Cleanup some imports (#398)

Author: christiangnrd

lib/mps/command_buf.jl

@@ -34,7 +34,7 @@ end
 MTL.enqueue!(cmdbuf::MPSCommandBuffer) = @inline MTL.enqueue!(cmdbuf.commandBuffer)
 MTL.commit!(cmdbuf::MPSCommandBuffer) = @inline MTL.commit!(cmdbuf.commandBuffer)
 
-function MPS.commit!(f::Base.Callable, cmdbuf::MPSCommandBuffer)
+function MTL.commit!(f::Base.Callable, cmdbuf::MPSCommandBuffer)
     enqueue!(cmdbuf)
     ret = f(cmdbuf)
     commit!(cmdbuf)

lib/mps/matrix.jl

@@ -42,7 +42,7 @@ for type in [UInt8,UInt16,UInt32,UInt64,Int8,Int16,Int32,Int64,Float16,Float32,C
     @eval Base.convert(::Type{MPSDataType}, ::Type{$type}) = $(Symbol(:MPSDataType, type))
     @eval jl_mps_to_typ[$(Symbol(:MPSDataType, type))] = $type
 end
-Base.sizeof(t::MPS.MPSDataType) = sizeof(jl_mps_to_typ[t])
+Base.sizeof(t::MPSDataType) = sizeof(jl_mps_to_typ[t])
 
 Base.convert(::Type{DataType}, mpstyp::MPSDataType) = jl_mps_to_typ[mpstyp]
 

src/compiler/library.jl

@@ -13,9 +13,9 @@
 
 using SHA: sha256
 using CEnum: @cenum
-using UUIDs: UUID, uuid1
+using UUIDs: UUID
 using Printf: @printf
-using CodecBzip2: transcode, Bzip2Compressor, Bzip2Decompressor
+using CodecBzip2: Bzip2Compressor, Bzip2Decompressor
 
 
 ## enums

src/compiler/reflection.jl

@@ -111,7 +111,7 @@ function extract_gpu_code(f, binary)
     metallib_section = findfirst(Sections(fat_handle[arch]), "__TEXT,__metallib")
     metallib_section === nothing && error("Could not find __metallib section in GPU binary")
     metallib_binary = read(metallib_section)
-    metallib = read(IOBuffer(metallib_binary), Metal.MetalLib)
+    metallib = read(IOBuffer(metallib_binary), MetalLib)
     # TODO: use this to implement a do-block device_code_air like CUDA.jl?
 
     # identify the kernel name

src/gpuarrays.jl

@@ -33,13 +33,6 @@ GPUArrays.griddim(ctx::mtlKernelContext)      = threadgroups_per_grid_1d()
 GPUArrays.global_index(ctx::mtlKernelContext) = thread_position_in_grid_1d()
 GPUArrays.global_size(ctx::mtlKernelContext)  = threads_per_grid_1d()
 
-# math
-
-@inline GPUArrays.cos(ctx::mtlKernelContext, x)  = cos(x)
-@inline GPUArrays.sin(ctx::mtlKernelContext, x)  = sin(x)
-@inline GPUArrays.sqrt(ctx::mtlKernelContext, x) = sqrt(x)
-@inline GPUArrays.log(ctx::mtlKernelContext, x)  = log(x)
-
 # memory
 
 @inline function GPUArrays.LocalMemory(::mtlKernelContext, ::Type{T}, ::Val{dims}, ::Val{id}

src/mapreduce.jl

@@ -34,7 +34,7 @@ end
     end
 
     # wait for all partial reductions
-    threadgroup_barrier(Metal.MemoryFlagThreadGroup)
+    threadgroup_barrier(MemoryFlagThreadGroup)
 
     # read from shared memory only if that warp existed
     val = if thread_index_in_threadgroup() <= fld1(threads_per_threadgroup_1d(), 32)