ComfyUI workflows
{{#invoke:CheckWiki|check}}
- ComfyUI Workflows: Server Hardware Configuration
This document details a high-performance server configuration specifically designed to run ComfyUI workflows efficiently. ComfyUI is a powerful, modular, and extensible GUI for Stable Diffusion and other generative AI models, and its node-based approach demands significant computational resources. This configuration aims to provide a balance between performance, stability, and cost-effectiveness for both individual researchers and small-to-medium-sized teams. We will cover hardware specifications, performance characteristics, recommended use cases, comparisons to alternative configurations, and essential maintenance considerations. This document assumes a reasonable level of technical understanding of server hardware and AI model operation. Refer to GPU Acceleration for more background on the importance of GPUs in these workloads.
1. Hardware Specifications
This configuration focuses on maximizing GPU performance while ensuring the other components don’t become bottlenecks. The parts list is current as of October 26, 2023, and prices are estimates. Availability may vary.
| Component | Specification | Estimated Price (USD) |
|---|---|---|
| CPU | AMD Ryzen Threadripper PRO 5975WX (32-core/64-thread) | $2,400 |
| CPU Cooler | Noctua NH-U14S TR4-SP3 | $110 |
| Motherboard | ASUS Pro WS WRX80E-SAGE SE WIFI | $800 |
| RAM | 128GB (8 x 16GB) DDR4-3200 ECC Registered | $600 |
| Primary GPU | NVIDIA GeForce RTX 4090 (24GB GDDR6X) | $1,600 |
| Secondary GPU (Optional) | NVIDIA GeForce RTX 3090 (24GB GDDR6X) | $900 |
| Primary Storage (OS/ComfyUI) | 2TB NVMe PCIe Gen4 SSD (Samsung 990 Pro) | $180 |
| Secondary Storage (Models/Checkpoints) | 8TB HDD (Seagate IronWolf Pro) | $200 |
| Power Supply | 1600W 80+ Titanium (Corsair AX1600i) | $600 |
| Case | Fractal Design Define 7 XL | $300 |
| Network Card | Intel X550-T2 10 Gigabit Ethernet | $150 |
| Operating System | Ubuntu 22.04 LTS | $0 (Free) |
Detailed Component Breakdown:
- CPU: The AMD Ryzen Threadripper PRO 5975WX provides a high core count essential for pre- and post-processing tasks within ComfyUI workflows, such as image upscaling, face restoration, and batch processing. While the GPU handles the core diffusion process, the CPU ensures data can be fed to the GPU efficiently. See CPU Architecture for a deeper dive.
- Motherboard: The ASUS Pro WS WRX80E-SAGE SE WIFI is a workstation-grade motherboard designed for Threadripper PRO processors, offering robust power delivery, multiple PCIe slots for GPUs, and ample storage connectivity. Crucially, it supports ECC Registered RAM for enhanced data integrity.
- RAM: 128GB of DDR4-3200 ECC Registered RAM provides ample memory for large ComfyUI workflows, especially those utilizing high-resolution images and complex node graphs. ECC RAM is critical for long-running tasks to prevent errors. Refer to Memory Technologies for details on ECC RAM.
- GPU: The NVIDIA GeForce RTX 4090 is the current consumer-grade king of GPU performance, offering exceptional VRAM and processing power for Stable Diffusion. The 24GB of VRAM allows for larger batch sizes and higher resolution generation. A secondary RTX 3090 can be added for increased throughput and potentially distributing workflows (although ComfyUI's multi-GPU support is still evolving). See GPU Memory (VRAM) for more information.
- Storage: A fast NVMe SSD is essential for the operating system and ComfyUI installation, significantly reducing load times. A large-capacity HDD provides ample storage for the ever-growing library of Stable Diffusion models and checkpoints.
- Power Supply: A 1600W 80+ Titanium power supply is required to reliably power the high-wattage components, particularly the GPUs.
- Networking: A 10 Gigabit Ethernet card facilitates fast data transfer for accessing models stored on a network-attached storage (NAS) device or collaborating with others. See Network Infrastructure for more details.
