4c36ddbdc2
- Add synor-compute crate for heterogeneous compute orchestration - Implement processor abstraction for CPU/GPU/TPU/NPU/LPU/FPGA/DSP - Add device registry with cross-vendor capability tracking - Implement task scheduler with work stealing and load balancing - Add energy-aware and latency-aware balancing strategies - Create spot market for compute resources with order matching - Add memory manager with tensor handles and cross-device transfers - Support processor capability profiles (H100, TPU v5p, Groq LPU, etc.) - Implement priority work queues with task decomposition Processor types supported: - CPU (x86-64 AVX512, ARM64 SVE, RISC-V Vector) - GPU (NVIDIA CUDA, AMD ROCm, Intel OneAPI, Apple Metal) - TPU (v2-v5p, Edge TPU) - NPU (Apple Neural Engine, Qualcomm Hexagon, Intel VPU) - LPU (Groq Language Processing Unit) - FPGA (Xilinx, Intel Altera) - DSP (TI, Analog Devices) - WebGPU and WASM runtimes
631 lines
20 KiB
Rust
631 lines
20 KiB
Rust
//! Synor Compute L2 - Heterogeneous Multi-Processor Compute Platform
|
|
//!
|
|
//! Provides decentralized compute services with:
|
|
//!
|
|
//! - **Heterogeneous Scheduling**: CPU + GPU + TPU + NPU + LPU working simultaneously
|
|
//! - **Consumer Device Mesh**: Mobile, browser, desktop devices contributing compute
|
|
//! - **90% Cost Reduction**: Zero margins, spot markets, electricity arbitrage
|
|
//! - **10x Speed**: Caching, speculative execution, optimal processor assignment
|
|
//!
|
|
//! # Architecture
|
|
//!
|
|
//! ```text
|
|
//! ┌─────────────────────────────────────────────────────────────────────────────┐
|
|
//! │ SYNOR COMPUTE L2 │
|
|
//! ├─────────────────────────────────────────────────────────────────────────────┤
|
|
//! │ │
|
|
//! │ ┌─────────────────────────────────────────────────────────────────────────┐ │
|
|
//! │ │ TASK DECOMPOSER │ │
|
|
//! │ │ Analyzes workload → Identifies subtasks → Maps to optimal processors │ │
|
|
//! │ └─────────────────────────────────────────────────────────────────────────┘ │
|
|
//! │ │ │
|
|
//! │ ▼ │
|
|
//! │ ┌─────────────────────────────────────────────────────────────────────────┐ │
|
|
//! │ │ HETEROGENEOUS SCHEDULER │ │
|
|
//! │ │ ┌──────┐ ┌──────┐ ┌──────┐ ┌──────┐ ┌──────┐ ┌──────┐ │ │
|
|
//! │ │ │ CPU │ │ GPU │ │ TPU │ │ NPU │ │ LPU │ │Custom│ │ │
|
|
//! │ │ │Queue │ │Queue │ │Queue │ │Queue │ │Queue │ │Queue │ │ │
|
|
//! │ │ └──────┘ └──────┘ └──────┘ └──────┘ └──────┘ └──────┘ │ │
|
|
//! │ └─────────────────────────────────────────────────────────────────────────┘ │
|
|
//! │ │
|
|
//! │ ┌─────────────────────────────────────────────────────────────────────────┐ │
|
|
//! │ │ UNIFIED MEMORY FABRIC │ │
|
|
//! │ │ Zero-copy data sharing │ Automatic placement │ Cache coherency │ │
|
|
//! │ └─────────────────────────────────────────────────────────────────────────┘ │
|
|
//! │ │
|
|
//! └─────────────────────────────────────────────────────────────────────────────┘
|
|
//! ```
|
|
//!
|
|
//! # Pricing
|
|
//!
|
|
//! | Resource | Unit | Price (SYNOR) |
|
|
//! |----------|------|---------------|
|
|
//! | GPU (consumer) | hour | 0.10 |
|
|
//! | GPU (datacenter) | hour | 0.50-4.00 |
|
|
//! | CPU | core/hour | 0.02 |
|
|
//! | Memory | GB/hour | 0.005 |
|
|
//! | Inference | 1M tokens | 0.10 |
|
|
|
|
#![allow(dead_code)]
|
|
|
|
pub mod device;
|
|
pub mod error;
|
|
pub mod market;
|
|
pub mod memory;
|
|
pub mod processor;
|
|
pub mod scheduler;
|
|
pub mod task;
|
|
|
|
pub use device::{
|
|
DeviceCapabilities, DeviceId, DeviceInfo, DeviceRegistry, DeviceStatus, DeviceType,
|
|
};
|
|
pub use error::ComputeError;
|
|
pub use market::{
|
|
Auction, AuctionId, CloudComparison, CpuTier as MarketCpuTier, GpuTier as MarketGpuTier,
|
|
MarketStats, Order, OrderBook, OrderId, OrderSide, OrderType, PricingEngine, ProviderListing,
|
|
ResourceType, SpotMarket, Trade,
|
|
};
|
|
pub use memory::{MemoryManager, TensorHandle, TransferPath, UnifiedMemory};
|
|
pub use processor::{
|
|
ComputeThroughput, CpuVariant, GpuVariant, NpuVariant, Operation, OperationType, Processor,
|
|
ProcessorCapabilities, ProcessorId, ProcessorType, TpuVersion,
|
|
};
|
|
pub use scheduler::{
|
|
HeterogeneousScheduler, LoadBalancer, Schedule, ScheduleResult, TaskAssignment, WorkQueue,
|
|
};
|
|
pub use task::{
|
|
ComputeTask, DecomposedWorkload, Task, TaskDecomposer, TaskId, TaskPriority, TaskResult,
|
|
TaskStatus,
|
|
};
|
|
|
|
use serde::{Deserialize, Serialize};
|
|
use std::collections::HashMap;
|
|
use std::sync::Arc;
|
|
|
|
use parking_lot::RwLock;
|
|
|
|
/// Compute node identifier.
|
|
#[derive(Clone, Copy, Debug, PartialEq, Eq, Hash, Serialize, Deserialize)]
|
|
pub struct NodeId(pub u64);
|
|
|
|
impl std::fmt::Display for NodeId {
|
|
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
|
|
write!(f, "node_{}", self.0)
|
|
}
|
|
}
|
|
|
|
/// Job identifier.
|
|
#[derive(Clone, Copy, Debug, PartialEq, Eq, Hash, Serialize, Deserialize)]
|
|
pub struct JobId(pub [u8; 32]);
|
|
|
|
impl JobId {
|
|
/// Creates a new job ID.
|
|
pub fn new() -> Self {
|
|
use rand::Rng;
|
|
let mut bytes = [0u8; 32];
|
|
rand::thread_rng().fill(&mut bytes);
|
|
JobId(bytes)
|
|
}
|
|
|
|
/// Creates from bytes.
|
|
pub fn from_bytes(bytes: [u8; 32]) -> Self {
|
|
JobId(bytes)
|
|
}
|
|
}
|
|
|
|
impl Default for JobId {
|
|
fn default() -> Self {
|
|
Self::new()
|
|
}
|
|
}
|
|
|
|
impl std::fmt::Display for JobId {
|
|
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
|
|
write!(f, "job_{}", hex::encode(&self.0[..8]))
|
|
}
|
|
}
|
|
|
|
/// Compute job specification.
|
|
#[derive(Clone, Debug, Serialize, Deserialize)]
|
|
pub struct ComputeJob {
|
|
/// Job ID.
|
|
pub id: JobId,
|
|
/// Owner address.
|
|
pub owner: [u8; 32],
|
|
/// Job type.
|
|
pub job_type: JobType,
|
|
/// Resource requirements.
|
|
pub resources: ResourceRequirements,
|
|
/// Input data reference (CID).
|
|
pub input_cid: Option<String>,
|
|
/// Maximum budget (in atomic SYNOR).
|
|
pub max_budget: u64,
|
|
/// Priority level.
|
|
pub priority: JobPriority,
|
|
/// Created timestamp.
|
|
pub created_at: u64,
|
|
/// Deadline (optional).
|
|
pub deadline: Option<u64>,
|
|
}
|
|
|
|
/// Job type classification.
|
|
#[derive(Clone, Debug, Serialize, Deserialize)]
|
|
pub enum JobType {
|
|
/// AI/ML training job.
|
|
Training {
|
|
framework: MlFramework,
|
|
model_cid: String,
|
|
dataset_cid: String,
|
|
epochs: u32,
|
|
batch_size: u32,
|
|
},
|
|
/// AI/ML inference job.
|
|
Inference {
|
|
model_cid: String,
|
|
input_format: String,
|
|
batch_size: u32,
|
|
},
|
|
/// Container workload.
