Dell PowerEdge R660 Specs and Features
Meet the PowerEdge R660 — Dense Compute for Demanding Workloads
-
1U dual-socket rack server — compact footprint engineered for maximum compute density per rack unit
-
Supports up to two 4th Gen or 5th Gen Intel Xeon Scalable processors — up to 64 cores total (4th Gen) or more with 5th Gen Emerald Rapids
-
32 DDR5 RDIMM slots supporting up to 8 TB of ECC memory — double the capacity of 15th-Gen DDR4 platforms
-
Up to 10 x 2.5" front drive bays — SAS 24 Gb, SATA, or NVMe Gen4; additional EDSFF E3.S Gen5 NVMe configurations available
-
PCIe Gen5 expansion — up to 3 low-profile PCIe slots for high-bandwidth networking and storage adapters
-
Supports optional Direct Liquid Cooling (DLC) for maximum thermal efficiency in high-TDP processor configurations
-
Integrated iDRAC9 with Redfish API, Quick Sync 2 wireless, and comprehensive OpenManage ecosystem support
-
Chassis dimensions: 42.8 mm H × 482 mm W × 822.88 mm D (with bezel) — 1.68" × 18.97" × 32.39"
Built for High-Density Virtualization, Database Analytics, and Mixed Workloads
-
High Density Virtualization — dual CPUs and up to 8 TB DDR5 RAM consolidate dozens of VMs in a single 1U chassis, reducing rack footprint and licensing costs
-
Dense Database Analytics (VDI) — fast DDR5 at up to 4800 MT/s and NVMe Gen4 storage eliminate memory and I/O bottlenecks in OLAP, OLTP, and virtual desktop environments
-
Mixed Workload Standardization — flexible drive bay configurations (LFF, SFF, NVMe, EDSFF) allow one chassis to serve multiple roles across your data center
-
HPC & Scientific Computing — up to 56 cores (4th Gen) / 64 cores (5th Gen) per system, paired with PCIe Gen5 fabric for compute-intensive scientific workloads
-
AI Inference & GPU Acceleration — up to 3 x 75 W single-wide GPUs in low-profile slots support ML inference, visualization, and distributed AI pipelines
-
Enterprise Middleware — ERP, CRM, and business-critical applications benefit from dual-socket NUMA architecture, redundant PSUs, and enterprise RAS features
-
Liquid-Cooled Dense Deployments — optional DLC (Direct Liquid Cooling) enables high-TDP processor configurations (up to 350 W) in high-density racks
4th & 5th Gen Intel Xeon Scalable — Up to 64 Cores and 350 W TDP
-
Up to two 4th Gen Intel Xeon Scalable (Sapphire Rapids) processors — up to 56 cores per CPU, up to 350 W TDP — or two 5th Gen Intel Xeon Scalable (Emerald Rapids) processors with up to 64 cores
-
Top 4th Gen SKU: Xeon Platinum 8480+ — 56 cores, 2 GHz base, 105 MB cache, 16 GT/s UPI, 350 W TDP, 4800 MT/s DDR5, 6 TB max memory
-
Intel Ultra Path Interconnect (UPI) — 3 UPI links per CPU at up to 16 GT/s for high-bandwidth multi-socket interconnect
-
PCIe Gen5 I/O fabric — up to 80 PCIe Gen5 lanes per processor at 32 GT/s, doubling Gen4 bandwidth for storage and networking
-
Built-in Intel accelerators — hardware-accelerated data analytics (IAA), cryptography (QAT), networking (DSA), and data compression at silicon level
-
TDP range from 250 W to 350 W (air cooling) — liquid cooling unlocks 350 W TDP SKUs including the 8470Q and 8471N in demanding deployments
-
DDR5 memory support up to 4800 MT/s (1 DIMM per channel) with 4th Gen / up to 5600 MT/s with 5th Gen; 16 DIMMs per socket, 32 total
-
Also supports Intel Xeon CPU Max (formerly codename HBM) series with high-bandwidth on-package memory for in-memory analytics workloads
DDR5 ECC RDIMM — 32 Slots, Up to 8 TB at 4800 MT/s
-
32 DDR5 RDIMM slots — 16 slots per CPU, fully populated for up to 8 TB of ECC-protected system RAM
-
Memory speeds: 4800 MT/s (1 DIMM per channel) and 4400 MT/s (2 DIMMs per channel) with 4th Gen Intel Xeon — up to 5600 MT/s with 5th Gen
-
Supported DIMM capacities: 16 GB, 32 GB, 64 GB, 128 GB, and 256 GB RDIMMs — maximum configuration: 32 × 256 GB = 8 TB
