fixed a sync problem

[ArrogantGao]

The main change is by force sync in the C code, the problem mentioned in #10 is fixed, results are shown as

julia> using GenericTensorNetworks, GenericTensorNetworks.Graphs

julia> using CUDA

julia> g = Graphs.random_regular_graph(200, 3)
{200, 300} undirected simple Int64 graph

julia> optimizer = TreeSA(ntrials=3)
TreeSA{Int64, StepRangeLen{Float64, Base.TwicePrecision{Float64}, Base.TwicePrecision{Float64}, Int64}, GreedyMethod{OMEinsumContractionOrders.MinSpaceOut}, Any}(20, 0.01:0.05:14.96, 3, 50, 1.0, 0.2, :greedy, 0, Any[], GreedyMethod{OMEinsumContractionOrders.MinSpaceOut}(OMEinsumContractionOrders.MinSpaceOut(), 1))

julia> gp = IndependentSet(g; optimizer=optimizer)
┌ Warning: target space complexity not found, got: 23.0, with time complexity 30.29749478594516, read-write complexity 25.905654474039626.
└ @ OMEinsumContractionOrders ~/.julia/packages/OMEinsumContractionOrders/WpwIz/src/treesa.jl:229
IndependentSet{OMEinsum.SlicedEinsum{Int64, OMEinsum.DynamicNestedEinsum{Int64}}, NoWeight}(OMEinsum.SlicedEinsum{Int64, OMEinsum.DynamicNestedEinsum{Int64}}(Int64[], 5, 5 -> 
├─ 147, 147∘5 -> 5
│  ├─ 147
│  └─ 147∘5, 5∘147 -> 147∘5
│     ├─ 182∘143∘147, 182∘143∘5 -> 147∘5
│     │  ├─ 147∘182, 143∘147 -> 182∘143∘147
│     │  │  ├─ 147∘182
│     │  │  └─ 143∘147
│     │  └─ 182∘136∘192∘143, 192∘136∘5 -> 182∘143∘5
│     │     ├─ 151∘178∘198∘33∘78∘146∘143∘196∘182, 33∘136∘78∘192∘146∘143∘198∘196∘178∘151 -> 182∘136∘192∘143
│     │     │  ⋮
│     │     │  
│     │     └─ 5∘192, 5∘136 -> 192∘136∘5
│     │        ⋮
│     │        
│     └─ 5∘147
└─ 5
), SimpleGraph{Int64}(300, [[40, 52, 126], [43, 113, 170], [10, 17, 97], [17, 96, 117], [136, 147, 192], [70, 75, 172], [26, 50, 144], [56, 139, 179], [177, 179, 192], [3, 56, 93]  …  [110, 122, 179], [5, 9, 142], [36, 45, 69], [75, 80, 127], [98, 102, 137], [54, 57, 119], [90, 125, 161], [15, 79, 163], [65, 77, 174], [18, 35, 47]]), NoWeight(), Dict{Int64, Int64}())

julia> contraction_complexity(gp)
Time complexity: 2^30.29749478594516
Space complexity: 2^23.0
Read-write complexity: 2^25.905654474039626

julia> @time CUDA.@sync solve(gp, SizeMax(); usecuda=true, T=Float32)
  4.446523 seconds (7.53 M allocations: 502.525 MiB, 3.03% gc time)
0-dimensional CuArray{Tropical{Float32}, 0, CUDA.Mem.DeviceBuffer}:
88.0ₜ

julia> using CuTropicalGEMM

julia> @time CUDA.@sync solve(gp, SizeMax(); usecuda=true, T=Float32)
  0.048345 seconds (121.74 k allocations: 6.030 MiB)
0-dimensional CuArray{Tropical{Float32}, 0, CUDA.Mem.DeviceBuffer}:
88.0ₜ

julia> @time CUDA.@sync solve(gp, SizeMax(); usecuda=true, T=Float32)
  0.053084 seconds (121.74 k allocations: 6.030 MiB)
0-dimensional CuArray{Tropical{Float32}, 0, CUDA.Mem.DeviceBuffer}:
88.0ₜ

julia> @time CUDA.@sync solve(gp, SizeMax(); usecuda=true, T=Float32)
  0.052982 seconds (121.74 k allocations: 6.030 MiB)
0-dimensional CuArray{Tropical{Float32}, 0, CUDA.Mem.DeviceBuffer}:
88.0ₜ

The result now is stable and the macro @time works properly.

