Skip to content

Instantly share code, notes, and snippets.

@ilkerkesen

ilkerkesen/reduction-timing1.log

Last active April 6, 2017 19:40
Show Gist options
  • Star 0 You must be signed in to star a gist
  • Fork 0 You must be signed in to fork a gist
  • Save ilkerkesen/4ebb8aa91fbbc877a8b0468a8600271c to your computer and use it in GitHub Desktop.
Save ilkerkesen/4ebb8aa91fbbc877a8b0468a8600271c to your computer and use it in GitHub Desktop.
Download ZIP
Knet Reduction Timings
Raw
reduction-timing1.log
6-element Array{Tuple{Int64,Int64},1}:
(100,100)
(500,500)
(100,1000)
(1000,100)
(100,60000)
(60000,100)
julia> a = randn(Float32, ds[1]...);
julia> k = KnetArray(a);
julia> @benchmark sum(k,1)
BenchmarkTools.Trial:
memory estimate: 240 bytes
allocs estimate: 9
--------------
minimum time: 4.098 μs (0.00% GC)
median time: 4.594 μs (0.00% GC)
mean time: 4.629 μs (0.00% GC)
maximum time: 14.002 μs (0.00% GC)
--------------
samples: 10000
evals/sample: 7
time tolerance: 5.00%
memory tolerance: 1.00%
julia> @benchmark sum(k,2)
BenchmarkTools.Trial:
memory estimate: 528 bytes
allocs estimate: 17
--------------
minimum time: 5.114 μs (0.00% GC)
median time: 5.965 μs (0.00% GC)
mean time: 6.115 μs (0.87% GC)
maximum time: 1.129 ms (46.89% GC)
--------------
samples: 10000
evals/sample: 6
time tolerance: 5.00%
memory tolerance: 1.00%
julia> a = randn(Float32, ds[2]...);
julia> k = KnetArray(a);
julia> @benchmark sum(k,1)
BenchmarkTools.Trial:
memory estimate: 272 bytes
allocs estimate: 11
--------------
minimum time: 4.197 μs (0.00% GC)
median time: 11.985 μs (0.00% GC)
mean time: 11.803 μs (0.48% GC)
maximum time: 1.522 ms (37.52% GC)
--------------
samples: 10000
evals/sample: 7
time tolerance: 5.00%
memory tolerance: 1.00%
julia> @benchmark sum(k,2)
BenchmarkTools.Trial:
memory estimate: 560 bytes
allocs estimate: 19
--------------
minimum time: 5.114 μs (0.00% GC)
median time: 53.434 μs (0.00% GC)
mean time: 52.513 μs (0.18% GC)
maximum time: 1.561 ms (59.62% GC)
--------------
samples: 10000
evals/sample: 6
time tolerance: 5.00%
memory tolerance: 1.00%
julia> a = randn(Float32, ds[3]...);
julia> k = KnetArray(a);
julia> @benchmark sum(k,1)
BenchmarkTools.Trial:
memory estimate: 336 bytes
allocs estimate: 15
--------------
minimum time: 4.206 μs (0.00% GC)
median time: 12.313 μs (0.00% GC)
mean time: 12.120 μs (0.42% GC)
maximum time: 1.293 ms (39.80% GC)
--------------
samples: 10000
evals/sample: 7
time tolerance: 5.00%
memory tolerance: 1.00%
julia> @benchmark sum(k,2)
BenchmarkTools.Trial:
memory estimate: 528 bytes
allocs estimate: 17
--------------
minimum time: 5.052 μs (0.00% GC)
median time: 25.351 μs (0.00% GC)
mean time: 24.905 μs (0.21% GC)
maximum time: 1.113 ms (46.85% GC)
--------------
samples: 10000
evals/sample: 6
time tolerance: 5.00%
memory tolerance: 1.00%
julia> a = randn(Float32, ds[4]...);
julia> k = KnetArray(a);
julia> @benchmark sum(k,1)
BenchmarkTools.Trial:
memory estimate: 256 bytes
allocs estimate: 10
--------------
minimum time: 4.178 μs (0.00% GC)
median time: 6.223 μs (0.00% GC)
mean time: 6.407 μs (1.00% GC)
maximum time: 1.527 ms (41.92% GC)
--------------
samples: 10000
evals/sample: 7
time tolerance: 5.00%
memory tolerance: 1.00%
julia> @benchmark sum(k,2)
a = randn(Float32, ds[5]...);
BenchmarkTools.Trial:
memory estimate: 624 bytes
allocs estimate: 23
--------------
minimum time: 5.145 μs (0.00% GC)
median time: 27.112 μs (0.00% GC)
mean time: 26.636 μs (0.36% GC)
maximum time: 1.025 ms (48.13% GC)
--------------
samples: 10000
evals/sample: 6