-
ECC (Error Correcting Code) protection on all DIMMs — detects and corrects single-bit errors in real time, critical for mission-critical workloads
-
Operating voltage: 1.1 V — DDR5 runs more efficiently than DDR4 (1.2 V), reducing per-DIMM power draw at higher speeds and densities
-
Single-rank (x8) and dual-rank / quad-rank (x4) RDIMM configurations supported for optimal capacity-to-bandwidth balance
-
Represents a major leap over the R650's DDR4 subsystem — 50% faster memory bus and ECC RDIMM-only design (no LRDIMM required for 8 TB)
Up to 10 × 2.5-Inch SAS/SATA/NVMe — Multiple Front Bay Configurations
-
Zero drives (no backplane) — stripped chassis for direct-attach or external storage deployments
-
8 × 2.5" NVMe (or NVMe RAID) — all-flash NVMe configuration for maximum IOPS; up to 122.88 TB raw capacity
-
8 × 2.5" SAS4/SATA or Universal — mixed HDD/SSD bay configuration for tiered storage deployments
-
10 × 2.5" SAS4/SATA or NVMe — maximum drive count front bay; up to 153.6 TB raw capacity with NVMe SSDs
-
EDSFF E3.S Gen5 NVMe configurations available (14- and 16-slot) — next-gen form factor for ultra-dense NVMe deployments with PCIe Gen5 bandwidth
-
Rear bay option — up to 2 × 2.5" SAS4/SATA (HDD/SSD) up to 30.72 TB, or up to 2 × EDSFF E3.S Gen5 NVMe rear modules
-
Supported drive types: SAS 24 Gb SSD (up to 7.68 TB), NVMe Gen4 SSD (up to 15.63 TB), SATA 6 Gb SSD (up to 3.84 TB), SAS 12 Gb 10K HDD (up to 2.4 TB)
-
Hot-swap drive bays on all configurations — zero downtime drive replacement during live operation
PERC H965i to S160 — Full Tier RAID Controller Portfolio
-
PERC H965i — Premium Performance (Avenger): 8 GB DDR4 NV cache, dual A15 1.2 GHz CPUs, x8 PCIe 3.0, 12 Gb SAS — maximum IOPS and SSD performance for demanding workloads
-
PERC H755 / H755N — Value Performance: proven 16G RAID controller covering the majority of production RAID 0/1/5/6/10/50/60 deployments
-
PERC H355 / HBA355i — Value: entry-level RAID and HBA for cost-optimized deployments without sacrificing data protection
-
PERC H965e (External): connects R660 to external JBOD arrays (12 Gb MD-series) for scalable capacity expansion beyond internal bays
-
HBA355e (External HBA): direct-attach to JBOD over SAS without RAID overhead for environments using software-defined storage
-
S160 Software RAID: OS-based RAID (RAID 0/1/5/10) using existing CPU cycles — ideal for hypervisor deployments with storage handled at software layer
-
fPERC (front PERC) form factor — does not consume a PCIe expansion slot, freeing all 3 riser slots for networking and accelerator cards
BOSS-N1 — Dedicated M.2 NVMe OS Boot Module
-
BOSS-N1 (Boot Optimized Server Storage) — dedicated M.2 NVMe module for OS boot, separated entirely from data storage drives
-
Does not consume a PCIe expansion slot — connects via a small form factor high-density connector directly to the system planar
-
Isolates OS I/O from workload I/O — eliminates contention between the operating system and production storage drives at the backplane level
-
BOSS-N1 replaces the 15G BOSS-S1 and BOSS-S2 SATA-based modules, delivering NVMe-speed OS boot performance without legacy SATA overhead
-
Supports mirrored M.2 SSD pairs for OS redundancy — if one boot drive fails, the system continues running on the second without manual intervention
-
Managed via iDRAC9 — drive health, warranty status, and firmware updates for the BOSS module are handled through the same interface as all other server components
-
Ideal for hypervisor deployments (VMware ESXi, Windows Server Hyper-V) where OS stability of the hypervisor host is critical to all hosted VMs
Up to 3 × 75 W Single-Wide GPUs with PCIe Gen5 Bandwidth
-
Supports up to 3 × 75 W single-wide (SW) GPUs in low-profile PCIe expansion slots — ideal for AI inference, remote visualization, and VDI acceleration