I also removed the unused files .travis.yml and Artifacts.toml, as mentioned in #12.

[GiggleLiu]

Please try this test case, I am afraid it is fully fixed:

using GenericTensorNetworks, GenericTensorNetworks.Graphs
using CUDA
using Random; Random.seed!(6)
g = Graphs.random_regular_graph(200, 3)
item(x::AbstractArray) = Array(x)[]
optimizer = TreeSA(ntrials=1)
gp = IndependentSet(g; optimizer=optimizer)
contraction_complexity(gp)
@time CUDA.@sync solve(gp, SizeMax(); usecuda=true, T=Float32)
using CuTropicalGEMM
@time CUDA.@sync solve(gp, SizeMax(); usecuda=true, T=Float32)

[ArrogantGao]

Please try this test case, I am afraid it is fully fixed:

using GenericTensorNetworks, GenericTensorNetworks.Graphs
using CUDA
using Random; Random.seed!(6)
g = Graphs.random_regular_graph(200, 3)
item(x::AbstractArray) = Array(x)[]
optimizer = TreeSA(ntrials=1)
gp = IndependentSet(g; optimizer=optimizer)
contraction_complexity(gp)
@time CUDA.@sync solve(gp, SizeMax(); usecuda=true, T=Float32)
using CuTropicalGEMM
@time CUDA.@sync solve(gp, SizeMax(); usecuda=true, T=Float32)

julia> using GenericTensorNetworks, GenericTensorNetworks.Graphs

julia> using CUDA

julia> using Random; Random.seed!(6)
TaskLocalRNG()

julia> g = Graphs.random_regular_graph(200, 3)
{200, 300} undirected simple Int64 graph

julia> item(x::AbstractArray) = Array(x)[]
item (generic function with 1 method)

julia> optimizer = TreeSA(ntrials=1)
TreeSA{Int64, StepRangeLen{Float64, Base.TwicePrecision{Float64}, Base.TwicePrecision{Float64}, Int64}, GreedyMethod{OMEinsumContractionOrders.MinSpaceOut}, Any}(20, 0.01:0.05:14.96, 1, 50, 1.0, 0.2, :greedy, 0, Any[], GreedyMethod{OMEinsumContractionOrders.MinSpaceOut}(OMEinsumContractionOrders.MinSpaceOut(), 1))