time tolerance: 5.00%
memory tolerance: 1.00%
julia>
julia> k = KnetArray(a);
julia> @benchmark sum(k,1)
BenchmarkTools.Trial:
memory estimate: 336 bytes
allocs estimate: 15
--------------
minimum time: 4.737 μs (0.00% GC)
median time: 554.659 μs (0.00% GC)
mean time: 527.203 μs (0.00% GC)
maximum time: 1.264 ms (0.00% GC)
--------------
samples: 9403
evals/sample: 1
time tolerance: 5.00%
memory tolerance: 1.00%
julia> @benchmark sum(k,2)
BenchmarkTools.Trial:
memory estimate: 528 bytes
allocs estimate: 17
--------------
minimum time: 5.738 μs (0.00% GC)
median time: 1.446 ms (0.00% GC)
mean time: 1.367 ms (0.00% GC)
maximum time: 1.601 ms (0.00% GC)
--------------
samples: 610
evals/sample: 6
time tolerance: 5.00%
memory tolerance: 1.00%
julia> a = randn(Float32, ds[6]...);
julia> k = KnetArray(a);
julia> @benchmark sum(k,1)
BenchmarkTools.Trial:
memory estimate: 256 bytes
allocs estimate: 10
--------------
minimum time: 4.316 μs (0.00% GC)
median time: 172.717 μs (0.00% GC)
mean time: 166.780 μs (0.00% GC)
maximum time: 173.993 μs (0.00% GC)
--------------
samples: 4264
evals/sample: 7
time tolerance: 5.00%
memory tolerance: 1.00%
julia> @benchmark sum(k,2)
BenchmarkTools.Trial:
memory estimate: 624 bytes
allocs estimate: 23
--------------
minimum time: 5.593 μs (0.00% GC)
median time: 1.489 ms (0.00% GC)
mean time: 1.424 ms (0.00% GC)
maximum time: 1.489 ms (0.00% GC)
--------------
samples: 585
evals/sample: 6
time tolerance: 5.00%
memory tolerance: 1.00%
Raw
reduction-timing2.log
6-element Array{Tuple{Int64,Int64},1}:
(100,100)
(500,500)
(100,1000)
(1000,100)
(100,60000)
(60000,100)
julia>
julia> a = randn(Float32, ds[1]...);
julia> k = KnetArray(a);
julia> @benchmark sum(k,1)
BenchmarkTools.Trial:
memory estimate: 96 bytes
allocs estimate: 3
--------------
minimum time: 3.926 μs (0.00% GC)
median time: 5.041 μs (0.00% GC)
mean time: 5.062 μs (0.00% GC)
maximum time: 12.898 μs (0.00% GC)
--------------
samples: 10000
evals/sample: 7
time tolerance: 5.00%
memory tolerance: 1.00%
julia> @benchmark sum(k,2)
BenchmarkTools.Trial:
memory estimate: 384 bytes
allocs estimate: 11
--------------
minimum time: 4.675 μs (0.00% GC)
median time: 6.425 μs (0.00% GC)
mean time: 6.577 μs (0.76% GC)
maximum time: 1.161 ms (42.95% GC)
--------------
samples: 10000
evals/sample: 7
time tolerance: 5.00%
memory tolerance: 1.00%
julia> a = randn(Float32, ds[2]...);
julia> k = KnetArray(a);
julia> @benchmark sum(k,1)
BenchmarkTools.Trial:
memory estimate: 128 bytes
allocs estimate: 5
--------------
minimum time: 3.978 μs (0.00% GC)
median time: 14.334 μs (0.00% GC)
mean time: 14.111 μs (0.00% GC)
maximum time: 16.470 μs (0.00% GC)
--------------
samples: 10000
evals/sample: 7
time tolerance: 5.00%
memory tolerance: 1.00%
julia> @benchmark sum(k,2)
BenchmarkTools.Trial:
memory estimate: 416 bytes
allocs estimate: 13
--------------
minimum time: 4.764 μs (0.00% GC)
median time: 56.472 μs (0.00% GC)
mean time: 55.516 μs (0.10% GC)
maximum time: 1.299 ms (43.10% GC)
--------------
samples: 10000
evals/sample: 6
time tolerance: 5.00%
memory tolerance: 1.00%
julia> a = randn(Float32, ds[3]...);
julia> k = KnetArray(a);
julia> @benchmark sum(k,1)
BenchmarkTools.Trial:
memory estimate: 128 bytes
allocs estimate: 5
--------------
minimum time: 4.006 μs (0.00% GC)
median time: 16.098 μs (0.00% GC)
mean time: 15.856 μs (0.00% GC)
maximum time: 17.634 μs (0.00% GC)
--------------
samples: 10000
evals/sample: 7
time tolerance: 5.00%
memory tolerance: 1.00%
julia> @benchmark sum(k,2)
BenchmarkTools.Trial:
memory estimate: 384 bytes
allocs estimate: 11
--------------
minimum time: 4.735 μs (0.00% GC)
median time: 26.163 μs (0.00% GC)
mean time: 25.771 μs (0.20% GC)
maximum time: 1.159 ms (43.67% GC)
--------------
samples: 10000
evals/sample: 7
time tolerance: 5.00%
memory tolerance: 1.00%
julia> a = randn(Float32, ds[4]...);
julia> k = KnetArray(a);
julia> @benchmark sum(k,1)
BenchmarkTools.Trial:
memory estimate: 96 bytes
allocs estimate: 3
--------------
minimum time: 4.021 μs (0.00% GC)
median time: 7.369 μs (0.00% GC)
mean time: 7.245 μs (0.00% GC)
maximum time: 12.383 μs (0.00% GC)
--------------
samples: 10000
evals/sample: 7
time tolerance: 5.00%
memory tolerance: 1.00%
julia> @benchmark sum(k,2)
BenchmarkTools.Trial:
memory estimate: 416 bytes
allocs estimate: 13
--------------
minimum time: 4.729 μs (0.00% GC)
median time: 32.814 μs (0.00% GC)
mean time: 32.321 μs (0.15% GC)
maximum time: 1.114 ms (44.75% GC)
--------------
samples: 10000
evals/sample: 7
time tolerance: 5.00%
memory tolerance: 1.00%
julia> a = randn(Float32, ds[5]...);
julia> k = KnetArray(a);
julia> @benchmark sum(k,1)
BenchmarkTools.Trial:
memory estimate: 128 bytes
allocs estimate: 5
--------------
minimum time: 4.038 μs (0.00% GC)
median time: 763.446 μs (0.00% GC)
mean time: 729.050 μs (0.00% GC)
maximum time: 1.396 ms (0.00% GC)
--------------
samples: 6813
evals/sample: 1
time tolerance: 5.00%
memory tolerance: 1.00%
julia> @benchmark sum(k,2)
a = randn(Float32, ds[6]...);
BenchmarkTools.Trial:
memory estimate: 384 bytes
allocs estimate: 11
--------------
minimum time: 5.064 μs (0.00% GC)
median time: 1.469 ms (0.00% GC)
mean time: 1.388 ms (0.00% GC)
maximum time: 1.579 ms (0.00% GC)
--------------
samples: 515
evals/sample: 7
time tolerance: 5.00%
memory tolerance: 1.00%
julia>
julia> k = KnetArray(a);
julia> @benchmark sum(k,1)
BenchmarkTools.Trial:
memory estimate: 96 bytes
allocs estimate: 3
--------------
minimum time: 4.082 μs (0.00% GC)
median time: 257.741 μs (0.00% GC)
mean time: 246.487 μs (0.00% GC)
maximum time: 259.281 μs (0.00% GC)
--------------
samples: 2530
evals/sample: 8
time tolerance: 5.00%
memory tolerance: 1.00%
julia> @benchmark sum(k,2)
BenchmarkTools.Trial:
memory estimate: 416 bytes
allocs estimate: 13
--------------
minimum time: 5.112 μs (0.00% GC)
median time: 1.730 ms (0.00% GC)
mean time: 1.655 ms (0.00% GC)
maximum time: 1.750 ms (0.00% GC)
--------------
samples: 432
evals/sample: 7
time tolerance: 5.00%
memory tolerance: 1.00%
Raw
reduction-timing3.log
6-element Array{Tuple{Int64,Int64},1}:
(100,100)
(500,500)
(100,1000)
(1000,100)
(100,60000)
(60000,100)
julia> a = randn(Float32, ds[1]...);
julia> k = KnetArray(a);
julia> @benchmark sum(k,1)
BenchmarkTools.Trial:
memory estimate: 240 bytes
allocs estimate: 9
--------------
minimum time: 4.087 μs (0.00% GC)
median time: 4.795 μs (0.00% GC)
mean time: 4.850 μs (0.00% GC)
maximum time: 14.486 μs (0.00% GC)
--------------
samples: 10000
evals/sample: 7
time tolerance: 5.00%
memory tolerance: 1.00%
julia> @benchmark sum(k,2)
BenchmarkTools.Trial:
memory estimate: 528 bytes
allocs estimate: 17
--------------
minimum time: 5.029 μs (0.00% GC)
median time: 6.055 μs (0.00% GC)
mean time: 6.244 μs (1.21% GC)
maximum time: 1.305 ms (58.06% GC)
--------------
samples: 10000
evals/sample: 6
time tolerance: 5.00%
memory tolerance: 1.00%
julia> a = randn(Float32, ds[2]...);
julia> k = KnetArray(a);
julia> @benchmark sum(k,1)
BenchmarkTools.Trial:
memory estimate: 272 bytes
allocs estimate: 11
--------------
minimum time: 4.113 μs (0.00% GC)
