Crate oxigdal_gpu_advanced 

Advanced GPU computing for geospatial operations with multi-GPU support.

This crate provides advanced GPU computing capabilities for OxiGDAL including:

• Multi-GPU orchestration and load balancing
• Advanced memory pool management
• Shader compilation and optimization
• GPU-accelerated terrain analysis
• GPU-based ML inference

Features

• Multi-GPU Support: Automatically detect and utilize multiple GPUs
• Memory Pooling: Efficient GPU memory management with sub-allocation
• Shader Optimization: Compile and optimize WGSL shaders
• Work Stealing: Dynamic load balancing across GPUs
• GPU Affinity: Thread-to-GPU pinning for optimal performance

Examples

use oxigdal_gpu_advanced::multi_gpu::{MultiGpuManager, SelectionStrategy};

// Create multi-GPU manager
let manager = MultiGpuManager::new(SelectionStrategy::LeastLoaded).await?;

// Print available GPUs
manager.print_gpu_info();

// Select best GPU for task
let gpu = manager.select_gpu()?;
println!("Selected GPU: {}", gpu.info.name);

Re-exports

pub use adaptive::AdaptiveConfig;

pub use adaptive::AdaptiveSelector;

pub use adaptive::Algorithm;

pub use adaptive::AlgorithmStats;

pub use adaptive::DeviceInfo;

pub use adaptive::ExecutionStrategy;

pub use adaptive::TuningParams;

pub use adaptive::WorkloadInfo;

pub use error::GpuAdvancedError;

pub use error::Result;

pub use gpu_ml::ActivationType;

pub use gpu_ml::GpuMlInference;

pub use gpu_ml::InferenceStats;

pub use gpu_ml::PoolType;

pub use gpu_terrain::GpuTerrainAnalyzer;

pub use gpu_terrain::TerrainMetrics;

pub use kernels::EdgeDetectionKernel;

pub use kernels::FftKernel;

pub use kernels::HistogramEqKernel;

pub use kernels::KernelParams;

pub use kernels::KernelRegistry;

pub use kernels::MatrixMultiplyKernel;

pub use kernels::MorphologyKernel;

pub use kernels::TextureAnalysisKernel;

pub use memory_compaction::CompactionConfig;

pub use memory_compaction::CompactionResult;

pub use memory_compaction::CompactionStats;

pub use memory_compaction::CompactionStrategy;

pub use memory_compaction::FragmentationInfo;

pub use memory_compaction::MemoryCompactor;

pub use memory_pool::MemoryAllocation;

pub use memory_pool::MemoryPool;

pub use memory_pool::MemoryPoolStats;

pub use multi_gpu::GpuDevice;

pub use multi_gpu::GpuDeviceInfo;

pub use multi_gpu::MultiGpuManager;

pub use multi_gpu::SelectionStrategy;

pub use multi_gpu::affinity::AffinityGuard;

pub use multi_gpu::affinity::AffinityManager;

pub use multi_gpu::affinity::AffinityStats;

pub use multi_gpu::device_manager::DeviceCapabilities;

pub use multi_gpu::device_manager::DeviceFilter;

pub use multi_gpu::device_manager::DeviceManager;

pub use multi_gpu::device_manager::DevicePerformanceClass;

pub use multi_gpu::load_balancer::LoadBalancer;

pub use multi_gpu::load_balancer::LoadStats;

pub use multi_gpu::sync::Barrier;

pub use multi_gpu::sync::Event;

pub use multi_gpu::sync::Fence;

pub use multi_gpu::sync::GpuSemaphore;

pub use multi_gpu::sync::SemaphoreGuard;

pub use multi_gpu::sync::SyncManager;

pub use multi_gpu::sync::SyncStats;

pub use multi_gpu::work_queue::BatchSubmitter;

pub use multi_gpu::work_queue::WorkQueue;

pub use multi_gpu::work_queue::WorkStealingQueue;

pub use pipeline_builder::Pipeline;

pub use pipeline_builder::PipelineBuilder;

pub use pipeline_builder::PipelineConfig;

pub use pipeline_builder::PipelineInfo;

pub use pipeline_builder::PipelineStage;

pub use profiling::BottleneckKind;

pub use profiling::BottleneckSeverity;

pub use profiling::GpuProfiler;

pub use profiling::KernelStats;

pub use profiling::PerformanceBottleneck;

pub use profiling::ProfileSession;

pub use profiling::ProfilingConfig;

pub use profiling::ProfilingMetrics;

pub use profiling::ProfilingReport;

pub use shader_compiler::CompiledShader;

pub use shader_compiler::CompilerStats;

pub use shader_compiler::ShaderCompiler;

pub use shader_compiler::ShaderPreprocessor;

pub use shader_compiler::analyzer::PerformanceClass;

pub use shader_compiler::analyzer::ShaderAnalysis;

pub use shader_compiler::analyzer::ShaderAnalyzer;

pub use shader_compiler::cache::CacheStats;

pub use shader_compiler::cache::ShaderCache;

pub use shader_compiler::optimizer::OptimizationConfig;

pub use shader_compiler::optimizer::OptimizationLevel;

pub use shader_compiler::optimizer::OptimizationMetrics;

pub use shader_compiler::optimizer::ShaderOptimizer;

Modules

adaptive

Adaptive algorithm selection and auto-tuning for GPUs.

error

Error types for advanced GPU operations.

gpu_ml

GPU-accelerated ML inference for geospatial data.

gpu_terrain

GPU-accelerated terrain analysis algorithms.

kernels

Advanced GPU compute kernels and WGSL shaders.

memory_compaction

GPU memory defragmentation and compaction.

memory_pool

Advanced GPU memory pool with sub-allocation and defragmentation.

multi_gpu

Multi-GPU orchestration and management.

pipeline_builder

GPU compute pipeline builder with fluent API.

profiling

GPU profiling and performance metrics.

shader_compiler

WGSL shader compiler and optimizer.

Structs

LibraryInfo

Library information

Constants

NAME

Library name

VERSION

Library version

Functions

info

Get library information