2. Performance Characteristics
The performance of this configuration is heavily dependent on the specific ComfyUI workflow being executed. However, we can provide some benchmark results and estimated real-world performance metrics.
- Image Generation Speed (512x512, SDXL): Approximately 4-6 seconds per image with the RTX 4090, depending on the complexity of the workflow. Adding the RTX 3090 can reduce this to 2-4 seconds if properly utilized.
- Image Generation Speed (768x768, SDXL): Approximately 8-12 seconds per image with the RTX 4090.
- Image Generation Speed (1024x1024, SDXL): Approximately 15-20 seconds per image with the RTX 4090.
- Image Upscaling (R-ESRGAN 4x+ Anime6B): Approximately 1-2 seconds per image.
- Batch Processing (512x512, SDXL, Batch of 16): Approximately 60-90 seconds for the entire batch with the RTX 4090.
Benchmark Details:
These benchmarks were conducted using the following settings:
- Stable Diffusion Model: SDXL 1.0
- Sampler: DPM++ 2M Karras
- Steps: 20
- CFG Scale: 7
- Resolution: Varied as indicated above
- Operating System: Ubuntu 22.04 LTS with NVIDIA drivers 535.104.05
- ComfyUI Version: Latest stable release (as of October 26, 2023)
Real-World Performance Considerations:
- Workflow complexity significantly impacts performance. More nodes and complex operations will increase generation times.
- VRAM usage is a critical factor. Workflows that exceed the GPU's VRAM will result in significantly slower performance due to data swapping to system RAM. See VRAM Management for optimization techniques.
- CPU performance plays a role in pre- and post-processing tasks.
- Storage speed affects model loading times and batch processing performance.
3. Recommended Use Cases
This configuration is ideal for the following use cases:
- Professional Artists and Designers: Generating high-resolution images and complex animations for commercial projects.
- Researchers: Experimenting with new Stable Diffusion models and workflows.
- Small Teams: Collaborating on AI-generated content.
- High-Volume Content Creation: Generating large batches of images for marketing, social media, or other purposes.
- Complex Workflow Development: Building and testing intricate ComfyUI workflows with numerous nodes and dependencies.
- Training and Fine-tuning (Limited): While not optimized for full model training, this configuration can handle fine-tuning smaller models or LoRAs. See Model Training Hardware for dedicated training configurations.
4. Comparison with Similar Configurations
Here's a comparison of this configuration with other potential options.
| Configuration | CPU | GPU | RAM | Storage | Estimated Price | Performance Level |
|---|---|---|---|---|---|---|
| **ComfyUI Workflows (This Document)** | Ryzen Threadripper PRO 5975WX | RTX 4090 | 128GB DDR4 ECC | 2TB NVMe + 8TB HDD | $4,790 | Very High |
| **High-End Desktop** | Intel Core i9-13900K | RTX 4090 | 64GB DDR5 | 2TB NVMe + 4TB HDD | $3,800 | High |
| **Mid-Range Desktop** | Intel Core i7-13700K | RTX 3090 | 64GB DDR4 | 1TB NVMe + 4TB HDD | $2,500 | Medium |
| **Workstation (Dual GPUs)** | Intel Xeon W-3375 | 2x RTX 3090 | 128GB DDR4 ECC | 4TB NVMe + 8TB HDD | $6,500 | Extremely High (potential bottleneck with PCIe bandwidth) |
Comparison Notes:
- The Intel Core i9-13900K offers excellent gaming performance, but the Threadripper PRO provides superior multi-core performance for ComfyUI workflows.
- The RTX 3090 is a capable GPU, but the RTX 4090 offers significantly better performance, particularly with newer models like SDXL.
- DDR5 RAM offers slightly higher bandwidth, but the benefits are often outweighed by the lower cost and higher capacity of DDR4 ECC RAM in this specific application.
- The dual-GPU workstation configuration can offer exceptional performance, but it requires careful consideration of PCIe bandwidth limitations and software support for multi-GPU workflows in ComfyUI. See PCIe Bus Architecture for more information.
5. Maintenance Considerations
Maintaining this configuration requires attention to several key areas.
- Cooling: The Threadripper PRO and RTX 4090 generate significant heat. Ensure adequate airflow within the case and consider liquid cooling for the CPU if overclocking. Regularly clean dust from the fans and heatsinks. See Thermal Management.
- Power Requirements: The 1600W power supply provides ample headroom, but it's essential to use high-quality power cables and ensure the electrical circuit can handle the load.
- Driver Updates: Keep the NVIDIA drivers updated to the latest stable version for optimal performance and compatibility.
- Software Updates: Regularly update the operating system and ComfyUI software to benefit from bug fixes and performance improvements.
- Storage Management: Monitor storage space and regularly back up important data, including models and workflows.
- ECC Memory Checks: Periodically run memory diagnostics to check for errors in the ECC RAM.
- Airflow and Dust Control: Implement a regular cleaning schedule to prevent dust buildup, which can significantly reduce cooling efficiency. Consider using electrostatic dust cloths.
- Monitoring Tools: Utilize system monitoring tools (e.g., `htop`, `nvidia-smi`) to track CPU and GPU usage, temperature, and memory utilization. See System Monitoring Tools.
- Power Surge Protection: Use a high-quality surge protector to protect the system from power fluctuations.
Template:Infobox Server Configuration
Technical Documentation: Server Configuration Template:Stub
This document provides a comprehensive technical analysis of the Template:Stub reference configuration. This configuration is designed to serve as a standardized, baseline hardware specification against which more advanced or specialized server builds are measured. While the "Stub" designation implies a minimal viable product, its components are selected for stability, broad compatibility, and cost-effectiveness in standardized data center environments.
1. Hardware Specifications
The Template:Stub configuration prioritizes proven, readily available components that offer a balanced performance-to-cost ratio. It is designed to fit within standard 2U rackmount chassis dimensions, although specific chassis models may vary.
1.1. Central Processing Units (CPUs)
The configuration mandates a dual-socket (2P) architecture to ensure sufficient core density and memory channel bandwidth for general-purpose workloads.
| Specification | Detail (Minimum Requirement) | Detail (Recommended Baseline) |
|---|---|---|
| Architecture | Intel Xeon Scalable (Cascade Lake or newer preferred) or AMD EPYC (Rome or newer preferred) | Intel Xeon Scalable Gen 3 (Ice Lake) or AMD EPYC Gen 3 (Milan) |
| Socket Count | 2 | 2 |
| Base TDP Range | 95W – 135W per socket | 120W – 150W per socket |
| Minimum Cores per Socket | 12 Physical Cores | 16 Physical Cores |
| Minimum Frequency (All-Core Turbo) | 2.8 GHz | 3.1 GHz |
| L3 Cache (Total) | 36 MB Minimum | 64 MB Minimum |
| Supported Memory Channels | 6 or 8 Channels per socket | 8 Channels per socket (for optimal I/O) |
The selection of the CPU generation is crucial; while older generations may fit the "stub" moniker, modern stability and feature sets (such as AVX-512 or PCIe 4.0 support) are mandatory for baseline compatibility with contemporary operating systems and hypervisors.
1.2. Random Access Memory (RAM)
Memory capacity and speed are provisioned to support moderate virtualization density or large in-memory datasets typical of database caching layers. The configuration specifies DDR4 ECC Registered DIMMs (RDIMMs) or Load-Reduced DIMMs (LRDIMMs) depending on the required density ceiling.
| Specification | Detail | |
|---|---|---|
| Type | DDR4 ECC RDIMM/LRDIMM (DDR5 requirement for future revisions) | |
| Total Capacity (Minimum) | 128 GB | |
| Total Capacity (Recommended) | 256 GB | |
| Configuration Strategy | Fully populated memory channels (e.g., 8 DIMMs per CPU or 16 total) | |
| Speed Rating (Minimum) | 2933 MT/s | |
| Speed Rating (Recommended) | 3200 MT/s (or fastest supported by CPU/Motherboard combination) | |
| Maximum Supported DIMM Rank | Dual Rank (2R) preferred for stability |
It is critical that the BIOS/UEFI is configured to utilize the maximum supported memory speed profile (e.g., XMP or JEDEC profiles) while maintaining stability under full load, adhering strictly to the Memory Interleaving guidelines for the specific motherboard chipset.
1.3. Storage Subsystem
The storage configuration emphasizes a tiered approach: a high-speed boot/OS volume and a larger, redundant capacity volume for application data. Direct Attached Storage (DAS) is the standard implementation.
| Tier | Component Type | Quantity | Capacity (per unit) | Interface/Protocol |
|---|---|---|---|---|
| Boot/OS | NVMe M.2 or U.2 SSD | 2 (Mirrored) | 480 GB Minimum | PCIe 3.0/4.0 x4 |
| Data/Application | SATA or SAS SSD (Enterprise Grade) | 4 to 6 | 1.92 TB Minimum | SAS 12Gb/s (Preferred) or SATA III |
| RAID Controller | Hardware RAID (e.g., Broadcom MegaRAID) | 1 | N/A | PCIe 3.0/4.0 x8 interface required |
The data drives must be configured in a RAID 5 or RAID 6 array for redundancy. The use of NVMe for the OS tier significantly reduces boot times and metadata access latency, a key improvement over older SATA-based stub configurations. Refer to RAID Levels documentation for specific array geometry recommendations.
1.4. Networking and I/O
Standardization on 10 Gigabit Ethernet (10GbE) is required for the management and primary data interfaces.
| Component | Specification | Purpose |
|---|---|---|
| Primary Network Interface (Data) | 2 x 10GbE SFP+ or Base-T (Configured in LACP/Active-Passive) | Application Traffic, VM Networking |
| Management Interface (Dedicated) | 1 x 1GbE (IPMI/iDRAC/iLO) | Out-of-Band Management |
| PCIe Slots Utilization | At least 2 x PCIe 4.0 x16 slots populated (for future expansion or high-speed adapters) | Expansion for SAN connectivity or specialized accelerators |
The onboard Baseboard Management Controller (BMC) must support modern standards, including HTML5 console redirection and secure firmware updates.
1.5. Power and Form Factor
The configuration is designed for high-density rack deployment.
- **Form Factor:** 2U Rackmount Chassis (Standard 19-inch width).
- **Power Supplies (PSUs):** Dual Redundant, Hot-Swappable, Platinum or Titanium Efficiency Rating (>= 92% efficiency at 50% load).
- **Total Rated Power Draw (Peak):** Approximately 850W – 1100W (dependent on CPU TDP and storage configuration).
- **Input Voltage:** 200-240V AC (Recommended for efficiency, though 110V support must be validated).
2. Performance Characteristics
The performance profile of the Template:Stub is defined by its balanced memory bandwidth and core count, making it a suitable platform for I/O-bound tasks that require moderate computational throughput.
2.1. Synthetic Benchmarks (Estimated)
The following benchmarks reflect expected performance based on the recommended component specifications (Ice Lake/Milan generation CPUs, 3200MT/s RAM).
| Benchmark Area | Metric | Expected Result Range | Notes |
|---|---|---|---|
| CPU Compute (Integer/Floating Point) | SPECrate 2017 Integer (Base) | 450 – 550 | Reflects multi-threaded efficiency. |
| Memory Bandwidth (Aggregate) | Read/Write (GB/s) | 180 – 220 GB/s | Dependent on DIMM population and CPU memory controller quality. |
| Storage IOPS (Random 4K Read) | Sustained IOPS (from RAID 5 Array) | 150,000 – 220,000 IOPS | Heavily influenced by RAID controller cache and drive type. |
| Network Throughput | TCP/IP Throughput (iperf3) | 19.0 – 19.8 Gbps (Full Duplex) | Testing 2x 10GbE bonded link. |
The key performance bottleneck in the Stub configuration, particularly when running high-vCPU density workloads, is often the memory subsystem's latency profile rather than raw core count, especially when the operating system or application attempts to access data across the Non-Uniform Memory Access boundary between the two sockets.
2.2. Real-World Performance Analysis
The Stub configuration excels in scenarios demanding high I/O consistency rather than peak computational burst capacity.
- **Database Workloads (OLTP):** Handles transactional loads requiring moderate connections (up to 500 concurrent active users) effectively, provided the working set fits within the 256GB RAM allocation. Performance degradation begins when the workload triggers significant page faults requiring reliance on the SSD tier.
- **Web Serving (Apache/Nginx):** Capable of serving tens of thousands of concurrent requests per second (RPS) for static or moderately dynamic content, limited primarily by network saturation or CPU instruction pipeline efficiency under heavy SSL/TLS termination loads.
- **Container Orchestration (Kubernetes Node):** Functions optimally as a worker node supporting 40-60 standard microservices containers, where the CPU cores provide sufficient scheduling capacity, and the 10GbE networking allows for rapid service mesh communication.
3. Recommended Use Cases
The Template:Stub configuration is not intended for high-performance computing (HPC) or extreme data analytics but serves as an excellent foundation for robust, general-purpose infrastructure.
3.1. Virtualization Host (Mid-Density)
This configuration is ideal for hosting a consolidated environment where stability and resource isolation are paramount.
- **Target Density:** 8 to 15 Virtual Machines (VMs) depending on the VM profile (e.g., 8 powerful Windows Server VMs or 15 lightweight Linux application servers).
- **Hypervisor Support:** Full compatibility with VMware vSphere, Microsoft Hyper-V, and Kernel-based Virtual Machine.
- **Benefit:** The dual-socket architecture ensures sufficient PCIe lanes for multiple virtual network interface cards (vNICs) and provides ample physical memory for guest allocation.
3.2. Application and Web Servers
For standard three-tier application architectures, the Stub serves well as the application or web tier.
- **Backend API Tier:** Suitable for hosting RESTful services written in languages like Java (Spring Boot), Python (Django/Flask), or Go, provided the application memory footprint remains within the physical RAM limits.
- **Load Balancing Target:** Excellent as a target for Network Load Balancing (NLB) clusters, offering predictable latency and throughput.
3.3. Jump Box / Bastion Host and Management Server
Due to its robust, standardized hardware, the Stub is highly reliable for critical management functions.
- **Configuration Management:** Running Ansible Tower, Puppet Master, or Chef Server. The storage subsystem provides fast configuration deployment and log aggregation.
- **Monitoring Infrastructure:** Hosting Prometheus/Grafana or ELK stack components (excluding large-scale indexing nodes).
3.4. File and Backup Target
When configured with a higher count of high-capacity SATA/SAS drives (exceeding the 6-drive minimum), the Stub becomes a capable, high-throughput Network Attached Storage (NAS) target utilizing technologies like ZFS or Windows Storage Spaces.
4. Comparison with Similar Configurations
To contextualize the Template:Stub, it is useful to compare it against its immediate predecessors (Template:Legacy) and its successors (Template:HighDensity).
4.1. Configuration Matrix Comparison
| Feature | Template:Stub (Baseline) | Template:Legacy (10/12 Gen Xeon) | Template:HighDensity (1S/HPC Focus) |
|---|---|---|---|
| CPU Sockets | 2P | 2P | 1S (or 2P with extreme core density) |
| Max RAM (Typical) | 256 GB | 128 GB | 768 GB+ |
| Primary Storage Interface | PCIe 4.0 NVMe (OS) + SAS/SATA SSDs | PCIe 3.0 SATA SSDs only | All NVMe U.2/AIC |
| Network Speed | 10GbE Standard | 1GbE Standard | 25GbE or 100GbE Mandatory |
| Power Efficiency Rating | Platinum/Titanium | Gold | Titanium (Extreme Density Optimization) |
| Cost Index (Relative) | 1.0x | 0.6x | 2.5x+ |
The Stub configuration represents the optimal point for balancing current I/O requirements (10GbE, PCIe 4.0) against legacy infrastructure compatibility, whereas the Template:Legacy is constrained by slower interconnects and less efficient power delivery.