|
|
Container {
|
|
image: String,
|
|
command: Vec<String>,
|
|
env: HashMap<String, String>,
|
|
},
|
|
/// Serverless function.
|
|
Serverless {
|
|
runtime: FunctionRuntime,
|
|
code_cid: String,
|
|
handler: String,
|
|
},
|
|
/// General compute (WASM).
|
|
Wasm {
|
|
module_cid: String,
|
|
entrypoint: String,
|
|
},
|
|
}
|
|
|
|
/// ML framework specification.
|
|
#[derive(Clone, Debug, Serialize, Deserialize)]
|
|
pub enum MlFramework {
|
|
PyTorch { version: String },
|
|
TensorFlow { version: String },
|
|
JAX { version: String },
|
|
ONNX,
|
|
}
|
|
|
|
/// Function runtime.
|
|
#[derive(Clone, Debug, Serialize, Deserialize)]
|
|
pub enum FunctionRuntime {
|
|
Node20,
|
|
Python312,
|
|
Rust,
|
|
Go,
|
|
Custom { image: String },
|
|
}
|
|
|
|
/// Job priority levels.
|
|
#[derive(Clone, Copy, Debug, PartialEq, Eq, PartialOrd, Ord, Serialize, Deserialize)]
|
|
pub enum JobPriority {
|
|
/// Background job, can be preempted.
|
|
Background = 0,
|
|
/// Normal priority.
|
|
Normal = 1,
|
|
/// High priority, faster scheduling.
|
|
High = 2,
|
|
/// Critical, guaranteed resources.
|
|
Critical = 3,
|
|
}
|
|
|
|
impl Default for JobPriority {
|
|
fn default() -> Self {
|
|
JobPriority::Normal
|
|
}
|
|
}
|
|
|
|
/// Resource requirements for a job.
|
|
#[derive(Clone, Debug, Default, Serialize, Deserialize)]
|
|
pub struct ResourceRequirements {
|
|
/// Minimum CPU cores.
|
|
pub min_cpu_cores: f32,
|
|
/// Minimum memory (GB).
|
|
pub min_memory_gb: f32,
|
|
/// GPU requirements.
|
|
pub gpu: Option<GpuRequirements>,
|
|
/// Preferred processor types (in priority order).
|
|
pub preferred_processors: Vec<ProcessorType>,
|
|
/// Maximum latency (ms) - for inference.
|
|
pub max_latency_ms: Option<u32>,
|
|
/// Requires distributed execution.
|
|
pub distributed: bool,
|
|
}
|
|
|
|
/// GPU resource requirements.
|
|
#[derive(Clone, Debug, Serialize, Deserialize)]
|
|
pub struct GpuRequirements {
|
|
/// Minimum number of GPUs.
|
|
pub min_count: u32,
|
|
/// Maximum number of GPUs.
|
|
pub max_count: u32,
|
|
/// Minimum VRAM per GPU (GB).
|
|
pub min_vram_gb: u32,
|
|
/// Minimum compute capability.
|
|
pub min_compute_capability: Option<(u8, u8)>,
|
|
/// Allow GPU sharing (MPS/MIG).
|
|
pub allow_sharing: bool,
|
|
}
|
|
|
|
/// Job execution status.
|
|
#[derive(Clone, Debug, Serialize, Deserialize)]
|
|
pub enum JobStatus {
|
|
/// Queued, waiting for resources.
|
|
Queued,
|
|
/// Resources allocated, starting.
|
|
Starting,
|
|
/// Running.
|
|
Running {
|
|
progress: f32,
|
|
assigned_nodes: Vec<NodeId>,
|
|
},
|
|
/// Completed successfully.
|
|
Completed {
|
|
result_cid: String,
|
|
duration_ms: u64,
|
|
cost: u64,
|
|
},
|
|
/// Failed.
|
|
Failed { error: String },
|
|
/// Cancelled by user.
|
|
Cancelled,
|
|
}
|
|
|
|
/// Compute node registration.
|
|
#[derive(Clone, Debug, Serialize, Deserialize)]
|
|
pub struct ComputeNode {
|
|
/// Node ID.
|
|
pub id: NodeId,
|
|
/// Owner address.
|
|
pub owner: [u8; 32],
|
|
/// Available processors.
|
|
pub processors: Vec<ProcessorInfo>,
|
|
/// Total memory (GB).
|
|
pub total_memory_gb: f32,
|
|
/// Available memory (GB).
