Dell PowerEdge R650xs — Dual Xeon 1U NVMe Server | Refurbished Review
Cost-Efficient 1U Dual-Socket Gen 15 — Built for Virtualization, Kubernetes, and Mid-Scale Databases
Enterprise Virtualization Standardization — The R650xs provides a single cost-efficient platform for mid-scale enterprise virtualization; dual 3rd Gen Xeon Scalable sockets with up to 64 physical cores, 1 TB DDR4 RDIMM, and PCIe Gen 4 bus deliver a compelling price-to-thread ratio for organizations standardizing on a single 1U dual-socket server model across the rack without the premium of maximum memory density configurations
Kubernetes and Container Orchestration — R650xs makes an ideal Kubernetes worker node: high core count (up to 64 cores), fast DDR4 memory at 3200 MT/s, NVMe Gen 4 local storage for container ephemeral I/O, and OCP 3.0 25 GbE networking for low-latency pod-to-pod communication; air cooling simplifies deployment in standard rack rows without liquid cooling infrastructure
Relational Database Tiers (Mid-Scale) — MySQL, PostgreSQL, Microsoft SQL Server, and Oracle deployments that operate within a 1 TB RAM footprint benefit from the R650xs dual-socket platform: PCIe Gen 4 NVMe delivers the low-latency I/O required for OLTP workloads, and the 64-core ceiling prevents CPU bottlenecks during peak transaction bursts — right-sized for departmental and regional database instances
High-Performance Technical Computing — CPU-bound workloads in simulation, CFD, genomics, financial risk modeling, and Monte Carlo analysis scale linearly across both Xeon Scalable sockets; OCP 3.0 NIC connectivity (up to 100 GbE) links R650xs nodes into low-latency MPI and RDMA clusters; 1U density maximizes node count per rack in compute-dense HPC deployments where memory capacity is secondary to raw core count
CI/CD Infrastructure and DevOps Build Farms — Continuous integration pipelines require high parallelism and fast local storage; 64 physical cores handle large simultaneous build and test job queues, NVMe local scratch minimizes compile times, and OCP 3.0 connectivity keeps artifact repositories and source code servers accessible at line rate; air cooling and standard rack integration reduce operational complexity in busy build environments
Software-Defined Networking (SDN) and NFV — OCP 3.0 with up to 100 GbE capability makes the R650xs a purpose-fit platform for network functions virtualization: vRouter, vFirewall, SD-WAN, and packet processing workloads benefit from direct PCIe Gen 4 NIC attachment with SR-IOV; NVMe local storage provides fast logging and telemetry capture; 1U density packs high port density per rack row
Software-Defined Storage Nodes — HBA355i pass-through mode enables OS-managed ZFS, Ceph OSD, and vSAN data node configurations across up to 10 NVMe or SAS drives; OCP 3.0 25 GbE provides storage fabric bandwidth for distributed object and block replication; dual-socket increases throughput for erasure-coded Ceph deployments where CPU is the bottleneck for encoding operations
Remote Office / Branch Office (ROBO) and Edge Data Centers — 1U chassis at standard 42.8 mm height fits any 19-inch rack; air-only cooling eliminates the need for liquid cooling infrastructure at edge sites; iDRAC9 with Quick Sync 2 optional BLE module enables smartphone-based field management without remote KVM gear; ReadyRails tool-less installation reduces site visit duration for IT generalist staff
Virtualization
Kubernetes
Databases
HPC
CI/CD
NFV / SDN
SDS
ROBO / Edge
3rd Gen Intel Xeon Scalable — Up to 2 × 32 Cores at 270 W with PCIe Gen 4 Native
Platform Architecture — Dual LGA4189 sockets on Intel C621A chipset; supports single or dual 3rd Gen Intel Xeon Scalable processors; Intel Ultra Path Interconnect provides up to 3 × UPI links per CPU at 11.2 GT/s (Gold 6 and Platinum) or 10.4 GT/s (Silver and Gold 4) for inter-socket NUMA bandwidth; 64 PCIe Gen 4 lanes per socket at 16 GT/s deliver double the per-lane bandwidth of Gen 15's predecessor platform
Top CPU — Xeon Platinum 8358 (32 Cores) — 32 cores / 64 threads, 2.6 GHz base, 48 MB L3 cache, 11.2 GT/s UPI, 250 W TDP, 3200 MT/s DDR4; the R650xs highest supported core count option — in dual-socket configuration delivers 64 cores / 128 threads from a single 1U air-cooled chassis; suited for demanding HPC, virtualization density, and large database workloads within the R650xs air-cooling thermal envelope
Xeon Gold 6338 (32 Cores) — 32 cores / 64 threads, 2.0 GHz base, 48 MB cache, 11.2 GT/s UPI, 205 W TDP, 3200 MT/s; lower TDP than the Platinum 8358 at identical core count — an optimal refurbished upgrade path for organizations maximizing core density on a power budget; dual 6338 configuration reaches 64 cores with lower electricity cost than dual Platinum
Xeon Gold 6 Series (Mid-Tier) — Gold 6 tier spans 10–32 cores; popular options include 6314 (32c, 2.4 GHz, 205W), 6342 (24c, 2.8 GHz, 230W), 6326 (16c, 2.9 GHz, 185W), 6312U (24c, single-socket only); all Gold 6 support 11.2 GT/s UPI and 3200 MT/s DDR4 — the mid-tier sweet spot for balanced virtualization hosts that need strong per-core performance and high memory bandwidth simultaneously
Xeon Gold 4 Series (Entry Dual Socket) — 4316 (20c, 2.3 GHz, 150W), 4314 (16c, 135W), 4310 (12c, 120W), 4309Y (8c, 105W); 10.4 GT/s UPI; 2666 MT/s DDR4 maximum; Gold 4-series is the most cost-efficient dual-socket option on the R650xs — right-sized for mid-scale file and application servers, light infrastructure workloads, and branch office deployments where dual-socket redundancy is more important than maximum throughput
Xeon Silver Series (Mid-Range) — 5318Y (24c, 2.1 GHz, 165W), 5320 (26c, 2.2 GHz, 185W), 5317 (12c, 3.0 GHz, 150W), 5315Y (8c, 3.2 GHz, 140W); 10.4 GT/s UPI; 2933 MT/s DDR4 maximum; Silver processors maximize base clock frequency relative to their core count — selected for latency-sensitive applications and middleware tiers where single-threaded throughput matters more than peak parallel compute
Air Cooling and TDP Range — R650xs supports processor TDPs from 105 W (Gold 4309Y) to 270 W with appropriate fan configuration; three fan tiers — Standard (STD), High Performance Silver (HPR Silver), and High Performance Gold (HPR Gold) — are selected automatically by iDRAC based on installed CPU TDP and chassis configuration; no Direct Liquid Cooling (DLC) option exists for the R650xs chassis; all processor SKUs operate under air cooling