julia> gp = IndependentSet(g; optimizer=optimizer)
┌ Warning: target space complexity not found, got: 24.0, with time complexity 31.121747025566243, read-write complexity 26.325278911340753.
└ @ OMEinsumContractionOrders ~/.julia/packages/OMEinsumContractionOrders/WpwIz/src/treesa.jl:229
IndependentSet{OMEinsum.SlicedEinsum{Int64, OMEinsum.DynamicNestedEinsum{Int64}}, NoWeight}(OMEinsum.SlicedEinsum{Int64, OMEinsum.DynamicNestedEinsum{Int64}}(Int64[], 86, 86 -> 
├─ 86∘192, 192∘86 -> 86
│  ├─ 86∘192
│  └─ 158∘49∘192, 86∘158∘49 -> 192∘86
│     ├─ 192∘158, 192∘49 -> 158∘49∘192
│     │  ├─ 158∘192, 158 -> 192∘158
│     │  │  ├─ 158∘192, 192 -> 158∘192
│     │  │  │  ⋮
│     │  │  │  
│     │  │  └─ 158
│     │  └─ 49∘192, 49 -> 192∘49
│     │     ├─ 49∘192
│     │     └─ 49
│     └─ 157∘86∘158∘106, 157∘106∘49 -> 86∘158∘49
│        ├─ 158∘106∘151∘157∘199∘125∘86∘21∘74∘138∘68∘30∘60∘58∘57, 138∘60∘125∘30∘58∘68∘74∘151∘158∘57∘199∘21∘106 -> 157∘86∘158∘106
│        │  ├─ 102∘57∘44∘158∘47, 106∘47∘151∘157∘102∘199∘125∘86∘57∘21∘74∘138∘68∘30∘60∘58∘44 -> 158∘106∘151∘157∘199∘125∘86∘21∘74∘138∘68∘30∘60∘58∘57
│        │  │  ⋮
│        │  │  
│        │  └─ 138∘60∘125∘30∘58∘68∘74∘151∘158∘57∘35∘199, 21∘106∘35 -> 138∘60∘125∘30∘58∘68∘74∘151∘158∘57∘199∘21∘106
│        │     ⋮
│        │     
│        └─ 49∘157, 49∘106 -> 157∘106∘49
│           ├─ 49∘157
│           └─ 49∘106
└─ 86
), SimpleGraph{Int64}(300, [[44, 158, 182], [66, 126, 167], [67, 85, 113], [10, 76, 105], [145, 148, 171], [56, 104, 180], [22, 41, 72], [149, 166, 173], [72, 90, 174], [4, 69, 129]  …  [55, 63, 137], [49, 86, 158], [15, 127, 176], [36, 102, 176], [120, 134, 178], [32, 94, 118], [30, 96, 113], [28, 81, 165], [45, 98, 189], [71, 75, 108]]), NoWeight(), Dict{Int64, Int64}())

julia> contraction_complexity(gp)
Time complexity: 2^31.121747025566243
Space complexity: 2^24.0
Read-write complexity: 2^26.325278911340753

julia> @time CUDA.@sync solve(gp, SizeMax(); usecuda=true, T=Float32)
 36.767120 seconds (65.42 M allocations: 4.263 GiB, 4.67% gc time, 0.13% compilation time)
0-dimensional CuArray{Tropical{Float32}, 0, CUDA.Mem.DeviceBuffer}:
89.0ₜ

julia> using CuTropicalGEMM
[ Info: Precompiling CuTropicalGEMM [c2b282c3-c9c2-431d-80f7-a1a0561ebe55]

julia> @time CUDA.@sync solve(gp, SizeMax(); usecuda=true, T=Float32)
  0.470510 seconds (526.07 k allocations: 33.198 MiB, 6.24% gc time)
0-dimensional CuArray{Tropical{Float32}, 0, CUDA.Mem.DeviceBuffer}:
89.0ₜ

julia> @time CUDA.@sync solve(gp, SizeMax(); usecuda=true, T=Float32)
  0.042993 seconds (115.21 k allocations: 5.698 MiB)
0-dimensional CuArray{Tropical{Float32}, 0, CUDA.Mem.DeviceBuffer}:
89.0ₜ

julia> @time CUDA.@sync solve(gp, SizeMax(); usecuda=true, T=Float32)
  0.065605 seconds (115.21 k allocations: 5.698 MiB)
0-dimensional CuArray{Tropical{Float32}, 0, CUDA.Mem.DeviceBuffer}:
89.0ₜ

The test passed, and I also double checked the result on another server.

Did you rebuild the binary after pull?

[GiggleLiu]

I just verified the correctness. Good job! After fixing the Project.toml and CI, we can release v0.1.

[GiggleLiu]

Please remove all dependencies that not directly used in src from the Project.toml
If you want a test environment, please add a Project.toml file to the test folder, example: https://github.com/TensorBFS/TensorInference.jl/tree/main/test
TestEnv.jl can help you start a test environment for debugging easily.

For packages like BenchmarkTools, they should not be included in the local environment.

[GiggleLiu]

We need to get this PR merged before closing issue #10

Approve means you get the permission to merge this PR directly.

[ArrogantGao]

We need to get this PR merged before closing issue #10

Ops, sorry about that, I reopened the the issue #10.