在<T> Alea.GPU 中访问 IList.Default.For

Question

我正在尝试访问在Alea.Gpu.Default.For之外声明的System.Collections.Generic.IList<T>的值。

[GpuManaged]
private void Evaluate_Caching(IList<TGenome> genomeList)
{
    var gpu = Gpu.Default;
    gpu.For(0, genomeList.Count - 1, i =>
    {
        TGenome genome = genomeList[i];
        TPhenome phenome = (TPhenome)genome.CachedPhenome;
        if (null == phenome)
        {   // Decode the phenome and store a ref against the genome.
            phenome = _genomeDecoder.Decode(genome);
            genome.CachedPhenome = phenome;
        }
        if (null == phenome)
        {   // Non-viable genome.
            genome.EvaluationInfo.SetFitness(0.0);
            genome.EvaluationInfo.AuxFitnessArr = null;
        }
        else
        {
            FitnessInfo fitnessInfo = _phenomeEvaluator.Evaluate(phenome);
            genome.EvaluationInfo.SetFitness(fitnessInfo._fitness);
            genome.EvaluationInfo.AuxFitnessArr = fitnessInfo._auxFitnessArr;
        }
    });
}

参考之前提出的一个问题，使用alea-gpu迭代自定义类的集合，我也在App.config中启用了<memory allowNonBlittableMemoryTransfer="true"/>。

但是我收到一个错误，因为

"Cannot get field "genomeList".
Possible reasons:
-> The field type is not supported.
-> In closure class, the field doesn't have [GpuParam] attribute.
Diagnostics:
-> Struct: ref(dyn{ i32 }(m:<Evaluate_Caching>b__0)):_O_
-> FieldName: genomeList
Source location stack:
-> in E:_turingProjects_csProjectsGpu_ProjectGpuParallelGenomeListEvaluator.cs(183,17-183,48)
-> at Evaluators.GpuParallelGenomeListEvaluator`2+<>c__DisplayClass17_0[SharpNeat.Genomes.Neat.NeatGenome,SharpNeat.Phenomes.IBlackBox].[Void <Evaluate_Caching>b__0(Int32)]
-> at Alea.Parallel.Device.DeviceFor.[Void Kernel(Int32, Int32, System.Action`1[System.Int32])]
-> at defining runtime64 (sm50,64bit)
Loading method as kernel:
-> Method: Alea.Parallel.Device.DeviceFor.[Void Kernel(Int32, Int32, System.Action`1[System.Int32])]
-> InstanceOpt: <None>
-> Argument.#0: 0
-> Argument.#1: 1999
-> Argument.#2: System.Action`1[System.Int32]

错误的可能原因是什么？在 GPU 中使用值的正确方法是什么？

Answer 1

目前，AleaGPU 仅适用于数组。列表通常需要动态内存分配，例如添加元素，这在 GPU 中效率不高。