-
All GPU slots feed from the PCIe Gen5 fabric — bidirectional bandwidth of up to 64 GB/s per x16 slot ensures GPUs are never I/O starved
-
AI Inference — run multiple inference workloads simultaneously across up to 3 GPU cards for edge AI, NLP tokenization, and image classification pipelines
-
Remote Visualization & VDI — GPU-accelerated virtual desktop sessions with hardware rendering support for CAD, media, and 3D applications
-
Combined with Intel's built-in accelerators (IAA, QAT, DSA) for offloading data prep, compression, and cryptographic tasks from GPU compute cores
-
Card edge power delivery — 75 W per PCIe slot without auxiliary power connectors, simplifying cabling and cooling in 1U form factor
-
GPU configuration managed via iDRAC9 — health monitoring, thermal thresholds, and firmware updates without physical access
Up to 3 PCIe Gen5 Slots — Flexible Riser Configurations for Every Deployment
-
Up to 3 low-profile PCIe slots — all slots are on the PCIe Gen5 fabric for maximum bandwidth to expansion cards
-
Slot 1 & Slot 2: PCIe Gen5 x16 FHFL — full-height/full-length capable slots for high-bandwidth adapters (networking, storage HBAs)
-
Slot 3: PCIe Gen5 x8/x16 LP — low-profile slot for additional adapters; x16 physical, x8 or x16 electrical depending on riser config
-
Multiple riser configurations: Riser 2A, 2P, 2Q, 2R, 2S, 3P, 3Q, 3R, 3S, 1P, 4P — supports 1- or 2-processor configurations and front/rear PERC options
-
No-riser option — available in config 0 for direct-attach storage deployments without expansion card overhead
-
75 W card edge power delivery per slot — supports full-power GPUs and network cards without auxiliary cabling in 1U form factor
-
Front PERC configurations available — controllers install in a dedicated front PERC slot without consuming any PCIe riser slot
-
Rear storage option (Config 7 with Riser 3P) — adds 2 × 2.5" rear bays while maintaining 1 PCIe expansion slot
OCP 3.0 Mezzanine Up to 100 GbE — Flex LOM and Dedicated iDRAC Port
-
OCP 3.0 mezzanine slot — PCIe Gen4, supports up to x16 width, up to 100 GbE per card, up to 4 ports, maximum 75 W — doesn't consume a PCIe riser slot
-
Supported OCP 3.0 cards: 25 GbE SFP28 (2- or 4-port, Intel or Broadcom), 10 GbE Base-T (2- or 4-port, Intel or Broadcom), 1 GbE Base-T (4-port)
-
Optional 2 × 1 GbE LOM (Flex I/O board) with Broadcom BCM5720 controller — independent of OCP slot, installs as rear I/O module
-
Dedicated iDRAC9 Ethernet port on the rear — always-on out-of-band management traffic isolated from production NICs on a separate network interface
-
OCP vs. rNDC vs. previous LOM Mezz: OCP 3.0 adds PCIe Gen4 (vs Gen3), SFF form factor, NC-SI, SNAPI, and WoL support in a single mezzanine slot
-
Both LOM and OCP cards can be installed simultaneously — system allows LOM + OCP + dedicated iDRAC port all active at once
-
Network adapters rigorously validated by Dell — firmware integrated with iDRAC management for health monitoring, port statistics, and firmware updates
700 W to 1800 W Hot-Swap Titanium & Platinum PSUs — Redundant N+1
-
700 W Titanium — 200–240 V AC or 240 V DC, 2,625 BTU/hr — ideal for lower-TDP CPU configurations; new 60 mm PSU form factor introduced in Gen 16
-
800 W Platinum — 100–240 V AC or 240 V DC, 3,000 BTU/hr — broadest input voltage range for global deployments
-
1100 W Titanium — 100–240 V AC or 240 V DC, 4,100 BTU/hr — standard choice for dual-CPU configurations with mid-range TDP processors
-
1100 W LVDC — −48 to −60 V DC, 4,625 BTU/hr — designed for telco data centers and DC-powered colocation facilities
-
1400 W Platinum — 100–240 V AC or 240 V DC, 5,250 BTU/hr — handles dual maximum-TDP processor configurations with full NVMe storage
-
1800 W Titanium — 200–240 V AC or 240 V DC, 6,610 BTU/hr — top-tier PSU for maximum configuration deployments with three GPUs and dual 350 W CPUs
-
All configurations support N+1 hot-swap redundancy — failed PSU replacement requires zero downtime and no system restart
-
Dell power monitoring accuracy: 1% — 5× better than the industry-standard 5%, enabling precise power capping and capacity planning
Air Cooling and Optional Direct Liquid Cooling — Comprehensive Thermal Control
-
Standard (STD) and Very High Performance (VHP) fan modules — up to 4 hot-plug dual-fan modules, allowing N+1 fan redundancy with zero downtime on single fan failure
-
Air cooling operating range: 10°C to 35°C ambient inlet (50°F to 95°F) — extended ambient temperature configurations available for A3/A4 ASHRAE-rated data centers
-
Optional Direct Liquid Cooling (DLC) — required for 350 W TDP processor configurations (e.g., 8470Q); connects to rack-level manifolds and a cooling distribution unit (CDU)
-
Closed-loop thermal fan speed control — feedback temperatures from CPU, DIMMs, chipset, ambient inlet, HDDs, and OCP dynamically regulate fan RPM in real time
-
Open-loop thermal control — uses system inventory and configuration data to pre-set appropriate fan speeds based on ambient temperature at startup
-
User-configurable thermal settings — adjustable via iDRAC9 / BIOS setup for workload-specific noise/performance tradeoffs (acoustic Category 4 & 5 rated)
-
Dell EIPT (Enterprise Infrastructure Planning Tool) — available at dell.com/calc to model power and thermal requirements for specific R660 configurations before deployment
-
ENERGY STAR compliant, 80 PLUS certified, and Climate Savers approved — proven enterprise power efficiency standards across all PSU configurations
Front, Rear, and Internal Ports — Full Connectivity for Management and Operation
-
Front ports: 1 × Dedicated iDRAC Direct micro-USB (zero-cable management access), 1 × USB 2.0 Type-A, 1 × VGA — all accessible without rear rack access
-
Rear ports: 1 × USB 2.0, 1 × USB 3.0, 1 × Dedicated iDRAC9 Ethernet (RJ45), 1 × VGA (optional — omitted in liquid cooling rear I/O configuration), 1 × Serial port (optional)
-
RJ45 networking ports from optional LOM card — 2 × 1 GbE RJ45 on rear I/O module when Flex LOM board is installed
-
Internal port: 1 × USB 3.0 (optional) — connects internal USB devices such as boot media or hardware dongles without occupying external USB ports
-
Dedicated iDRAC Ethernet port is always isolated from production network traffic — out-of-band management over a separate physical interface for maximum security
-
iDRAC Direct (front micro-USB) — connect a laptop directly to the front of the server for initial configuration, diagnostics, and firmware updates without network connectivity
-
Quick Resource Locator (QRL) label on chassis — scan with a smartphone to access service manuals, setup guides, and iDRAC credentials specific to this service tag
Cyber Resilient Architecture — Silicon Root of Trust to Data-at-Rest Encryption
-
Silicon Root of Trust — cryptographic chain of trust originating in hardware; iDRAC refuses to boot if any firmware has been tampered with since last known-good state
-
Cryptographically signed firmware — every firmware image (BIOS, iDRAC, PERC, NIC) is digitally signed; unsigned or modified firmware is rejected at load time
-
Secure Boot — UEFI Secure Boot prevents unauthorized or modified bootloaders and OS kernels from loading; keys managed through iDRAC or BIOS setup
-
Secure Erase — NIST 800-88-compliant cryptographic erasure of all storage media (SAS, SATA, NVMe, BOSS, PERC cache) before drive removal or system decommission
-
Secured Component Verification (SCV) — hardware integrity check that validates all installed components against the factory manifest; detects unauthorized substitutions
