btmit

"Unfortuantely" was simply in regard to having to give you a code snippet instead of a proper PR. Also, as I learn more about Julia marcos, I believe the following...

You're right, that allocates. This doesn't seem to: ```julia macro tuplegen(expr) Expr(:tuple, eval(expr)...) end ```

In addition to unexpected typing there is a noteworthy performance penalty. ```julia ua_raw = SMatrix{3,3}(ua) @btime UpperTriangular(inv($ua_raw)) # 4.666 ns (0 allocations: 0 bytes) @btime inv($ua) # 162.800 ns (1...

I'm not familiar with the implementation - just coming at it from the user side. To work around this I currently have to 1) copy sparse matrix to vector of...

Sure, I'll give it a shot. First, I want to say that I now understand why you think this is a bad idea and if the efficiency of `dropzeros!` was...

I understand why `zero(::Type{Tv})` is needed for the variable `xs` (which is sparse), but I'm missing why it is being evaluated for `y`. If `y` were sparse and `xs` were...

Just checking in to see if anyone had made progress fixing this bug. I found my way here from: https://discourse.julialang.org/t/sparse-broadcasting-division-with-a-dense-vector-yields-a-dense-array/15005

I'm running into this issue as well. ``` using CuArrays, FFTW a = cu(rand(ComplexF32, 16, 8, 4, 32)) p = plan_fft!(a, 2) ``` The error seems to apply to `plan_fft`...

I believe you can do this with the various options in `cufftPlanMany()` in CUFFT. The problem I found with `permutedims` is that it can't be done inplace, so this approach...

I will try and figure it out as soon as I have a minute. I think I might be able to track down some old code. As another data point,...