|
|
pub available_memory_gb: f32,
|
|
/// Network bandwidth (Gbps).
|
|
pub bandwidth_gbps: f32,
|
|
/// Geographic region.
|
|
pub region: String,
|
|
/// Stake amount (for PoS).
|
|
pub stake: u64,
|
|
/// Reputation score (0-100).
|
|
pub reputation: u32,
|
|
/// Current status.
|
|
pub status: NodeStatus,
|
|
}
|
|
|
|
/// Processor information on a node.
|
|
#[derive(Clone, Debug, Serialize, Deserialize)]
|
|
pub struct ProcessorInfo {
|
|
/// Processor ID (local to node).
|
|
pub id: ProcessorId,
|
|
/// Processor type.
|
|
pub processor_type: ProcessorType,
|
|
/// Capabilities.
|
|
pub capabilities: ProcessorCapabilities,
|
|
/// Current utilization (0.0 - 1.0).
|
|
pub utilization: f32,
|
|
/// Current temperature (Celsius).
|
|
pub temperature: Option<f32>,
|
|
}
|
|
|
|
/// Node status.
|
|
#[derive(Clone, Copy, Debug, PartialEq, Eq, Serialize, Deserialize)]
|
|
pub enum NodeStatus {
|
|
/// Online and accepting jobs.
|
|
Online,
|
|
/// Online but not accepting new jobs.
|
|
Draining,
|
|
/// Offline.
|
|
Offline,
|
|
/// Maintenance mode.
|
|
Maintenance,
|
|
}
|
|
|
|
/// Compute cluster manager.
|
|
pub struct ComputeCluster {
|
|
/// Registered nodes.
|
|
nodes: RwLock<HashMap<NodeId, ComputeNode>>,
|
|
/// Device registry.
|
|
device_registry: Arc<DeviceRegistry>,
|
|
/// Heterogeneous scheduler.
|
|
scheduler: Arc<HeterogeneousScheduler>,
|
|
/// Spot market.
|
|
spot_market: Arc<SpotMarket>,
|
|
/// Memory manager.
|
|
memory_manager: Arc<MemoryManager>,
|
|
/// Active jobs.
|
|
jobs: RwLock<HashMap<JobId, ComputeJob>>,
|
|
}
|
|
|
|
impl ComputeCluster {
|
|
/// Creates a new compute cluster.
|
|
pub fn new() -> Self {
|
|
let device_registry = Arc::new(DeviceRegistry::new());
|
|
let scheduler = Arc::new(HeterogeneousScheduler::new(device_registry.clone()));
|
|
let spot_market = Arc::new(SpotMarket::new());
|
|
let memory_manager = Arc::new(MemoryManager::new());
|
|
|
|
Self {
|
|
nodes: RwLock::new(HashMap::new()),
|
|
device_registry,
|
|
scheduler,
|
|
spot_market,
|
|
memory_manager,
|
|
jobs: RwLock::new(HashMap::new()),
|
|
}
|
|
}
|
|
|
|
/// Registers a compute node.
|
|
pub fn register_node(&self, node: ComputeNode) -> Result<(), ComputeError> {
|
|
let id = node.id;
|
|
|
|
// Register processors with device registry
|
|
for proc in &node.processors {
|
|
self.device_registry.register_processor(id, proc.clone())?;
|
|
}
|
|
|
|
self.nodes.write().insert(id, node);
|
|
Ok(())
|
|
}
|
|
|
|
/// Unregisters a compute node.
|
|
pub fn unregister_node(&self, node_id: NodeId) -> Result<(), ComputeError> {
|
|
self.device_registry.unregister_node(node_id)?;
|
|
self.nodes.write().remove(&node_id);
|
|
Ok(())
|
|
}
|
|
|
|
/// Submits a job for execution.
|
|
pub async fn submit_job(&self, job: ComputeJob) -> Result<JobId, ComputeError> {
|
|
let job_id = job.id;
|
|
|
|
// Decompose job into tasks
|
|
let tasks = self.decompose_job(&job)?;
|
|
|
|
// Schedule tasks
|
|
let schedule = self.scheduler.schedule(tasks).await?;
|
|
|
|
// Store job
|
|
self.jobs.write().insert(job_id, job);
|
|
|
|
// Execute schedule (async)
|
|
tokio::spawn({
|
|
let scheduler = self.scheduler.clone();
|
|
async move {
|
|
let _ = scheduler.execute(&schedule.schedule).await;
|
|
}
|
|
});
|
|
|
|
Ok(job_id)
|
|
}
|
|
|
|
/// Gets job status.