Single-Socket Configuration — R650xs ships and operates with a single processor installed; CPU2 slot is left open; iDRAC and Lifecycle Controller automatically adjust memory population rules, PCIe slot availability, and fan profile for single-socket operation; adding a second matched processor later enables full dual-socket performance without chassis replacement — a practical phased deployment option for growing workloads
16-Slot DDR4 RDIMM — Up to 1 TB at 3200 MT/s — 8 Channels per CPU
16 DIMM Slots — 8 Channels per CPU (1 DPC) — 16 total DDR4 RDIMM slots distributed as 8 per processor across 8 memory channels at 1 DIMM per channel; at 1 DPC (one DIMM per channel) DDR4 RDIMM operates at its maximum rated speed of 3200 MT/s without bandwidth penalty — the R650xs single-DIMM-per-channel architecture ensures each memory channel runs at full speed regardless of total installed capacity
RDIMM Up to 1 TB — DDR4 Only — Registered ECC DDR4 DIMMs only; available sizes 8 GB (1Rx8), 16 GB (2Rx8), 32 GB (2Rx8), and 64 GB (2Rx4); maximum capacity 1 TB (16 × 64 GB RDIMMs) at 3200 MT/s; the R650xs is RDIMM-exclusive — no LRDIMM or UDIMM support; no Intel Optane Persistent Memory 200 Series support; RDIMM-only architecture keeps platform cost lower while delivering full DDR4 bandwidth at supported capacities
Memory Speed by CPU Tier — DDR4 maximum operating speed: 3200 MT/s (Gold 6 and Platinum at 1 DPC); 2933 MT/s (Silver series); 2666 MT/s (Gold 4 series); memory speed scales with CPU tier — when configuring an R650xs for memory-bandwidth-intensive workloads such as HPC or analytics, select Gold 6 or Platinum CPUs to achieve 3200 MT/s on all populated channels
Typical Configuration Scales — 8-slot populated (4 per CPU): 128 GB (8 × 16 GB) / 256 GB (8 × 32 GB) — mid-tier virtualization; 12-slot: 384 GB (12 × 32 GB) / 768 GB (12 × 64 GB) — database and analytics; 16-slot fully populated: 512 GB (16 × 32 GB) / 1 TB (16 × 64 GB) — maximum capacity; even 4-DIMM single-socket entry configurations of 64 GB scale up to full 1 TB dual-socket by adding matched DIMMs without chassis changes
ECC and RAS Features — Registered DIMMs provide hardware error correction at the DIMM register buffer; supports Single Device Data Correction (SDDC), Demand and Patrol Scrubbing, Memory Rank Sparing, and Multi-Rank ECC; iDRAC9 reports correctable and uncorrectable memory errors in real-time event logs — enabling predictive DIMM replacement before uncorrectable errors cause unplanned downtime on production workloads
R650xs vs R650 Full — Memory Trade-Off — The R650 (full) supports 32 RDIMM/LRDIMM slots and 4 TB maximum; R650xs is intentionally constrained to 16 RDIMM slots and 1 TB — the XS designation reflects a cost-optimized platform design that eliminates LRDIMM expense, Intel Optane PMem licensing, and the associated planar complexity; organizations that require 1 TB or less of DDR4 and do not need GPU support receive full PCIe Gen 4 and OCP 3.0 Gen 15 platform benefits at lower platform cost
Upgrade from R640 Memory Comparison — R640 provided up to 24 DIMM slots at 2933 MT/s (Gold 6/Platinum) and supported LRDIMM up to 3 TB; R650xs reduces slot count (16 vs 24) but raises DDR4 speed to 3200 MT/s; organizations migrating R640 workloads to R650xs gain per-channel bandwidth improvement (+267 MT/s per channel) and Gen 4 platform benefits even as total DIMM slot count decreases — workloads within the 1 TB RDIMM ceiling see direct performance improvement
Up to 10+2 NVMe or 4 × 3.5-inch LFF — 92.1 TB Max in 1U
10 × 2.5-inch Front — Up to 76.8 TB — Maximum density front configuration with 10 × SFF hot-plug bays for SAS/SATA HDD or SSD or NVMe Gen 4 SSDs; max raw capacity 76.8 TB using 7.68 TB NVMe SSDs; universal backplane supports mixed SAS and NVMe within the same R650xs chassis — combine NVMe for hot-tier IOPS with SAS HDDs for capacity tier without deploying a separate JBOD shelf
10+2 Configuration — Up to 12 Drives Total — 10 front 2.5-inch bays plus 2 optional rear 2.5-inch SAS/SATA/NVMe hot-plug bays via rear-drive cage; rear bays add up to 15.36 TB (2 × 7.68 TB NVMe); rear drives are ideal for a dedicated OS mirror, journal/log tier, or cache device without consuming any front data bay slots — 12 active storage paths from a single 1U server
8 × 2.5-inch SAS/SATA/NVMe — Eight SFF hot-plug front bays; max 61.44 TB; supports SAS 12 Gb/s, SATA 6 Gb/s, and NVMe Gen 4 on universal backplane; reduced-bay chassis for workloads that prioritize lower upfront cost or have modest storage requirements — PCIe Gen 4 NVMe performance is identical in the 8-bay versus 10-bay configuration on a per-drive basis
4 × 3.5-inch LFF SAS/SATA — Four Large Form Factor hot-plug front bays; max 64 TB (4 × 16 TB SAS/SATA HDDs) or up to 7.68 TB per-drive SSD; optimized for capacity-first workloads: archival repositories, backup landing zones, sequential video surveillance storage, and file servers where per-drive capacity outweighs IOPS; lower drive count simplifies RAID parity rebuild times on large HDDs
NVMe Gen 4 Drive Options — 2.5-inch U.2 NVMe Gen 4 SSDs up to 7.68 TB per slot; Gen 4 direct-attach backplane provides the full 16 GT/s per-slot PCIe bandwidth — each NVMe U.2 SSD operates at its rated interface speed without sharing PCIe lanes through a SAS expander; NVMe random read IOPS in the millions per slot enable in-chassis all-flash performance approaching dedicated AFA arrays
SAS and SATA Drive Options — 2.5-inch SAS 12 Gb/s: 10K HDDs (600 GB–2.4 TB), SAS SSDs (400 GB–15.36 TB); 2.5-inch SATA 6 Gb/s SSDs (240 GB–3.84 TB); 3.5-inch SATA 6 Gb/s 7.2K (2–18 TB); 3.5-inch SAS 12 Gb/s 7.2K (2–18 TB); M.2 SATA for BOSS-S1 boot; microSD 16/32/64 GB for IDSDM boot — full spectrum of drive types accommodated by the R650xs backplane and RAID stack
External JBOD Expansion — HBA355e external SAS 12 Gb/s rear-panel port enables connection to Dell MD14xx PowerVault JBODs and ME4 storage arrays; PERC H840 external RAID controller expands to additional JBOD shelves for capacity growth beyond internal bays; dual-path SAS fabric from HBA355e provides fail-safe path redundancy to attached JBODs for production SAN-attached R650xs deployments
PERC H755N NVMe RAID + H355 Front PERC — Full 15G RAID Stack in 1U
PERC H755 (Premium RAID — SAS/SATA) — 12 Gb/s SAS + SATA controller on PCIe Gen 4 with NV Flash-Backed write cache; RAID 0, 1, 5, 6, 10, 50, 60; NV cache preserves write data through unexpected power events without battery or supercapacitor maintenance cycles; highest-endurance RAID tier for OLTP databases, ERP transaction journals, and write-heavy RAID 6 arrays on the R650xs platform