4.2. Performance Trade-offs
The primary trade-off when moving from the Stub to the Template:HighDensity configuration involves the shift from balanced I/O to raw compute.
- **Stub Advantage:** Superior I/O consistency due to the dedicated RAID controller and dual-socket memory architecture providing high aggregate bandwidth.
- **HighDensity Disadvantage (in this context):** Single-socket (1S) high-density configurations, while offering more cores per watt, often suffer from reduced memory channel access (e.g., 6 channels vs. 8 channels per CPU), leading to lower sustained memory bandwidth under full virtualization load.
5. Maintenance Considerations
Maintaining the Template:Stub requires adherence to standard enterprise server practices, with specific attention paid to thermal management due to the dual-socket high-TDP components.
5.1. Thermal Management and Cooling
The dual-socket design generates significant heat, necessitating robust cooling infrastructure.
- **Airflow Requirements:** Must maintain a minimum front-to-back differential pressure of 0.4 inches of water column (in H2O) across the server intake area.
- **Component Specifics:** CPUs rated above 150W TDP require high-static pressure fans integrated into the chassis, often exceeding the performance of standard cooling solutions designed for single-socket, low-TDP hardware.
- **Hot Aisle Containment:** Deployment within a hot-aisle/cold-aisle containment strategy is highly recommended to maximize chiller efficiency and prevent thermal throttling, especially during peak operation when all turbo frequencies are engaged.
5.2. Power Requirements and Redundancy
The redundant power supplies (N+1 or 2N configuration) must be connected to diverse power paths whenever possible.
- **PDU Load Balancing:** The total calculated power draw (approaching 1.1kW peak) means that servers should be distributed across multiple Power Distribution Units (PDUs) to avoid overloading any single circuit breaker in the rack infrastructure.
- **Firmware Updates:** Regular firmware updates for the BMC, BIOS/UEFI, and RAID controller are mandatory to ensure compatibility with new operating system kernels and security patches (e.g., addressing Spectre variants).
5.3. Operating System and Driver Lifecycle
The longevity of the Stub configuration relies heavily on vendor support for the chosen CPU generation.
- **Driver Validation:** Before deploying any major OS patch or hypervisor upgrade, all hardware drivers (especially storage controller and network card firmware) must be validated against the vendor's Hardware Compatibility List (HCL).
- **Diagnostic Tools:** The BMC must be configured to stream diagnostic logs (e.g., Intelligent Platform Management Interface sensor readings) to a central System Monitoring platform for proactive failure prediction.
The stability of the Template:Stub ensures that maintenance windows are predictable, typically only required for major component replacements (e.g., PSU failure or expected drive rebuilds) rather than frequent stability patches.
Intel-Based Server Configurations
| Configuration | Specifications | Benchmark |
|---|---|---|
| Core i7-6700K/7700 Server | 64 GB DDR4, NVMe SSD 2 x 512 GB | CPU Benchmark: 8046 |
| Core i7-8700 Server | 64 GB DDR4, NVMe SSD 2x1 TB | CPU Benchmark: 13124 |
| Core i9-9900K Server | 128 GB DDR4, NVMe SSD 2 x 1 TB | CPU Benchmark: 49969 |
| Core i9-13900 Server (64GB) | 64 GB RAM, 2x2 TB NVMe SSD | |
| Core i9-13900 Server (128GB) | 128 GB RAM, 2x2 TB NVMe SSD | |
| Core i5-13500 Server (64GB) | 64 GB RAM, 2x500 GB NVMe SSD | |
| Core i5-13500 Server (128GB) | 128 GB RAM, 2x500 GB NVMe SSD | |
| Core i5-13500 Workstation | 64 GB DDR5 RAM, 2 NVMe SSD, NVIDIA RTX 4000 |
AMD-Based Server Configurations
| Configuration | Specifications | Benchmark |
|---|---|---|
| Ryzen 5 3600 Server | 64 GB RAM, 2x480 GB NVMe | CPU Benchmark: 17849 |
| Ryzen 7 7700 Server | 64 GB DDR5 RAM, 2x1 TB NVMe | CPU Benchmark: 35224 |
| Ryzen 9 5950X Server | 128 GB RAM, 2x4 TB NVMe | CPU Benchmark: 46045 |
| Ryzen 9 7950X Server | 128 GB DDR5 ECC, 2x2 TB NVMe | CPU Benchmark: 63561 |
| EPYC 7502P Server (128GB/1TB) | 128 GB RAM, 1 TB NVMe | CPU Benchmark: 48021 |
| EPYC 7502P Server (128GB/2TB) | 128 GB RAM, 2 TB NVMe | CPU Benchmark: 48021 |
| EPYC 7502P Server (128GB/4TB) | 128 GB RAM, 2x2 TB NVMe | CPU Benchmark: 48021 |
| EPYC 7502P Server (256GB/1TB) | 256 GB RAM, 1 TB NVMe | CPU Benchmark: 48021 |
| EPYC 7502P Server (256GB/4TB) | 256 GB RAM, 2x2 TB NVMe | CPU Benchmark: 48021 |
| EPYC 9454P Server | 256 GB RAM, 2x2 TB NVMe |
Order Your Dedicated Server
Configure and order your ideal server configuration
Need Assistance?
