@@ -53,18 +53,23 @@ cpu_mat = rand(rng, Float32, m, n)
5353gpu_mat =  MtlMatrix {Float32} (cpu_mat)
5454gpu_mat_long =  MtlMatrix {Float32} (rand (rng, Float32, m_long, n_long))
5555gpu_vec =  reshape (gpu_mat, length (gpu_mat))
56+ gpu_vec_long =  reshape (gpu_mat_long, length (gpu_mat_long))
5657gpu_arr_3d =  reshape (gpu_mat, (m, 40 , 25 ))
5758gpu_arr_4d =  reshape (gpu_mat, (m, 10 , 10 , 10 ))
5859gpu_mat_ints =  MtlMatrix {Int64} (rand (rng, - 10 : 10 , m, n))
5960gpu_mat_long_ints =  MtlMatrix {Int64} (rand (rng, - 10 : 10 , m_long, n_long))
6061gpu_vec_ints =  reshape (gpu_mat_ints, length (gpu_mat_ints))
6162
62- #  let group = addgroup!(group, "reverse")
63- #      group["1d"] = @benchmarkable Metal.@sync reverse($gpu_vec)
64- #      group["2d"] = @benchmarkable Metal.@sync reverse($gpu_mat; dims=1)
65- #      group["1d_inplace"] = @benchmarkable Metal.@sync reverse!($gpu_vec)
66- #      group["2d_inplace"] = @benchmarkable Metal.@sync reverse!($gpu_mat; dims=1)
67- #  end
63+ let  group =  addgroup! (group, " reverse" 
64+     group[" 1d" =  @benchmarkable  Metal. @sync  reverse ($ gpu_vec)
65+     group[" 1dL" =  @benchmarkable  Metal. @sync  reverse ($ gpu_vec_long)
66+     group[" 2d" =  @benchmarkable  Metal. @sync  reverse ($ gpu_mat; dims= 1 )
67+     group[" 2dL" =  @benchmarkable  Metal. @sync  reverse ($ gpu_mat_long; dims= 1 )
68+     group[" 1d_inplace" =  @benchmarkable  Metal. @sync  reverse! ($ gpu_vec)
69+     group[" 1dL_inplace" =  @benchmarkable  Metal. @sync  reverse! ($ gpu_vec_long)
70+     group[" 2d_inplace" =  @benchmarkable  Metal. @sync  reverse! ($ gpu_mat; dims= 1 )
71+     group[" 2dL_inplace" =  @benchmarkable  Metal. @sync  reverse! ($ gpu_mat_long; dims= 2 )
72+ end 
6873
6974#  'evals=1' added to prevent hang when running benchmarks of CI
7075#  TODO : Investigate cause and properly fix.
0 commit comments