PERC H755N (NVMe RAID) — NVMe-native RAID controller on PCIe Gen 4 with NV write cache; RAID 0, 1, 5, 6, 10, 50, 60 across NVMe Gen 4 U.2 SSDs; the H755N enables hardware-level data protection on all-NVMe configurations without operating NVMe drives in unprotected JBOD mode under OS software RAID — critical for production databases and latency-sensitive analytics arrays where NVMe RAID rebuild speed and data integrity are non-negotiable
PERC H745 (Value Performance RAID) — 12 Gb/s SAS + SATA on PCIe Gen 4 with write-back cache; RAID 0, 1, 5, 6, 10, 50, 60; mid-tier between H755 and H345 for virtualization hosts, moderate-intensity databases, and mixed workloads requiring RAID parity and write caching at lower cost than the H755 premium; broadly compatible with all SAS/SATA drive types supported on the R650xs
PERC H355 Front PERC — Zero PCIe Slot Consumption — 12 Gb/s SAS fPERC installs in a dedicated small-form-factor planar slot (not a user PCIe slot); RAID 0, 1, 10; all three user-accessible PCIe Gen 4 slots remain available for NICs, HBAs, or other add-in cards when H355 fPERC is the sole RAID controller; essential for R650xs configurations requiring RAID plus multiple full-height PCIe add-in cards simultaneously
PERC H345 (Entry Hardware RAID) — 12 Gb/s SAS + SATA entry controller on PCIe Gen 4; RAID 0, 1, 10; no write cache; appropriate for read-dominant workloads, hypervisor boot arrays, development and lab environments where sustained write-intensive RAID rebuild and caching overhead is not a workload requirement
HBA355i and HBA355e (Pass-Through) — HBA355i internal 12 Gb/s SAS pass-through HBA for OS-managed ZFS, Ceph OSD, vSAN, and software-defined storage configurations; HBA355e rear external 12 Gb/s SAS connects to JBODs and tape libraries; both present drives directly to the OS without a RAID layer — mandatory for software storage platforms that manage their own parity, mirroring, and fault tolerance
PERC S150 (Software RAID) — Firmware-managed SATA + NVMe software RAID; RAID 0, 1, 5, 10; no additional PCIe card required; lowest-cost RAID option for non-critical data tiers, single-drive OS installations, and development environments; RAID activity runs on the Xeon Scalable processor — kept away from production storage arrays under sustained application load
BOSS-S1 Mirrored M.2 Boot — OS Storage Dedicated and Isolated from All Data Bays
BOSS-S1 Module (Primary Boot Option) — Boot Optimized Storage Solution S1 installs in a dedicated rear internal riser slot that consumes no front data bays and no user PCIe expansion slots; houses dual M.2 SATA SSDs configured in hardware RAID 1 mirror; M.2 SSD failure is transparent — the mirror rebuilds from the surviving drive while the server continues operating; BOSS-S1 is the purpose-built OS boot solution for Gen 15 PowerEdge platforms keeping boot I/O completely separate from application storage paths
BOSS-S1 Drive Capacity — 240 GB M.2 SATA Read-Intensive SSDs; mirrored pair provides 240 GB usable (RAID 1) for OS image and boot partition; SATA 6 Gb/s interface; endurance class optimized for OS boot and read-heavy patterns; appropriate for VMware ESXi, Windows Server, and Linux OS installations including full hypervisor with management agents
Boot Isolation Architecture — BOSS-S1 M.2 SATA bandwidth runs on a fully dedicated path to the rear internal riser — completely isolated from the SAS/SATA/NVMe backplane, all PERC controllers, and all user PCIe slots; OS boot I/O and application storage I/O never share a controller, backplane, or PCIe lane; parallel boot and data I/O pipelines at all times eliminate the BOSS-to-data storage interference present on single-controller configurations
IDSDM (Internal Dual SD Module) — Alternative boot device: dual microSD in hardware RAID 1 mirror; available in 16 GB, 32 GB, and 64 GB; supported for VMware ESXi and Citrix XenServer hypervisor boot deployments; microSD write endurance is optimized for read-heavy OS boot patterns; IDSDM shares the same dedicated internal slot family as BOSS-S1 — one or the other may be installed, not both simultaneously
Internal USB 3.0 (Optional) — Optional internal USB 3.0 dongle (40 × 16 × 8 mm) installs cleanly inside the chassis; provides persistent in-chassis USB storage for hardware security keys, supplemental configuration media, and deployment tooling; not externally accessible and does not occupy any front-panel USB port; useful for UEFI Secure Boot key provisioning workflows managed inside the chassis
Boot Device Selection in iDRAC and Lifecycle Controller — Boot device order is configurable via iDRAC9 web interface, RACADM CLI, Redfish API, and Lifecycle Controller BIOS setup; BOSS-S1 and IDSDM appear as standard SATA storage devices in the boot sequence; Lifecycle Controller OS deployment guided workflow integrates BOSS-S1 and IDSDM as valid first-stage boot targets with OS partition layout options — no manual driver injection required for certified OS installations
Up to 3 × PCIe Gen 4 LP Slots — Plus OCP 3.0 on a Dedicated Gen 4 × 8 Lane
Up to 3 × Low-Profile PCIe Gen 4 Slots — Maximum riser configuration provides three PCIe Gen 4 × 16 user slots (low-profile / half-height); PCIe Gen 4 at 16 GT/s per lane delivers 32 GB/s bidirectional bandwidth per slot — double the 16 GB/s available on the R640's Gen 3 platform; supports simultaneous 25 GbE NIC, 32G Fibre Channel HBA, and NVMe PCIe add-in card installations without bandwidth contention
Riser Configurations — R1+R2a: 3 × LP slots (1 × x16 CPU1 + 1 × x8 CPU1 + 1 × x16 CPU2); R1+R2b SNAPI: 2 × LP slots with dedicated SNAP I/O path for socket-direct OCP; R1+R2c: 2 × LP slots (alternate CPU routing); R1 only (single socket): 1 × LP x16; riser configuration determines PCIe topology between CPU1, CPU2, and add-in cards — dual-socket R1+R2a provides maximum 3-slot expansion with CPU-affine PCIe routing for NUMA-sensitive workloads
OCP 3.0 Slot — Dedicated Gen 4 × 8 Lanes — One integrated OCP 3.0 mezzanine slot on PCIe Gen 4 × 8 independent of the three user PCIe slots; supports Intel, Broadcom/Emulex, Mellanox, QLogic, and SolarFlare SFF OCP 3.0 NIC cards at 1 GbE to 100 GbE; OCP 3.0 bandwidth and slot count are additive — all three user PCIe slots remain fully available for HBAs, add-in NICs, FPGAs, and NVMe expansion cards simultaneously