|
|
pub fn get_job_status(&self, job_id: &JobId) -> Option<JobStatus> {
|
|
self.jobs.read().get(job_id).map(|_| JobStatus::Queued)
|
|
}
|
|
|
|
/// Cancels a job.
|
|
pub fn cancel_job(&self, job_id: &JobId) -> Result<(), ComputeError> {
|
|
if self.jobs.write().remove(job_id).is_some() {
|
|
Ok(())
|
|
} else {
|
|
Err(ComputeError::JobNotFound(*job_id))
|
|
}
|
|
}
|
|
|
|
/// Gets cluster statistics.
|
|
pub fn stats(&self) -> ClusterStats {
|
|
let nodes = self.nodes.read();
|
|
let jobs = self.jobs.read();
|
|
|
|
let total_nodes = nodes.len();
|
|
let online_nodes = nodes.values().filter(|n| n.status == NodeStatus::Online).count();
|
|
|
|
let total_gpus: usize = nodes
|
|
.values()
|
|
.flat_map(|n| &n.processors)
|
|
.filter(|p| matches!(p.processor_type, ProcessorType::Gpu(_)))
|
|
.count();
|
|
|
|
let total_memory: f32 = nodes.values().map(|n| n.total_memory_gb).sum();
|
|
|
|
ClusterStats {
|
|
total_nodes,
|
|
online_nodes,
|
|
total_gpus,
|
|
total_memory_gb: total_memory,
|
|
active_jobs: jobs.len(),
|
|
queued_jobs: jobs.values().filter(|_| true).count(), // Simplified
|
|
}
|
|
}
|
|
|
|
/// Decomposes a job into schedulable tasks.
|
|
fn decompose_job(&self, job: &ComputeJob) -> Result<Vec<Task>, ComputeError> {
|
|
let decomposer = TaskDecomposer::new();
|
|
decomposer.decompose(job)
|
|
}
|
|
}
|
|
|
|
impl Default for ComputeCluster {
|
|
fn default() -> Self {
|
|
Self::new()
|
|
}
|
|
}
|
|
|
|
/// Cluster statistics.
|
|
#[derive(Clone, Debug, Default, Serialize, Deserialize)]
|
|
pub struct ClusterStats {
|
|
/// Total registered nodes.
|
|
pub total_nodes: usize,
|
|
/// Online nodes.
|
|
pub online_nodes: usize,
|
|
/// Total GPUs across cluster.
|
|
pub total_gpus: usize,
|
|
/// Total memory (GB).
|
|
pub total_memory_gb: f32,
|
|
/// Active jobs.
|
|
pub active_jobs: usize,
|
|
/// Queued jobs.
|
|
pub queued_jobs: usize,
|
|
}
|
|
|
|
/// Pricing calculator for compute operations.
|
|
#[derive(Clone, Debug)]
|
|
pub struct ComputePricing {
|
|
/// GPU cost per hour by type.
|
|
pub gpu_hourly: HashMap<GpuTier, u64>,
|
|
/// CPU cost per core-hour.
|
|
pub cpu_core_hour: u64,
|
|
/// Memory cost per GB-hour.
|
|
pub memory_gb_hour: u64,
|
|
/// Network egress per GB.
|
|
pub network_egress_gb: u64,
|
|
/// Inference per million tokens.
|
|
pub inference_per_million_tokens: u64,
|
|
}
|
|
|
|
/// GPU pricing tiers.
|
|
#[derive(Clone, Copy, Debug, PartialEq, Eq, Hash, Serialize, Deserialize)]
|
|
pub enum GpuTier {
|
|
/// Consumer GPUs (RTX 30xx, 40xx).
|
|
Consumer,
|
|
/// Professional GPUs (RTX A series).
|
|
Professional,
|
|
/// Data center GPUs (A100).
|
|
DataCenter,
|
|
/// Latest generation (H100).
