Clinton Tepper
Clinton Tepper
Sure, hope this helps- - [ ] zeros - [ ] ones - [ ] fill - [ ] cumprod - [ ] device2host - [ ] reduce (two examples)...
Updated checklist: - [ ] zeros - [ ] ones - [ ] fill - [ ] cumprod - [ ] device2host - [ ] reduce + - [X] reduce...
I couldn't find any 1.6 versions where this worked, so it must be a 1.5.4->1.6 issue. However, as a workaround, I just modified the adjoint to explicitly accept a CuMatrix...
Thanks- no dice with #932. CUDA#master compiles but doesn't work on my machine, even with GPUArrays#master, GPUCompiler#master, LLVM#master - most array operations just spit out garbage.
Didn't work on CUDA#master. The workaround is still ok though.
Nevermind, I see Gadfly doesn't work with this :(
For posterity: As a work around, one can use the Images, FileIO, and ImageMagick (Windows) packages to open images saved by Gadfly as follows: ```julia draw(PNG("ExcessReturns.png",9inch, 7inch),p) im = load("ExcessReturns.png")...
Thank you! That's great news.
Using the new PR- There is a new and rather bizarre issue. Starting from a new instance of Julia, the first time I run the code it opens the image...
Cool, thanks @mpastell and @tlnagy! @tlnagy , the `using Gadfly` is in the same code block as plot. Also, the plot is wrapped in a function, if that affects anything.