-
System Lockdown — prevents firmware, configuration, and hardware changes while the system is under lockdown; requires iDRAC9 Enterprise or Datacenter license
-
Data at Rest Encryption — Self-Encrypting Drives (SEDs) with local key management (LKM) or external key management (EKMS) for full-volume encryption at the drive level
-
TPM 2.0 FIPS 140-2 & CC-TCG certified — hardware-based key storage and attestation; also available as TPM 2.0 China NationZ for regional compliance requirements
iDRAC9, OpenManage, and Redfish API — Unified Lifecycle Management
-
iDRAC9 (Integrated Dell Remote Access Controller 9) — out-of-band management controller embedded in every R660; operates independently of the host OS at all times
-
iDRAC RESTful API with Redfish — industry-standard REST API for programmable, vendor-agnostic server management; integrates with Ansible, Terraform, Python, and automation frameworks
-
iDRAC Direct (front micro-USB) — physical management port for laptop-based initial setup, diagnostics, and firmware updates without a network connection
-
Quick Sync 2 wireless module — Bluetooth-based mobile management; configure and monitor the server from a smartphone using the OpenManage Mobile app
-
iDRAC Service Module — lightweight host OS agent that passes in-band health telemetry (CPU/memory utilization, OS events) to iDRAC for unified monitoring
-
OpenManage Enterprise — centralized console for managing fleets of Dell servers; plus plug-ins for Power Manager, SupportAssist, and Update Manager
-
Third-party integrations: BMC TrueSight, Microsoft SCCM, Red Hat Ansible Modules, VMware vCenter & vRealize Operations Manager
-
Bezel options: optional LCD bezel for local health/status display, or security bezel to restrict front-panel physical access
Ubuntu, Windows Server, RHEL, SLES, and VMware ESXi — Dell-Validated Compatibility
-
Canonical Ubuntu Server LTS — Dell-validated Ubuntu LTS releases; preferred OS for cloud-native, Kubernetes, and open-source workloads on Xeon Scalable platforms
-
Windows Server with Hyper-V — full Windows Server LTSC support with Hyper-V hypervisor; ideal for Microsoft-centric enterprise deployments and VDI
-
Red Hat Enterprise Linux (RHEL) — enterprise Linux for production workloads; certified by both Red Hat and Dell for full hardware compatibility
-
SUSE Linux Enterprise Server (SLES) — SAP-certified Linux platform; SLES on Dell PowerEdge is a validated stack for SAP HANA and SAP S/4HANA
-
VMware ESXi — VMware HCL certified; R660 with ESXi supports vSphere, vSAN, vCenter, and NSX for full software-defined data center (SDDC) deployments
-
Full interoperability details at Dell.com/OSsupport — includes driver versions, firmware requirements, and validated configuration guides for each OS
-
iDRAC9 Lifecycle Controller — deploy OS and drivers directly from iDRAC without external boot media; reduces manual steps in large-scale bare-metal provisioning
R660 vs R650 — Key Improvements from Gen 15 to Gen 16
-
CPU platform upgrade — 3rd Gen Xeon Scalable (Ice Lake, Socket P14) → 4th Gen Xeon Scalable (Sapphire Rapids, Socket E); new socket, new PCIe Gen5 fabric, new DDR5 memory controller
-
Memory technology leap — DDR4 RDIMM/LRDIMM at 3200 MT/s → DDR5 RDIMM at 4800 MT/s (50% faster bus); 8 TB max achievable with RDIMM alone (no LRDIMM required)
-
PCIe bandwidth doubled — Gen 15 supported 3 × PCIe Gen4 slots → Gen 16 adds PCIe Gen5 slots (Slots 1–2: Gen5 x16) alongside Gen4 — full 32 GT/s × lane throughput
-
PERC controller tier added — H965i (Avenger) Premium Performance tier is new in Gen 16; Gen 15's top controller was H755 — H965i adds 8 GB NV cache and dual-CPU RAID architecture
-
BOSS upgrade — BOSS-S1/S2 (SATA M.2) → BOSS-N1 (NVMe M.2); OS boot performance dramatically improved with NVMe latency and throughput
-
PSU maximum increased — 15G max was 1400 W Platinum → 16G adds 1800 W Titanium; supports higher-TDP DLC processor configurations
-
Software RAID upgraded — S150 → S160; expanded RAID level and drive type support in the software controller
| Feature | R660 (Gen 16) | R650 (Gen 15) |
|---|---|---|
| CPU Generation | 4th / 5th Gen Xeon Scalable | 3rd Gen Xeon Scalable |
| Max Cores | 64 cores (5th Gen) | 40 cores |
| Memory Type | DDR5 RDIMM | DDR4 RDIMM / LRDIMM |
| Memory Speed | 4800 MT/s (4th Gen) / 5600 MT/s (5th Gen) | 3200 MT/s |
| Max Memory | 8 TB (RDIMM only) | 8 TB (LRDIMM) / 2 TB (RDIMM) |
| PCIe Generation | Gen5 + Gen4 | Gen4 only |
| Top PERC Controller | H965i (8 GB NV cache) | H755 |
| Boot Module | BOSS-N1 (NVMe M.2) | BOSS-S1 / BOSS-S2 (SATA M.2) |
| Max PSU | 1800 W Titanium | 1400 W Platinum |
ProSupport, Deployment Services, and Power Efficiency — Complete Lifecycle Support
-
ProSupport Plus — predictive issue detection via SupportAssist AI; proactive alerts before failures occur, with 24/7 expert access and on-site dispatch
-
ProSupport — 24/7 telephone and remote access support with direct access to Dell engineering for complex hardware and firmware issues
-
Dell Residency Services — dedicated on-site Dell engineers embedded within your team for complex platform deployments, migrations, and optimization projects
-
Custom Deployment Services — factory-integrated BIOS, firmware, OS, and application configurations so servers arrive ready to deploy; reduces on-site setup time to minutes
-
Data Migration Services — assisted migration from legacy infrastructure; Dell engineers plan, execute, and validate data movement to minimize downtime and risk
-
Dell TechDirect — self-service warranty management, remote diagnostics, and parts dispatch portal for organizations managing their own hardware support
-
Power efficiency: ENERGY STAR certified, 80 PLUS Titanium/Platinum PSUs, ASHRAE A3/A4 extended temperature support — meets enterprise sustainability and data center efficiency standards
-
Dell Technologies Consulting — strategic IT assessments, workload modernization roadmaps, and cloud integration planning for long-term infrastructure optimization
Frequently Asked Questions — Dell PowerEdge R660
The Dell PowerEdge R660 supports up to 8 TB of DDR5 ECC memory across 32 DIMM slots (16 per socket) in a dual-socket 1U chassis — twice the capacity of the Gen 15 R650 it replaces, critical for in-memory analytics, large virtualisation pools, and SAP HANA.
The R660 supports up to 10 × 2.5-inch SFF, 24 × NVMe U.2, or 16 × EDSFF E3.S Gen 5 drives depending on backplane selection — one of the densest 1U storage footprints in the industry.
The R660 supports 4th Gen Intel Xeon Scalable processors (Sapphire Rapids, LGA4677) in a dual-socket configuration with up to 56 cores and 112 threads per CPU. PCIe Gen 5.0 slots double available bandwidth versus the R650’s PCIe Gen 4 platform.
Yes. Express Computer Systems stocks professionally reconditioned Dell PowerEdge R660 rack servers tested, cleaned, and configured to your exact workload. Shop refurbished Dell R660 servers at ECS.
The R660 (Gen 16) replaces the R650 (Gen 15) with 4th Gen Xeon Scalable (Sapphire Rapids) vs 3rd Gen (Ice Lake-SP), DDR5 with 8 TB capacity vs 4 TB DDR4, PCIe Gen 5.0 vs Gen 4.0, new EDSFF E3.S drive support, iDRAC9 with OpenManage Enterprise, and enhanced on-die accelerators (AMX, DSA, QAT) for AI, compression, and encryption.
Ready to Deploy a Refurbished Dell PowerEdge R660?
Express Computer Systems specializes in professionally reconditioned Dell PowerEdge servers — thoroughly tested, firmware-updated, and backed by our team of hardware experts. Get enterprise-grade Gen 16 performance at a fraction of the cost of new.
Start building your custom server today