Closed cortner closed 1 year ago
this is ready to merge as well, we can do both together tomorrow.
I've partially fixed the performance now. There is probably more to be found, but I'm happy for now. This is ready to merge for me. @CheukHinHoJerry please confirm.
Test 1: with a minimal model
Nel = 4
polys = legendre_basis(8)
wf = BFwf(Nel, polys; ν=3)
we get
[ Info: ψ
4.780 μs (6 allocations: 672 bytes)
[ Info: ∇ψ
12.250 μs (30 allocations: 71.36 KiB)
[ Info: Δψ
38.750 μs (153 allocations: 95.41 KiB)
[ Info: ∇Δψ
57.667 μs (183 allocations: 297.28 KiB)
Test 2: with a larger model
Nel = 8
polys = legendre_basis(10)
wf = BFwf(Nel, polys; ν=4)
we get
[ Info: ψ
78.875 μs (6 allocations: 1.52 KiB)
[ Info: ∇ψ
159.083 μs (30 allocations: 797.11 KiB)
[ Info: Δψ
633.291 μs (183 allocations: 889.19 KiB)
[ Info: ∇Δψ
726.209 μs (245 allocations: 3.19 MiB)
Sure! Maybe we can merge later today if you are free. I will be at IAM in the afternoon.
Since there are no conflicts, I think I can just merge then. (I'm at home again today)
CC @CheukHinHoJerry