1) (*.f32 2 (+.f32 (log.f32 (*.f32 (cbrt.f32 (pow.f32 alpha (PI.f32))) (cbrt.f32 (pow.f32 alpha (PI.f32))))) (log.f32 (cbrt.f32 (pow.f32 alpha (PI.f32)))))))) | |
| 0.6b | (*.f32 (/.f32 (+.f32 alpha 1) (PI.f32)) (/.f32 (+.f32 alpha -1) (*.f32 (log.f32 (*.f32 alpha alpha)) (fma.f32 (fma.f32 alpha alpha -1) (*.f32 cosTheta cosTheta) 1)))) | |
| 0.6b | (cbrt.f32 (/.f32 (pow.f32 (fma.f32 alpha alpha -1) 3) (pow.f32 (*.f32 (*.f32 (PI.f32) (log.f32 (*.f32 alpha alpha))) (fma.f32 (fma.f32 alpha alpha -1) (*.f32 cosTheta cosTheta) 1)) 3))) | |
| ✓ | 0.5b | (/.f32 (-.f32 (*.f32 alpha alpha) 1) (*.f32 (*.f32 (PI.f32) (log.f32 (*.f32 alpha alpha))) (+.f32 1 (log.f32 (pow.f32 (exp.f32 (fma.f32 alpha alpha -1)) (*.f32 cosTheta cosTheta)))))) |
| 0.5b | (/.f32 (-.f32 (*.f32 alpha alpha) 1) (*.f32 (*.f32 2 (*.f32 (PI.f32) (log.f32 alpha))) (+.f32 1 (*.f32 (*.f32 (-.f32 (*.f32 alpha alpha) 1) cosTheta) cosTheta)))) | |
| 0.7b | (pow.f32 (sqrt.f32 (/.f32 (fma.f32 alpha alpha -1) (*.f32 (log.f32 (pow.f32 (*.f32 alpha alpha) (PI.f32))) (fma.f32 (fma.f32 alpha alpha -1) (*.f32 cosTheta cosTheta) 1)))) 2) |
Compiled 9920 to 6479 computations (34.7% saved)
Total 0.4b remaining (89%)
Threshold costs 0.4b (89%)
Compiled 26192 to 17662 computations (32.6% saved)
| 1× | egg-herbie |
| 3× | *-commutative_binary32 |
| 2× | +-commutative_binary32 |
| 1× | sub-neg_binary32 |
| 1× | 1-exp_binary32 |
Useful iterations: 0 (0.0ms)
| Iter | Nodes | Cost |
|---|---|---|
| 0 | 16 | 35 |
| 1 | 24 | 35 |
| 2 | 25 | 35 |
| 3 | 23 | 35 |
Compiled 487 to 306 computations (37.2% saved)
Loading profile data...