median time: 14.870 μs (0.00% GC)
mean time: 14.636 μs (0.59% GC)
maximum time: 1.749 ms (49.04% GC)
--------------
samples: 10000
evals/sample: 7
time tolerance: 5.00%
memory tolerance: 1.00%
julia> @benchmark sum(k,2)
BenchmarkTools.Trial:
memory estimate: 560 bytes
allocs estimate: 19
--------------
minimum time: 5.160 μs (0.00% GC)
median time: 56.003 μs (0.00% GC)
mean time: 55.045 μs (0.17% GC)
maximum time: 1.521 ms (62.01% GC)
--------------
samples: 10000
evals/sample: 6
time tolerance: 5.00%
memory tolerance: 1.00%
julia> a = randn(Float32, ds[3]...);
julia> k = KnetArray(a);
julia> @benchmark sum(k,1)
BenchmarkTools.Trial:
memory estimate: 336 bytes
allocs estimate: 15
--------------
minimum time: 4.250 μs (0.00% GC)
median time: 16.657 μs (0.00% GC)
mean time: 16.407 μs (0.32% GC)
maximum time: 1.245 ms (42.08% GC)
--------------
samples: 10000
evals/sample: 7
time tolerance: 5.00%
memory tolerance: 1.00%
julia> @benchmark sum(k,2)
BenchmarkTools.Trial:
memory estimate: 528 bytes
allocs estimate: 17
--------------
minimum time: 5.104 μs (0.00% GC)
median time: 25.693 μs (0.00% GC)
mean time: 25.243 μs (0.21% GC)
maximum time: 1.095 ms (49.05% GC)
--------------
samples: 10000
evals/sample: 6
time tolerance: 5.00%
memory tolerance: 1.00%
julia> a = randn(Float32, ds[4]...);
julia> k = KnetArray(a);
julia> @benchmark sum(k,1)
BenchmarkTools.Trial:
memory estimate: 256 bytes
allocs estimate: 10
--------------
minimum time: 4.091 μs (0.00% GC)
median time: 6.459 μs (0.00% GC)
mean time: 6.588 μs (0.93% GC)
maximum time: 1.445 ms (42.24% GC)
--------------
samples: 10000
evals/sample: 7
time tolerance: 5.00%
memory tolerance: 1.00%
julia> @benchmark sum(k,2)
BenchmarkTools.Trial:
memory estimate: 624 bytes
allocs estimate: 23
--------------
minimum time: 5.278 μs (0.00% GC)
median time: 31.534 μs (0.00% GC)
mean time: 30.976 μs (0.32% GC)
maximum time: 1.020 ms (51.32% GC)
--------------
samples: 10000
evals/sample: 6
time tolerance: 5.00%
memory tolerance: 1.00%
julia> a = randn(Float32, ds[5]...);
julia> k = KnetArray(a);
julia> @benchmark sum(k,1)
BenchmarkTools.Trial:
memory estimate: 336 bytes
allocs estimate: 15
--------------
minimum time: 4.506 μs (0.00% GC)
median time: 810.902 μs (0.00% GC)
mean time: 762.506 μs (0.00% GC)
maximum time: 1.746 ms (0.00% GC)
--------------
samples: 6521
evals/sample: 1
time tolerance: 5.00%
memory tolerance: 1.00%
julia> @benchmark sum(k,2)
BenchmarkTools.Trial:
memory estimate: 528 bytes
allocs estimate: 17
--------------
minimum time: 5.408 μs (0.00% GC)
median time: 1.589 ms (0.00% GC)
mean time: 1.519 ms (0.00% GC)
maximum time: 1.734 ms (0.00% GC)
--------------
samples: 549
evals/sample: 6
time tolerance: 5.00%
memory tolerance: 1.00%
julia> a = randn(Float32, ds[6]...);
julia> k = KnetArray(a);
julia> @benchmark sum(k,1)
BenchmarkTools.Trial:
memory estimate: 256 bytes
allocs estimate: 10
--------------
minimum time: 4.286 μs (0.00% GC)
median time: 173.004 μs (0.00% GC)
mean time: 167.080 μs (0.00% GC)
maximum time: 174.756 μs (0.00% GC)
--------------
samples: 4258
evals/sample: 7
time tolerance: 5.00%
memory tolerance: 1.00%
julia> @benchmark sum(k,2)
BenchmarkTools.Trial:
memory estimate: 624 bytes
allocs estimate: 23
--------------
minimum time: 5.744 μs (0.00% GC)
median time: 1.723 ms (0.00% GC)
mean time: 1.648 ms (0.00% GC)
maximum time: 1.758 ms (0.00% GC)
--------------
samples: 506
evals/sample: 6
time tolerance: 5.00%
memory tolerance: 1.00%
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment