Unable to load DLL 'CpuMathNative': The specified module could not be found

[BalsamIbrahim]

salam all
I face this problem when I use Microsoft.ML to create clustering
I hope to help me
System.DllNotFoundException: 'Unable to load DLL 'CpuMathNative': The specified module could not be found. (Exception from HRESULT: 0x8007007E)'
thank you

[frank-dong-ms-zz]

@BalsamIbrahim Thanks for using ML.NET, could you please provide a repro project so we can investigate further?

[BalsamIbrahim]

@frank-dong-ms
project type : Class Library
Target : 4.6.1
what I do : I have list from Latitude and longitude and I need do clustering for them in my plugin

[BalsamIbrahim]

this my code 
        public void cluster (List<Entity> prescriptions , int total_seconds_allowed, int iteration)
        {
            try
            {
                var max = prescriptions.Count / 3;
                max = total_seconds_allowed > 3600 ? max * 2 : max;
                var max_clusters = max + iteration;
                max_clusters = max_clusters > prescriptions.Count ? prescriptions.Count : max_clusters;

                var listpoints = new List<InputDate>();
               //prescriptions.ForEach(item => listpoints.Add( new InputDate { Feature1 =item.GetAttributeValue<decimal>("new_latitude"), Feature2 = item.GetAttributeValue<decimal>("new_longitude") }));
               // listpoints = listpoints.Where(c => c.Feature1 != 0 && c.Feature2 != 0).ToList();
              
                List<InputDate> pointsValues = prescriptions
                    .Select(value => { return new InputDate(value.GetAttributeValue<decimal>("new_longitude"), value.GetAttributeValue<decimal>("new_latitude")); })
                    .ToList();
                var mlContext = new MLContext();
       
                IDataView dataView = mlContext.Data.LoadFromEnumerable(pointsValues);
                var options = new KMeansTrainer.Options
                {
                    NumberOfClusters = max_clusters,
                    OptimizationTolerance = 1e-6f,
                    NumberOfThreads = 1,InitializationAlgorithm = KMeansTrainer.InitializationAlgorithm.Random,FeatureColumnName = "Features"  
                };
                var pipeline = mlContext.Transforms.Concatenate(
                    "Features",
                    "Latitude", 
                    "Longitude")
                    .Append(mlContext.Clustering.Trainers.KMeans(options));

                var model = pipeline.Fit(dataView);
            }
            catch (Exception ex)
            {

                throw;
            }
            

        }
        public class InputDate
        {
            [LoadColumn(0)]
            public float Latitude;
            [LoadColumn(1)]
            public float Longitude;
            [LoadColumn(2)]
            public string Label;
            public InputDate(decimal longitude, decimal latitude)
            {
                Latitude = Convert.ToSingle(latitude) ;
                Longitude = Convert.ToSingle(longitude);
            }
        }

[frank-dong-ms-zz]

@BalsamIbrahim Thanks, could you please provide a complete repro project or solution, and your repro steps? Previous code piece is not enough for me to investigate the issue. Also, the missing DLL 'CpuMathNative' is coming from nuget package 'Microsoft.ML.CpuMath', you can find this dll under path 'runtimes\win-x64\nativeassets\netstandard2.0\CpuMathNative.dll', could you verify if the nuget package contains that dll?

[BalsamIbrahim]

okay because I speak Arabic not English so please can you afford me ??
I'll upload some picture
references

packages

I am now briefly clarifying the project
First I have a group of packages for people
They are receive through the driver
Objective: To give the driver the best route to deliver all packages to people
From points I want to create the best path using the microsoft.ML

[frank-dong-ms-zz]

@BalsamIbrahim are you able to provide the whole/complete project?

[BalsamIbrahim]

@frank-dong-ms YES I HAVE COMPLETE PROJECT
I'LL COMPRESS PROJECT AND UPLOAD IT

[BalsamIbrahim]

class name Scheduale

[frank-dong-ms-zz]

@BalsamIbrahim Thanks I can reproduce your issue and I will start to work on a fix.

[frank-dong-ms-zz]

@BalsamIbrahim If possible you can use dotnet core for your class library project Plugin and your issue will go away if you use dotnet core.

[BalsamIbrahim]

okay, I try use .net core but I face problem in this

this plugins use frame work

[frank-dong-ms-zz]

@BalsamIbrahim is it possible to change the Plugin project to use .net core?

[frank-dong-ms-zz]

@BalsamIbrahim another work around is to add package reference of Microsoft.ML.CpuMath to the project consuming the plugin project and set x64 for both Debug and Release:
#4870 (comment)

[BalsamIbrahim]

@BalsamIbrahim another work around is to add package reference of Microsoft.ML.CpuMath to the project consuming the plugin project and set x64 for both Debug and Release:
#4870 (comment)

it's work thanks

[frank-dong-ms-zz]

Close this issue now, feel free to reopen if any necessary.