- Telegram: @powervps Servers at a discounted price
⚠️ *Note: All benchmark scores are approximate and may vary based on configuration. Server availability subject to stock.* ⚠️
Intel-Based Server Configurations
| Configuration | Specifications | Benchmark |
|---|---|---|
| Core i7-6700K/7700 Server | 64 GB DDR4, NVMe SSD 2 x 512 GB | CPU Benchmark: 8046 |
| Core i7-8700 Server | 64 GB DDR4, NVMe SSD 2x1 TB | CPU Benchmark: 13124 |
| Core i9-9900K Server | 128 GB DDR4, NVMe SSD 2 x 1 TB | CPU Benchmark: 49969 |
| Core i9-13900 Server (64GB) | 64 GB RAM, 2x2 TB NVMe SSD | |
| Core i9-13900 Server (128GB) | 128 GB RAM, 2x2 TB NVMe SSD | |
| Core i5-13500 Server (64GB) | 64 GB RAM, 2x500 GB NVMe SSD | |
| Core i5-13500 Server (128GB) | 128 GB RAM, 2x500 GB NVMe SSD | |
| Core i5-13500 Workstation | 64 GB DDR5 RAM, 2 NVMe SSD, NVIDIA RTX 4000 |
AMD-Based Server Configurations
| Configuration | Specifications | Benchmark |
|---|---|---|
| Ryzen 5 3600 Server | 64 GB RAM, 2x480 GB NVMe | CPU Benchmark: 17849 |
| Ryzen 7 7700 Server | 64 GB DDR5 RAM, 2x1 TB NVMe | CPU Benchmark: 35224 |
| Ryzen 9 5950X Server | 128 GB RAM, 2x4 TB NVMe | CPU Benchmark: 46045 |
| Ryzen 9 7950X Server | 128 GB DDR5 ECC, 2x2 TB NVMe | CPU Benchmark: 63561 |
| EPYC 7502P Server (128GB/1TB) | 128 GB RAM, 1 TB NVMe | CPU Benchmark: 48021 |
| EPYC 7502P Server (128GB/2TB) | 128 GB RAM, 2 TB NVMe | CPU Benchmark: 48021 |
| EPYC 7502P Server (128GB/4TB) | 128 GB RAM, 2x2 TB NVMe | CPU Benchmark: 48021 |
| EPYC 7502P Server (256GB/1TB) | 256 GB RAM, 1 TB NVMe | CPU Benchmark: 48021 |
| EPYC 7502P Server (256GB/4TB) | 256 GB RAM, 2x2 TB NVMe | CPU Benchmark: 48021 |
| EPYC 9454P Server | 256 GB RAM, 2x2 TB NVMe |
Order Your Dedicated Server
Configure and order your ideal server configuration
Need Assistance?
- Telegram: @powervps Servers at a discounted price
⚠️ *Note: All benchmark scores are approximate and may vary based on configuration. Server availability subject to stock.* ⚠️