SNAP I/O Socket-Direct Technology — SNAP I/O (R1+R2b riser) allows CPU0 and CPU1 to each directly access a dedicated OCP NIC port without crossing the UPI inter-socket link; on standard NIC configurations traffic from the NUMA-remote CPU traverses UPI adding latency; SNAP I/O routes each CPU's I/O to its corresponding NIC port natively — reducing inter-socket network latency and freeing UPI bandwidth for cache coherency traffic
Fibre Channel and InfiniBand Expansion — PCIe Gen 4 LP slots accommodate 16G/32G Fibre Channel HBAs (QLogic QLE, Emulex LPe series) for SAN fabric attachment; low-profile InfiniBand HDR25 (50 Gb/s) NICs expand high-speed interconnect capability beyond OCP 3.0; all LP form-factor cards from major vendors fit the R650xs riser without bracket modification in a standard 4-post rack deployment
NVMe Add-In PCIe Expansion — PCIe Gen 4 x16 slots accept LP NVMe add-in cards (AIC) for storage capacity beyond the front backplane; U.2 PCIe Gen 4 bridge cards in LP form factor route between the PCIe slot and additional U.2 NVMe drives via rear drive cage — a storage expansion option for configurations where all 10 front bays are occupied and additional NVMe capacity is needed beyond the 2 rear bay maximum
No GPU Support — The R650xs chassis does not support GPU add-in cards; the air-cooling-only thermal design and 1U chassis depth are not validated for GPU thermal loads or double-wide GPU form factors; organizations requiring GPU acceleration (NVIDIA T-series, AMD Instinct, or similar) should evaluate the Dell PowerEdge R650 (full) which supports up to 3 × 75 W single-width GPUs, or the R750xa/R750 for multi-GPU deployments
Dual 1 GbE LOM + OCP 3.0 Up to 100 GbE — Dedicated iDRAC Management Port
Embedded Dual-Port 1 GbE LOM — Broadcom BCM5720 dual-port 1 GbE LAN on Motherboard on the rear panel; supports Wake-on-LAN, PXE network boot, and iDRAC Shared LOM mode; iDRAC Shared LOM routes out-of-band management traffic over one LOM port when a dedicated iDRAC network is not available — a useful option for branch and edge sites without a dedicated management VLAN infrastructure
Dedicated iDRAC9 Management Port — One dedicated RJ-45 iDRAC9 GbE management NIC on the rear panel; operates independently of both LOM data ports; physically distinct management path eliminates the risk that a production network incident or host OS failure blocks iDRAC remote access — required for iDRAC9 Enterprise and Datacenter license full out-of-band functionality including virtual console and virtual media
OCP 3.0 Slot — 1 GbE to 100 GbE — Integrated OCP 3.0 NIC slot on 8 × PCIe Gen 4 lanes beyond user PCIe slots; multi-vendor NIC ecosystem spans 1 GbE (Intel 4-port BT), 10 GbE (Intel/Broadcom SFP+/BT), 25 GbE (Mellanox/Broadcom SFP28), 100 GbE (Mellanox ConnectX single-port QSFP28); OCP 3.0 standardization means a NIC card validated for one Gen 15 server is compatible across the entire Gen 15 PowerEdge portfolio
10 GbE Options via OCP 3.0 — Intel (SFP+ 2-port, BT 2-port, BT 4-port), Broadcom/Emulex (SFP+ 2-port, BT 2-port, BT 4-port), QLogic/Marvell (SFP+ 2-port); 10GBASE-T avoids optical transceivers with Cat6a/Cat7 copper runs up to 100 m; SFP+ enables direct ToR switch connectivity via fiber LC or Twinax DAC cables; both interfaces are served from the same OCP 3.0 slot with zero additional PCIe card cost
25 GbE Options via OCP 3.0 — Broadcom/Emulex Ethernet OCP (SFP28 2-port, SFP28 4-port), Intel E810 (SFP28 2-port), Mellanox ConnectX-5 (SFP28 2-port), SolarFlare (SFP28 2-port); 25 GbE delivers 2.5× the per-port bandwidth of 10 GbE in the same slot footprint — the correct tier for VM cluster VM-motion fabric, iSCSI/NFS storage traffic, and Kubernetes pod-to-pod overlay networks requiring consistent low-latency throughput
100 GbE Option via OCP 3.0 — Mellanox ConnectX-6 single-port QSFP28 100 GbE OCP 3.0 card provides 100 Gb/s connectivity from the OCP slot without consuming any user PCIe slot — enabling 100 GbE network function virtualization, RDMA over Converged Ethernet (RoCE), and NVMe over Fabrics (NVMe-oF) target connectivity at the maximum OCP gen 3.0 throughput supported by the PCIe Gen 4 × 8 host interface
PCIe Add-in NICs via Expansion Slots — With R1+R2a 3-slot riser, additional LP NIC cards expand beyond OCP bandwidth; popular options include Mellanox ConnectX-5/6 25/100 GbE LP and Intel E810 25 GbE LP; supports iSCSI, NFS, NVMe-oF over Ethernet (RoCE v2), and RDMA fabrics; 16G/32G Fibre Channel HBAs simultaneously coexist with OCP NIC in a 3-slot configuration — serving both IP storage and FC SAN traffic from a single R650xs chassis
600 W Platinum to 1800 W Titanium — Hot-Swap 1+1 Redundant Across Seven PSU Options
600 W Mixed Mode 80 PLUS Platinum — Entry PSU tier for cost–optimized single-socket configurations or dual-socket deployments with low-TDP CPUs and minimal drive count; Mixed Mode auto-sensing AC (100–240 V) plus 240 VDC; Platinum efficiency tier; suitable for Silver or Gold 4 processor single-socket configurations where total system TDP remains well under 600 W at peak
700 W Flex Slot 80 PLUS Titanium — Titanium efficiency at 700 W; superior energy efficiency per watt compared to same-wattage Platinum units; recommended for mid-tier dual-socket configurations (Silver or Gold 4 CPUs, 8–12 drives, standard OCP NIC) where 700 W headroom comfortably covers peak draw and Titanium efficiency premium offsets PSU cost over a 3+ year deployment lifecycle
800 W Mixed Mode 80 PLUS Platinum — 800 W at high-line (200–240 V) with Mixed Mode 240 VDC support; Platinum efficiency; covers dual-socket Gold 6 configurations with standard drive populations and OCP 25 GbE; the mid-range PSU option for R650xs deployments that require more headroom than 700 W Titanium but do not need the full 1100 W+ tier for sustained dual-socket peak loads
1100 W Mixed Mode 80 PLUS Titanium — 1100 W at high-line; Titanium efficiency rated at 80% (10%), 94% (20%), 96% (50%), 91.5% (100%); 96% efficiency at 50% load makes this the most efficient PSU option for dual-socket R650xs deployments at 50–70% utilization; reduces electricity cost measurably over multi-year deployments versus same-wattage Platinum units — the preferred PSU for always-on dual-socket configurations
1100 W DC Hot-Swap (−48 V to −60 V) — Dedicated −48 V DC PSU for carrier, telco, and NEBS-adjacent data center power plants; 1100 W rated at nominal DC rail; same hot-swap PSU form factor as AC units — both PSU bays accept mixed AC and DC units for power-feed diversity in hybrid AC/DC facilities; enables R650xs standardization in telco central office and edge data center deployments with DC power distribution
