I am now using MKL.
a simple test:
Warning: both Julia and NumPy are linked with MKL, which may cause conflicts and crashes (#433).
└ @ PyCall C:\Users\test\.julia\packages\PyCall\a5Jd3\src\numpy.jl:73
The kernel appears to have died. It will restart automatically.
how to fix the bug?
This is a reported issue in PyCall.jl
Seems like there is no real solution for this problem yet! (https://github.com/JuliaPy/PyPlot.jl/issues/428).
If you are not very particular about PyPlot, then try Plots.jl+GR
May I ask, is PyPlot plotting speed slower vs Plots.jl?
Plots.jl is library which uses multiple backends to plot. PyPlot is also one of the backends for Plots.jl - so you can think of it as PyPlot itself with little bit extra code from Plots.jl.
But GR backend of Plots.jl is supposed to be better that PyPlot.