Scalar indexing in continuous callbacks with EnsembleGPUArray

[ChrisRackauckas]

29d90b2

is marks continuous callback usage broken because of scalar indexing in https://github.com/JuliaDiffEq/DiffEqBase.jl/blob/master/src/init.jl#L150 . @maleadt I thought this was fixed on CuArrays.jl?

[maleadt]

Yes, findall is in CuArrays 1.4.0 JuliaGPU/CuArrays.jl#446
Using old package? Working with an array wrapper?

[ChrisRackauckas]

Interesting....

using DiffEqGPU, CuArrays, OrdinaryDiffEq, Test

function lorenz(du,u,p,t)
 @inbounds begin
     du[1] = p[1]*(u[2]-u[1])
     du[2] = u[1]*(p[2]-u[3]) - u[2]
     du[3] = u[1]*u[2] - p[3]*u[3]
 end
 nothing
end

CuArrays.allowscalar(false)
u0 = Float32[1.0;0.0;0.0]
tspan = (0.0f0,100.0f0)
p = (10.0f0,28.0f0,8/3f0)
prob = ODEProblem(lorenz,u0,tspan,p)
const pre_p = [rand(Float32,3) for i in 1:10]
prob_func = (prob,i,repeat) -> remake(prob,p=pre_p[i].*p)

c_condition = function (u,t,integrator)
    @inbounds u[1] - 3
end

c_affect! = function (integrator)
    @inbounds integrator.u[1] += 20
end

callback_prob = ODEProblem(lorenz,u0,tspan,p,callback=ContinuousCallback(c_condition,c_affect!,save_positions=(false,false)))
callback_monteprob = EnsembleProblem(callback_prob, prob_func = prob_func)
solve(callback_monteprob,Tsit5(),EnsembleGPUArray(),trajectories=2,saveat=1.0f0)

The MWE works on my laptop but not my desktop