Performance Monitoring
Track GPU performance, earnings optimization, and system health to maximize revenue and ensure reliable service delivery.
Dashboard Analytics Overview
Section titled “Dashboard Analytics Overview”Access performance metrics through Dashboard → Containers or Dashboard → My GPUs tabs.
Key Performance Indicators
Section titled “Key Performance Indicators”Utilization Rate
- Target:
>
85%
during available hours - Calculation:
(Active rental hours / Total available hours) × 100
- Low utilization
(<70%)
indicates pricing or availability issues
Revenue Per Hour
- Formula:
Total earnings ÷ Total available hours
- Compare against market rates for similar hardware
- Track weekly trends to identify optimization opportunities
Session Duration
- Average rental length across all bookings
- Longer sessions typically indicate satisfied renters
Sub-30
minute sessions may signal performance issues
Real-Time GPU Monitoring
Section titled “Real-Time GPU Monitoring”Temperature Management
Section titled “Temperature Management”Critical thresholds for common GPUs:
NVIDIA RTX 40 Series:
- Normal operation:
65-75°C
- Throttling begins:
83°C
- Emergency shutdown:
90°C
NVIDIA RTX 30 Series:
- Normal operation:
70-78°C
- Throttling begins:
83°C
- Emergency shutdown:
93°C
AMD RX 7000/6000:
- Normal operation:
75-85°C
- Throttling begins:
90°C
- Emergency shutdown:
95°C
Memory Usage Optimization
Section titled “Memory Usage Optimization”VRAM Allocation:
- Monitor through Dashboard → Containers → GPU Monitor
- Efficient allocation:
85-95%
during active rentals - Memory leaks: Usage remains high between rentals
- Clean state: Less than 500MB usage when idle
System RAM Impact:
- AI workloads:
16-32GB
recommended - Mining operations:
8-16GB
sufficient - Gaming/streaming:
32GB+
optimal
Power Consumption Analysis
Section titled “Power Consumption Analysis”Track power efficiency to optimize electrical costs:
Power Limit Tuning:
- Default:
100%
power limit (maximum performance) - Efficiency mode:
80-90%
limit (reduced heat/noise) - Calculate:
Revenue per watt = Earnings ÷ Power consumed
Cost-Benefit Analysis:
Revenue optimization example (RTX 4090):`100%` power: `$0.60/hour`, `450W = $1.33/kWh``85%` power: `$0.54/hour`, `320W = $1.69/kWh`
Earnings Optimization
Section titled “Earnings Optimization”Dynamic Pricing Strategy
Section titled “Dynamic Pricing Strategy”Market Position Analysis:
- Check competitor rates weekly via marketplace filters
- Price
5-10%
below similar hardware initially - Increase by
$0.05/hour
after achieving>90%
utilization
Demand Pattern Recognition:
- Peak hours: Evenings
(6-11 PM local time)
- Weekend premium:
15-25%
rate increase typical - Holiday demand: Gaming workloads increase
40-60%
Performance-Based Pricing:
- High-end cooling:
+10-15%
premium justified - Consistent
99%+
uptime:+20%
premium - Below-average performance:
-15-25%
discount required
Availability Scheduling
Section titled “Availability Scheduling”Optimal Availability Windows:
24/7
availability:35-50%
higher monthly earnings- Peak-only
(6 PM - 12 AM)
: Higher hourly rates, lower volume - Business hours
(9 AM - 5 PM)
: Enterprise customers, steady demand
Minimum Session Management:
1-hour
minimum: Maximum flexibility, higher turnover costs4-hour
minimum: Reduced overhead, better for intensive workloads8+
hour minimum: Enterprise focus, stable income
System Health Monitoring
Section titled “System Health Monitoring”Hardware Degradation Tracking
Section titled “Hardware Degradation Tracking”GPU Health Metrics:
- Core clock stability:
<2%
variance during load - Memory error rate:
<0.1
errors per hour acceptable - Fan curve efficiency: Monitor
RPM vs temperature
correlation
Early Warning Signs:
- Increasing temperatures with same workloads
- Declining benchmark scores over time
- Higher error rates in mining or compute tasks
- Unusual system crashes or instability
Network Performance Impact
Section titled “Network Performance Impact”Connection Quality Requirements:
- Upload speed: Minimum
50 Mbps
for responsive access - Latency:
<50ms
to major cloud providers optimal - Jitter:
<10ms
variation for smooth experience
Bandwidth Usage Patterns:
- AI training:
10-100 MB/s
sustained transfers - Mining:
<1 MB/s
steady-state - Gaming/streaming:
50-200 MB/s
peak usage
Troubleshooting Common Issues
Section titled “Troubleshooting Common Issues”Low Utilization Diagnosis
Section titled “Low Utilization Diagnosis”Pricing Analysis:
- Compare rates against similar hardware in marketplace
- Check booking rate:
Views → Inquiries → Completed rentals
- Review renter feedback for performance complaints
Technical Issues:
- Container startup time
>60
seconds: Investigate system resources - Connection failures: Verify port forwarding and firewall rules
- Performance complaints: Run GPU benchmarks during idle periods
Performance Degradation
Section titled “Performance Degradation”Systematic Diagnosis:
- Baseline Testing: Run GPU stress tests monthly
- Thermal Analysis: Check cooling system efficiency
- Driver Updates: Maintain latest stable GPU drivers
- System Resource: Monitor CPU, RAM, and storage impact
Resolution Priority:
- Critical (revenue impact
>50%
): Address within4
hours - High (revenue impact
20-50%
): Resolve within24
hours - Medium (revenue impact
<20%
): Schedule maintenance window
Advanced Analytics
Section titled “Advanced Analytics”Revenue Forecasting
Section titled “Revenue Forecasting”Historical Analysis:
- Track monthly earnings trends
- Identify seasonal patterns
- Calculate growth rate
month-over-month
Market Expansion Opportunities:
- Underserved time zones: International renter access
- Specialized workloads: AI inference, rendering, scientific computing
- Bulk reservation: Enterprise customers seeking dedicated access
Competitive Intelligence
Section titled “Competitive Intelligence”Market Position Tracking:
- Weekly rate surveys of comparable hardware
- Utilization rate benchmarking against network averages
- Feature comparison: Cooling, overclocks, additional software
Performance Optimization:
- A/B test pricing strategies over
2-week
periods - Monitor booking patterns after rate adjustments
- Track renter retention and repeat bookings
Reporting and Analytics
Section titled “Reporting and Analytics”Daily Operations Dashboard
Section titled “Daily Operations Dashboard”Access through Dashboard → Analytics (if available) or compile manually:
Key Metrics Summary:
- Total active hours today
- Revenue generated
(gross/net)
- Average GPU temperature and utilization
- Number of unique renters served
Weekly Performance Review
Section titled “Weekly Performance Review”Revenue Analysis:
- Week-over-week growth percentage
- Utilization rate trends
- Average session duration changes
- Pricing effectiveness evaluation
Technical Performance:
- System uptime percentage
- Average response times
- Hardware health score trends
- Support ticket volume and resolution
Monthly Optimization Reports
Section titled “Monthly Optimization Reports”Financial Summary:
- Total earnings vs. electricity costs
- Revenue per available hour
- Market share within your hardware category
- ROI calculation including depreciation
Strategic Planning:
- Hardware upgrade recommendations
- Market expansion opportunities
- Competitive positioning analysis
- Growth trajectory projections