Windows 中的未来多进程并行

Question

我正在尝试实施并行过程，以便将栅格数据转换为坡向、坡度等地形产品。我使用以下代码使用 future 这样做：

dem = raster("./dem/dem.asc")
output = "./output/"
crs(dem) <- epsg

plan(multiprocess, workers = availableCores()-1, gc = TRUE)
f1 %<-% terrain(dem, filename = paste0(output,"01_slope.asc"), opt = "slope", unit = 'degrees', neighbors = 8)
f2 %<-% terrain(dem, filename = paste0(output,"02_aspect.asc"), opt = "aspect", unit = 'degrees', neighbors = 8)
f1; f2

这些进程并行开始，但有些奇怪：它同时生成两个文件，每个文件都有自己的名字，但是，两个文件完全相同（在这种情况下，两个栅格都是斜率栅格） . 我做错了什么？

Answer 1

我终于用上了这个功能并且成功了：

dempath = "./dem/dem.asc"
output = "./output/"
crs(dem) <- epsg    

#CREATE DEM PRODUCTS IN PARALLEL
demproductspar <- function(dempath, output, epsg = crs('+init=epsg:25829'), cor = 3){

    print ('Leyendo dem...')
    dem = raster(dempath)
    crs(dem) <- epsg
    plan(multiprocess, workers = cor, gc = TRUE)
    plan(multiprocess, workers = availableCores(), gc = TRUE)

    f1 <- future({ terrain(dem, filename = paste0(output,"01_slope.asc"), opt = "slope", unit = 'degrees', neighbors = 8) })
    f2 <- future({
        terrain(dem, filename = paste0(output,"02_aspect.asc"), opt = "aspect", unit = 'degrees', neighbors = 8)
        aspectreclass(aspectdir = paste0(output,"02_aspect.asc"), output = paste0(output,"02r_aspect.asc")) })
    f3 <- future({ terrain(dem, filename = paste0(output,"03_roughness.asc"), opt = "roughness", unit = 'degrees', neighbors = 8) })
    f4 <- future({ terrain(dem, filename = paste0(output,"04_tri.asc"), opt = "tri", unit = 'degrees', neighbors = 8) })
    f5 <- future({ terrain(dem, filename = paste0(output,"05_tpi.asc"), opt = "tpi", unit = 'degrees', neighbors = 8) })

    print ('Procesando dem...')
    f1; f2; f3; f4; f5
    gc()
}

demproductspar(dempath, output)

Windows 中的未来多进程并行

Future multiprocess parallel in Windows

parallel-processing

future

r

raster

terrain