1400 W Mixed Mode 80 PLUS Platinum — 1400 W at high-line; Platinum efficiency; provides headroom above the 1100 W Titanium for dual-socket high-TDP configurations (2 × 250 W Platinum CPUs + 10 NVMe drives + OCP 100 GbE + 3 PCIe LP add-in cards) operating simultaneously at peak load; the correct PSU tier when total system TDP calculation approaches or exceeds the 1100 W PSU ceiling
1800 W Mixed Mode 80 PLUS Titanium — Top-tier PSU for maximum dual-socket TDP configurations with full drive complement; Titanium efficiency at 1800 W; addresses worst-case power draw under simultaneous ful CPU TDP, all-NVMe drive spin-up, and maximum OCP + PCIe add-in card power draw — provides substantial headroom margin for configurations specifying two 270 W Platinum CPUs and maximum storage density
1+1 Hot-Swap Redundant Design — Both PSU bays are active simultaneously in load-sharing mode; single PSU failure triggers immediate full load transfer to the surviving unit with zero service interruption; hot-swap replacement while the server operates; mismatched wattage PSUs trigger an iDRAC/LCD mismatch alert — always replace with matched wattage PSUs to maintain BIOS-validated operating conditions
Air-Cooled Only — Up to 7 Hot-Plug Fans with ASHRAE A2/A3/A4 Certification
Air Cooling Only — No DLC Option — The R650xs is a pure air-cooled platform; no Direct Liquid Cooling (DLC) option is available or required; the 1U chassis with up to 7 hot-plug dual-fan modules provides sufficient thermal capacity for all supported processor SKUs including up to 270 W TDP configurations; air-only cooling simplifies deployment in any standard data center row without liquid coolant plumbing, leak detection, and supply/return manifold infrastructure
Up to 7 Hot-Plug Fan Modules — Up to seven hot-plug dual-fan modules in the R650xs chassis; N+1 fan redundancy allows continuous operation when one fan module fails; iDRAC immediately instructs remaining modules to ramp speed to maintain all component temperatures within specification; hot-plug replacement returns the chassis to full redundancy without system downtime or service window
Three Fan Profiles — STD, HPR Silver, HPR Gold — Standard (STD) fan profile for lower-TDP configurations; High Performance Silver (HPR-S) for mid-TDP Gold/Silver processors; High Performance Gold (HPR-G) for high-TDP Gold 6 and Platinum configurations; iDRAC automatically selects and enforces the correct fan profile at POST based on installed CPU TDP, chassis configuration, and ambient inlet temperature — no manual fan speed configuration required
Dell Multi Vector Cooling — Real-time closed-loop fan control using inlet temperature sensors, individual component thermal telemetry (processors, DIMMs, storage controllers, OCP NIC, HDDs/SSDs), and workload power draw; fan speed is driven by the most demanding component in the system at any moment and adjusted at sub-second intervals; MVC minimizes fan speed — reducing acoustical output and fan wear — while maintaining all components within acceptable operating temperatures
ASHRAE A2 — Standard Operating Range — Standard operating temperature: 10–35°C (50–95°F) up to 950 m altitude with standard derating above; all processor, drive, NIC, and controller options are validated for ASHRAE A2 at standard configuration; review Dell EMC PowerEdge R650xs Thermal Restriction tables before deploying high-TDP CPU configurations at elevated altitude or above 35°C ambient
ASHRAE A3 and A4 Extended Operation — Certain processor and drive configurations are validated for ASHRAE A3 (40°C / 104°F) and A4 (45°C / 113°F) extended ambient operation; A3/A4 certification enables R650xs deployment in high-density telco shelters, container data centers, and edge locations where ambient cooling cannot maintain A2 conditions; verify specific component combination compliance in the R650xs Technical Specifications document before A3/A4 deployment
Acoustic Output and Sound Cap Mode — At 23°C ambient idle, the R650xs produces substantially lower acoustical output than GPU-equipped 1U servers; iDRAC BIOS "Sound Cap" mode limits maximum fan speed for noise-sensitive attended deployments at the expense of a minor performance reduction at peak load; Sound Cap is recommended for lab, staging, and attended-space deployments where standard server noise levels are inappropriate without acoustic enclosures
iDRAC Direct Front Access — USB 2.0 Front + USB 3.0 Rear + Dual VGA
Front Panel — iDRAC Direct Micro-AB USB — Dedicated front-panel micro-USB iDRAC Direct port provides zero-network local iDRAC access at 169.254.0.1 from a directly connected laptop or workstation; configure BIOS settings, retrieve hardware inventory, review System Event Log, deploy firmware, and initiate OS installation from Lifecycle Controller without a network connection to the iDRAC management port — critical for initial staging at the rack and ROBO field service
Front Panel — USB 2.0 and VGA — One USB 2.0 Type-A port (0.5 A) for keyboard, mouse, and bootable OS installation media; one VGA DB-15 for direct console access at the front of the chassis; front VGA eliminates the need to slide or re-cable the server to reach the rear panel during initial BIOS configuration, troubleshooting, and OS installation in dense rack deployments
Rear Panel — USB 2.0 and USB 3.0 — One USB 2.0 (0.5 A, Type-A) and one USB 3.0 (0.9 A, Type-A) on the rear panel; USB 3.0 at 5 Gb/s supports external high-speed storage, OS installation media, and backup devices; both ports are accessible simultaneously with the iDRAC and LOM ports on the rear I/O plate — consistent with the standard Gen 15 PowerEdge rear-panel layout across all R650xs configurations
Rear Panel — Dual LOM, iDRAC, VGA — Two RJ-45 1 GbE LOM ports for production data networking; one dedicated RJ-45 iDRAC9 GbE management port; one rear VGA DB-15; all three Ethernet ports accessible simultaneously without mode switching or port-sharing configuration; combined with OCP 3.0 NIC port(s) and PCIe add-in NIC ports, the R650xs rear supports up to 5+ simultaneous Ethernet connections in a standard rack rear-panel footprint
Internal USB 3.0 and Video Controller — Optional internal USB 3.0 dongle (40 × 16 × 8 mm) for persistent in-chassis storage; embedded Matrox G200 video controller with 16 MB frame buffer; supports resolutions up to 1920 × 1200 at 60 Hz on both front and rear VGA simultaneously; the G200 is sufficient for iDRAC console, BIOS setup, OS installation, and remote KVM workflows — no discrete display adapter required in the R650xs
Front Panel Status LCD (Optional) — Optional LCD bezel (in place of security bezel) provides system health status, IP address, power state, and alert indicators directly on the front panel; readable at a glance in open-access server rooms and data center aisles without requiring SSH or iDRAC web sessions; not compatible with the security bezel variant — only one front bezel type may be installed per chassis at a time
Silicon Root of Trust — Hardware-Anchored Cyber Resilient Architecture Across the Full Lifecycle
Silicon Root of Trust — iDRAC9 boot chain validation is anchored in immutable silicon cryptographic keys embedded at manufacturing; no firmware in the R650xs boot chain executes without passing hardware-level root-of-trust verification; tampered, replaced, or maliciously modified firmware components are detected before execution — eliminating the persistent firmware attack vector targeting server management subsystems in multi-tenant and shared infrastructure environments
Cryptographically Signed Firmware — Every firmware update package covering BIOS, iDRAC, PERC controllers, OCP NIC firmware, BOSS-S1 controller, and all peripheral subsystems carries a Dell EMC cryptographic signature; the platform verifies signatures before applying any update regardless of delivery method (iDRAC web, Lifecycle Controller, Ansible, RACADM, or USB); unsigned or tampered packages are automatically rejected before installation begins
UEFI Secure Boot — OS bootloader and early-boot driver signatures are validated at startup against the UEFI authorized signature database; unauthorized or revoked boot-stage software is blocked before the OS kernel loads; required for PCI-DSS, HIPAA, FedRAMP, CIS Level 2 Benchmark, and DISA STIG compliance frameworks that mandate validated boot chain integrity for server infrastructure
System Lockdown — Locks BIOS, iDRAC, and firmware configurations against unauthorized modification; requires iDRAC9 Enterprise or Datacenter license; all override attempts are recorded in the iDRAC Lifecycle Log with timestamp and source identity; physical presence override available for authorized recovery; hardens insider threat and unauthorized configuration drift risk for R650xs deployments shared across multiple operators or teams
TPM 1.2 / 2.0 — FIPS and CC-TCG Certified — TPM 1.2 and TPM 2.0 hardware security modules are supported; both FIPS 140-2 Level 1 and CC-TCG certified; TPM state is bound to the R650xs system planar and cannot be migrated to a different hardware platform; optional TCM 2.0 (China NationZ) for Chinese Cryptography Standards compliance; TPM enables BitLocker, dm-crypt, vTPM, and other OS-level encryption key management bound to the physical server
Secure Enterprise Key Manager (SEKM) — iDRAC9 optional SEKM integration connects to an external KMIP-compatible key management server for externally managed self-encrypting drive (SED) key lifecycle; SEKM ensures drive encryption keys are not stored on the server — drives encrypted with SEKM cannot be decrypted if physically removed from the R650xs and attached to another system outside the KMIP governance perimeter
Secure Component Verification (SCV) and System Erase — SCV enables cryptographic verification that installed hardware components (CPUs, DIMMs, drives, PERC controllers) match the original factory bill of materials — detecting unauthorized component substitution; System Erase via Lifecycle Controller performs NIST SP 800-88 compliant media sanitization across all HDDs, SSDs, NVMe drives, and system RAM without a deployed OS — mandatory before R650xs decommission, redeployment, or component resale
iDRAC9 Out-of-Band Management — OpenManage Portfolio with Redfish API and Automation
iDRAC9 License Tiers — iDRAC9 Express ships standard on all R650xs configurations; Enterprise adds dedicated GbE NIC for out-of-band management, virtual console, virtual media, power capping, and System Lockdown; Datacenter adds telemetry streaming, extended Redfish scripting, custom thermal controls (delta-T, PCIe inlet control), and GPU management APIs; license upgrades are digital key activations — no hardware change required to move between iDRAC tiers on deployed R650xs servers
iDRAC Service Module (iSM) — Optional lightweight in-band agent running inside the host OS extends iDRAC visibility to OS-level health: process-level status, Windows/Linux crash analysis, software inventory, and OS power consumption metrics; iSM data streams into iDRAC event logs and OpenManage Enterprise dashboard without additional license cost; runs on all certified R650xs OS families without kernel module conflicts
iDRAC Direct and Quick Sync 2 — iDRAC Direct micro-USB front port provides instant local browser-zero access at 169.254.0.1; optional Quick Sync 2 BLE wireless module enables smartphone-based iDRAC management via OpenManage Mobile over Bluetooth — configure initial IP addresses, review hardware inventory, and check system health without a laptop, console cable, or network uplink at the iDRAC port; especially valuable for field technicians at remote and edge sites
Lifecycle Controller — Embedded firmware-based provisioning engine performs bare-metal OS installation, RAID configuration, firmware updates, driver deployment, and BIOS/iDRAC profile application without a deployed OS or PXE infrastructure; change audit trail is persisted in the Lifecycle Controller log for change management and audit reporting; Lifecycle Controller GUI wizard guides first-time setup from blank chassis to running OS in a single sequential workflow
Redfish RESTful API and Automation — DMTF-standard Redfish API on iDRAC9 enables full lifecycle management via curl, Python, PowerShell, Ansible, and Terraform; RACADM CLI provides batch scripted administration for firmware, BIOS, and iDRAC configuration; Dell EMC GitHub Scripting Libraries provide ready-made Python and PowerShell automation scripts for common R650xs management workflows; DSU and DRM support repository-based fleet firmware management across mixed-generation PowerEdge deployments
OpenManage Enterprise Portfolio — OpenManage Enterprise (OME) for one-to-many R650xs lifecycle management across fleets; OpenManage Power Manager for rack and row power capping, energy analytics, and peak demand management; OpenManage SupportAssist for predictive issue detection and automated technical support case creation; Dell EMC Repository Manager for change-controlled firmware workflow governance across R650xs and mixed PowerEdge fleets
Third-Party ITSM and Monitoring Integrations — OpenManage Integrations: VMware vCenter (OMIVV), Microsoft System Center (OMIMSSC), ServiceNow (OMISNOW), Ansible Modules for iDRAC; monitoring platform connections: Nagios Core, Nagios XI, IBM Tivoli, Micro Focus Operations Manager, BMC Truesight; all major enterprise ITSM and monitoring platforms can receive iDRAC health events, metrics, and lifecycle actions from R650xs servers without bespoke integration development
Broad OS Certification — RHEL, SLES, Ubuntu, Windows Server, VMware ESXi, and Citrix
Red Hat Enterprise Linux (RHEL) — Certified for RHEL 7.9 and RHEL 8.x Server x86_64; Lifecycle Controller delivers RHEL driver packs for bare-metal OS installation without PXE or external media; OpenManage Ansible Modules for iDRAC integrate into RHEL-based GitOps pipelines; RHEL High Availability Pacemaker/Corosync clustering on dual-socket R650xs configurations supports active/passive database and NFS-HA workloads without additional clustering hardware
SUSE Linux Enterprise Server (SLES) — Certified for SLES 15 SP2 x86_64 including SLES High Availability Extension (HAE); Lifecycle Controller driver delivery for bare-metal SLES installation; SLES HA cluster on R650xs supports Oracle Database failover, SAP application server HA, and NFS HA patterns; SLES GEO Clustering available for geographically distributed disaster recovery deployments connecting multiple R650xs sites
Canonical Ubuntu Server — Certified for Ubuntu 20.04 LTS; Ansible roles, Terraform providers, and Python iDRAC REST clients natively align with Ubuntu-based DevOps toolchains; cloud-init support for automated bare-metal provisioning through Lifecycle Controller network boot; Ubuntu 20.04 validated for Docker and Kubernetes worker node deployments — the most common container platform OS choice for R650xs Kubernetes cluster nodes
Microsoft Windows Server with Hyper-V — Certified for Windows Server 2019 and 2022 with Hyper-V; Windows Server Failover Cluster (WSFC) supported for SQL Server Always On Availability Groups on R650xs dual-socket NUMA; driver delivery through Lifecycle Controller or Windows Update integration; Microsoft WSSD certification on OCP 3.0 10 GbE iSCSI and SMB Direct configurations validates Windows-based SDS and hyper-converged deployments
VMware vSphere ESXi — Certified for vSphere 6.7 U3, 7.0 U1, and 7.0 U2; Dell OpenManage Integration for VMware vCenter (OMIVV) surfaces R650xs iDRAC firmware lifecycle inside vCenter UI; ESXi vSAN ReadyNode configurations on R650xs NVMe builds enable hyper-converged deployments with in-server all-flash storage; Quick Sync 2 BLE module paired with OpenManage Mobile enables field-technician ESXi host provisioning tasks without vCenter credential access
Citrix XenServer — Certified for Citrix XenServer 8.2 LTSR for VDI and application delivery; IDSDM microSD boot is validated for XenServer deployments using SD media as the hypervisor volume; R650xs dual-socket OCP 3.0 25 GbE NICs provide the display-protocol-delivery bandwidth required for high-density Citrix session density without dedicated GPU acceleration in software-rendered XenDesktop deployments
Lifecycle Controller Driver Delivery — Embedded driver packs in Lifecycle Controller firmware cover all six certified OS families; guided OS installation wizard manages RAID pre-configuration (PERC or fPERC), BIOS settings, NIC driver selection, and OS partition layout from a blank chassis to a running OS in a single workflow; no PXE server, WDS deployment share, or network ISO required — enabling single-person rack-and-provision at remote sites without enterprise deployment infrastructure
ReadyRails II + Stab-In/Drop-In Sliding Rails — Tool-Less 4-Post with CMA and SRB
ReadyRails II Sliding Rails — Drop-in chassis installation; tool-less in 19-inch EIA-310-E square and unthreaded round hole 4-post racks including all Dell rack generations (4220, 4220D, 4820, Titan); tooled installation in threaded 4-post racks; supports full extension of the R650xs out of the rack for complete access to drives, PSUs, fans, DIMMs, and PCIe add-in cards without removing the server — a single service technician can service a running R650xs in the rack without assistance
Stab-In/Drop-In Sliding Rails — Supports both Drop-in and Stab-in installation in square, unthreaded round, and threaded round hole 4-post racks; Stab-in requires attaching inner rail members to the chassis sides first (recommend 2-person lift for 1U); full extension serviceability for all field-replaceable units; CMA and SRB compatible — outer CMA brackets are removable to reduce rail footprint depth and eliminate rear PDU clearance conflicts in deep-cabinet deployments
ReadyRails Static Rails — Stab-in installation only; tool-less in 19-inch square/unthreaded round hole 4-post racks; tooled in threaded 4-post and 2-post relay/open-frame racks; static rails do not extend — no cable management arm or strain relief bar compatibility; use for 2-post telco relay racks, branch IDF closets, and any deployment where full slide-out serviceability is a non-requirement and rack depth is constrained
Cable Management Arm (CMA) — Available with ReadyRails II and Stab-In/Drop-In sliding rails; tool-less left/right swap without additional brackets; ventilated strap baskets retain power, data, and management cables during server slide-out service events; mount CMA on the side opposite the PSU bay in single-PSU R650xs configurations to keep the hot-swap PSU access clear while the CMA is fully loaded with cables
Strain Relief Bar (SRB) — Tool-less attachment to both sliding rail types; two depth positions accommodate different cable bundle sizes and rear rack depths; organizes production data, power, and management cables into separated clean bundles without the arm rotation mechanism of the CMA — preferred for very-high-density rows where CMA arm travel would conflict with adjacent 1U servers or rear-panel PDUs mounted in the same column
Chassis Dimensions and Weight — Height: 42.8 mm (1.7 in, 1U); Width: 482 mm (18.97 in); Depth: 734.95 mm without bezel (48.8 mm CMA connector depth add-on); Maximum weight: 21.9 kg (10-bay chassis at maximum drive/component population); verify floor load and rack weight capacity for fully-populated 42U racks loaded exclusively with dual R650xs 2-PSU configurations at maximum drive density
R650xs vs R640 — Gen 15 Advances in CPU, PCIe, OCP, and Storage Controllers