|
|
Premium,
|
|
}
|
|
|
|
impl Default for ComputePricing {
|
|
fn default() -> Self {
|
|
let mut gpu_hourly = HashMap::new();
|
|
gpu_hourly.insert(GpuTier::Consumer, 100_000_000); // 0.10 SYNOR
|
|
gpu_hourly.insert(GpuTier::Professional, 300_000_000); // 0.30 SYNOR
|
|
gpu_hourly.insert(GpuTier::DataCenter, 2_000_000_000); // 2.00 SYNOR
|
|
gpu_hourly.insert(GpuTier::Premium, 4_000_000_000); // 4.00 SYNOR
|
|
|
|
Self {
|
|
gpu_hourly,
|
|
cpu_core_hour: 20_000_000, // 0.02 SYNOR
|
|
memory_gb_hour: 5_000_000, // 0.005 SYNOR
|
|
network_egress_gb: 50_000_000, // 0.05 SYNOR
|
|
inference_per_million_tokens: 100_000_000, // 0.10 SYNOR
|
|
}
|
|
}
|
|
}
|
|
|
|
impl ComputePricing {
|
|
/// Estimates cost for a job.
|
|
pub fn estimate(&self, job: &ComputeJob, duration_hours: f32) -> u64 {
|
|
let mut cost = 0u64;
|
|
|
|
// CPU cost
|
|
cost += (self.cpu_core_hour as f32 * job.resources.min_cpu_cores * duration_hours) as u64;
|
|
|
|
// Memory cost
|
|
cost += (self.memory_gb_hour as f32 * job.resources.min_memory_gb * duration_hours) as u64;
|
|
|
|
// GPU cost
|
|
if let Some(gpu) = &job.resources.gpu {
|
|
let tier = GpuTier::Consumer; // Simplified
|
|
let gpu_cost = self.gpu_hourly.get(&tier).unwrap_or(&100_000_000);
|
|
cost += (*gpu_cost as f32 * gpu.min_count as f32 * duration_hours) as u64;
|
|
}
|
|
|
|
cost
|
|
}
|
|
}
|
|
|
|
#[cfg(test)]
|
|
mod tests {
|
|
use super::*;
|
|
|
|
#[test]
|
|
fn test_job_id() {
|
|
let id1 = JobId::new();
|
|
let id2 = JobId::new();
|
|
assert_ne!(id1.0, id2.0);
|
|
}
|
|
|
|
#[test]
|
|
fn test_compute_cluster() {
|
|
let cluster = ComputeCluster::new();
|
|
let stats = cluster.stats();
|
|
assert_eq!(stats.total_nodes, 0);
|
|
}
|
|
|
|
#[test]
|
|
fn test_pricing() {
|
|
let pricing = ComputePricing::default();
|
|
|
|
let job = ComputeJob {
|
|
id: JobId::new(),
|
|
owner: [0u8; 32],
|
|
job_type: JobType::Inference {
|
|
model_cid: "model123".to_string(),
|
|
input_format: "json".to_string(),
|
|
batch_size: 32,
|
|
},
|
|
resources: ResourceRequirements {
|
|
min_cpu_cores: 4.0,
|
|
min_memory_gb: 16.0,
|
|
gpu: Some(GpuRequirements {
|
|
min_count: 1,
|
|
max_count: 1,
|
|
min_vram_gb: 16,
|
|
min_compute_capability: None,
|
|
allow_sharing: false,
|
|
}),
|
|
..Default::default()
|
|
},
|
|
input_cid: None,
|
|
max_budget: 1_000_000_000,
|
|
priority: JobPriority::Normal,
|
|
created_at: 0,
|
|
deadline: None,
|
|
};
|
|
|
|
let cost = pricing.estimate(&job, 1.0);
|
|
assert!(cost > 0);
|
|
}
|
|
|
|
#[test]
|
|
fn test_node_registration() {
|
|
let cluster = ComputeCluster::new();
|
|
|
|
let node = ComputeNode {
|
|
id: NodeId(1),
|
|
owner: [1u8; 32],
|
|
processors: vec![ProcessorInfo {
|
|
id: ProcessorId(0),
|
|
processor_type: ProcessorType::Cpu(CpuVariant::X86_64 {
|
|
avx: processor::AvxSupport::Avx512,
|
|
}),
|
|
capabilities: ProcessorCapabilities::default(),
|
|
utilization: 0.0,
|
|
temperature: Some(45.0),
|
|
}],
|
|
total_memory_gb: 64.0,
|
|
available_memory_gb: 60.0,
|
|
bandwidth_gbps: 10.0,
|
|
region: "us-east".to_string(),
|
|
stake: 1000,
|
|
reputation: 100,
|
|
status: NodeStatus::Online,
|
|
};
|
|
|
|
cluster.register_node(node).unwrap();
|
|
assert_eq!(cluster.stats().total_nodes, 1);
|
|
}
|
|
}
|