Processor Platform — 3rd Gen vs 2nd Gen Xeon Scalable — R650xs introduces 3rd Gen Intel Xeon Scalable (Ice Lake-SP, LGA4189) versus R640's 2nd Gen (Cascade Lake-SP, LGA3647); Ice Lake-SP adds native PCIe Gen 4 controller integrated in the CPU die, increases maximum cores per socket to 32 (from 28 on R640's top Gold 6258R), and improves IPC per core across all workload types — a direct compute performance improvement with no platform cost premium for PCIe Gen 4 access
PCIe Generation — Gen 4 vs Gen 3 — R650xs provides up to 3 × PCIe Gen 4 × 16 slots at 16 GT/s per lane; R640 was limited to PCIe Gen 3 at 8 GT/s; each PCIe Gen 4 slot delivers 32 GB/s bidirectional bandwidth versus 16 GB/s on R640 — NVMe Gen 4 SSD performance, 25 GbE NIC throughput, and FPGA data pipeline bandwidth all double on the same hardware; this is a platform-level improvement not requiring any additional investment per server
OCP Upgrade — 3.0 Gen 4 vs rNDC Gen 3 — R650xs integrates OCP 3.0 on PCIe Gen 4 × 8 supporting validated NIC cards from Intel, Broadcom, Mellanox, QLogic, and SolarFlare at up to 100 GbE; R640 used a proprietary rNDC (rack Network Daughter Card) slot on PCIe Gen 3 × 8 limited to Dell-qualified rNDC form factors only; OCP 3.0 multi-vendor open standard eliminates NIC vendor lock-in and enables NIC technology upgrades across all Gen 15 PowerEdge platforms with one card form factor
Memory Speed Improvement — 3200 MT/s vs 2933 MT/s — R650xs DDR4 operates at 3200 MT/s maximum (Gold 6 and Platinum CPUs) versus R640's 2933 MT/s ceiling; R650xs operates at 1 DIMM per channel providing the maximum per-channel bandwidth without multi-rank penalties; R640 with 24 DIMM slots at 12 per CPU ran 2 DPC configurations that could reduce effective speed; workloads that are DDR4 bandwidth-bound see a measurable throughput improvement migrating from R640 to R650xs
RAID Controllers — 15G Stack vs 14G Stack — R650xs introduces PERC H755N (NVMe RAID with NV Flash-Backed cache) which has no R640 equivalent; R640 was limited to PERC H740P (SAS/SATA) for premium RAID; fPERC (H345/H355) on R650xs uses a dedicated small-form-factor planar slot — freeing all user PCIe slots; R640 mini-PERC shared PCIe slot resources; BOSS-S1 on R650xs occupies a dedicated rear riser vs R640's original BOSS module in a shared slot
NVMe Drive Bay Expansion — R650xs supports up to 10 NVMe front bays + 2 optional NVMe rear bays (12 total); R640 supported up to 10 NVMe through a PCIe Gen 3 bridge card; R650xs routes CPU PCIe Gen 4 lanes directly to the front backplane without a bridge card — eliminating the Gen 3 bandwidth bottleneck on R640 NVMe configurations and increasing per-drive bandwidth from 8 GT/s (Gen 3 bridge) to 16 GT/s (Gen 4 direct attach)
Memory Slot Trade-Off — The R640 provided 24 DIMM slots (12 per CPU, 6 channels per CPU) supporting up to 3 TB LRDIMM; R650xs reduces to 16 RDIMM-only slots (8 per CPU, 8 channels × 1 DPC) with 1 TB maximum; organizations whose workloads require more than 1 TB RAM should evaluate the Dell PowerEdge R650 (full) platform which provides 32 slots and LRDIMM support up to 4 TB; R650xs is purpose-designed for workloads within the 1 TB ceiling that prioritize PCIe Gen 4 speed, OCP 3.0, and cost-efficiency over maximum memory density
| Feature | PowerEdge R650xs (Gen 15) | PowerEdge R640 (Gen 14) |
|---|---|---|
| CPU Platform | LGA4189 — 3rd Gen Xeon Scalable (Ice Lake-SP) | LGA3647 — 2nd Gen Xeon Scalable (Cascade Lake-SP) |
| Max Cores per Socket | 32 (Xeon Gold 6338 / Platinum 8358) | 28 (Xeon Gold 6258R) |
| DIMM Slots | 16 (8 per CPU, RDIMM only) | 24 (12 per CPU, RDIMM/LRDIMM) |
| Max DDR4 Capacity | 1 TB (RDIMM only) | 3 TB (LRDIMM) / 1.5 TB (RDIMM) |
| Max Memory Speed | 3200 MT/s DDR4 (1 DPC, Gold 6 / Platinum) | 2933 MT/s DDR4 |
| Intel Optane PMem | Not supported | 10 × PMem 200 Series (select CPU configs) |
| PCIe Generation | Up to 3 × Gen 4 (LP × 16) | Up to 3 × Gen 3 (LP × 16) |
| OCP / Network Slot | OCP 3.0 (PCIe Gen 4 × 8, multi-vendor) | rNDC (PCIe Gen 3 × 8, proprietary) |
| NVMe Drive Bays | Up to 10 front + 2 rear (Gen 4 direct) | Up to 10 (PCIe Gen 3 bridge card) |
| RAID Controllers | PERC H345/H355/H745/H755/H755N (15G) | PERC H330/H730P/H740P/H840/HBA330 (14G) |
| BOSS Module | BOSS-S1 (dedicated rear riser) | BOSS S1 (original shared slot) |
| Max PSU Support | 1800 W 80 PLUS Titanium | 2400 W (dual-socket HPC) |
| GPU Support | Not supported | Limited LP GPU (select configurations) |
Frequently Asked Questions — Dell PowerEdge R650xs
The Dell PowerEdge R650xs supports up to 1 TB of DDR4 ECC RDIMM memory across 16 DIMM slots (8 per socket) running at up to 3200 MT/s with 3rd Gen Xeon Gold 6 or Platinum processors — using 16 × 64 GB registered DIMMs at one DIMM per channel for maximum DDR4 bandwidth. LRDIMM and Intel Optane Persistent Memory are not supported on the R650xs platform.
The R650xs supports up to 12 hot-plug drives in a 1U chassis — 10 front 2.5-inch bays (SAS, SATA, or NVMe Gen 4) plus 2 optional rear 2.5-inch bays. Alternatively the chassis configures with 8 × 2.5-inch or 4 × 3.5-inch LFF front bays. Maximum raw capacity is 76.8 TB using 10 × 7.68 TB NVMe SSDs, plus 15.36 TB in 2 rear bays.
The R650xs supports 3rd Gen Intel Xeon Scalable processors (Ice Lake-SP) in a dual LGA4189 socket configuration, covering the Silver, Gold 4, Gold 6, and Platinum families with up to 32 cores per socket (64 total cores dual-socket). PCIe Gen 4 is natively integrated in each CPU — delivering double the per-lane bandwidth of the previous-generation R640 platform without any additional PCIe hardware.
Yes. Express Computer Systems stocks professionally reconditioned Dell PowerEdge R650xs servers tested, cleaned, and configured to your exact workload specifications. Shop refurbished Dell R650xs servers at ECS.
The R650xs (Gen 15) replaces the R640 (Gen 14) with 3rd Gen Intel Xeon Scalable (Ice Lake-SP, up to 32 cores) vs 2nd Gen (Cascade Lake, up to 28 cores), native PCIe Gen 4.0 vs Gen 3.0, OCP 3.0 open-standard NIC slot replacing the proprietary rNDC, and DDR4 at 3200 MT/s vs 2933 MT/s. The R640 offered 24 DIMM slots and LRDIMM support up to 3 TB — the R650xs reduces to 16 RDIMM-only slots (1 TB max) as a deliberate cost-efficiency trade for organizations that do not require maximum memory density.
Ready to Deploy the Dell PowerEdge R650xs?
Express Computer Systems stocks professionally reconditioned Dell PowerEdge R650xs rack servers configured to your workload. Every unit is tested, cleaned, and ready to ship — build your custom R650xs configuration today.
Start building your custom server today