Dell PowerEdge R6625 — Server Review
Dual-Socket Density for HPC, Virtualization, and EDSFF Gen5 NVMe Workloads
Dense Virtualization — 256 total cores across two EPYC 9004 sockets with hardware-enforced AMD SEV per-VM memory isolation; more VMs per rack unit than any prior 2-socket 1U PowerEdge.
High Performance Computing (HPC) — dual-socket EPYC 9004 delivers up to 32 MB L3 cache per core, DDR5 memory bandwidth, and PCIe Gen5 I/O for scientific and simulation workloads.
Hyper-Converged Infrastructure (HCI) — up to 16 EDSFF E3.S Gen5 NVMe drives in 1U support all-flash HCI with high NVMe density per rack unit.
Software-Defined Storage (SDS) — direct-attach NVMe Gen5 pairs with high dual-socket core count for distributed block and object storage nodes.
Database Consolidation — 6 TB DDR5 capacity with fast NVMe I/O enables in-memory and high-throughput transactional database operations at scale.
Network Functions Virtualization (NFV) — 256 cores and PCIe Gen5 bandwidth support telco-grade NFV with OpenStack for low-latency packet processing.
Virtual Desktop Infrastructure (VDI) — high dual-socket core count maximizes user session density; optional 75W GPU support handles graphics-intensive desktop workloads.
Dual AMD EPYC™ 9004 (Genoa) — Up to 256 Total Cores per Chassis
Dual-socket SP5 platform — two AMD EPYC 9004 series processors share the same 1U chassis via the xGMI3 high-speed interconnect at 32 GT/s for coherent shared memory access.
Up to 128 Zen4 or Zen4c cores per socket — dual-socket configuration delivers up to 256 total cores; Zen4c SKUs maximize thread density for highly parallel workloads at lower TDP.
5nm TSMC process technology — AMD’s Genoa generation improves performance-per-watt versus prior EPYC generations, enabling more compute in a thermally constrained 1U chassis.
Up to 4.1 GHz boost clock — high single-thread frequency benefits latency-sensitive applications, transactional databases, and VDI user density.
Up to 400W TDP per socket — high-TDP configurations require Direct Liquid Cooling (DLC); standard air cooling supports up to 300W TDP configurations.
32 MB L3 cache per core — largest available x86 L3 cache per core; reduces memory access latency for data-intensive workloads running on both sockets.
AMD Infinity Guard — hardware-level AMD SEV (Secure Encrypted Virtualization) and AMD SME (Secure Memory Encryption) protect per-VM and per-process memory at silicon level.
24 DDR5 DIMM Slots — Up to 6 TB at 4800 MT/s
24 × 288-pin DDR5 RDIMM slots — 12 slots per processor; each CPU has its own DDR5 memory subsystem with independent bandwidth per socket.
Up to 6 TB total capacity — achieved with 256 GB 3DS RDIMM modules (octa-rank); supports 16 GB single-rank through 256 GB octa-rank DIMMs across all speed grades.
4800 MT/s operating speed — DDR5 dual sub-channel architecture delivers greater bandwidth per slot than DDR4 at equivalent slot counts.
ECC registration required — all DIMMs must be RDIMM or 3DS RDIMM with ECC enabled; no UDIMMs supported in R6625 configurations.
Capacity by DIMM type — 32 GB dual-rank: up to 768 GB; 64 GB dual-rank: up to 1.5 TB; 128 GB quad-rank: up to 3 TB; 256 GB octa-rank: up to 6 TB.
Note: 256 GB RDIMM not supported in 4×3.5-inch chassis — the 4 x 3.5-inch LFF chassis configuration does not support octa-rank 256 GB DIMMs; maximum is 3 TB in that config.
Bandwidth advantage over R6525 — DDR5 at 4800 MT/s exceeds the R6525’s DDR4 maximum (3200 MT/s), increasing memory bandwidth for HPC and in-memory database workloads.
PERC H965i Through H355 — Full RAID Coverage Across All Drive Tiers
PERC H965i (PERC 12) — 24G SAS4 RAID-on-Chip controller with 8 GB NV cache; highest-tier internal RAID with full RAID 0/1/5/6/10/50/60 support across SAS, SATA, and NVMe.
PERC H755 / H755N — PERC 11 12G SAS3 RAID controllers with 8 GB write cache; H755N is the NVMe-optimized variant for direct-attach NVMe RAID configurations.
PERC H355 — entry-level PERC 11 controller for cost-optimized RAID configurations; supports 12G SAS/SATA with RAID 0/1/5/10.
HBA355i / HBA355e — PERC 11 non-RAID passthrough HBAs for direct OS-visible SAS/SATA; HBA355i internal, HBA355e for external SAS array connectivity.
HBA465e / HBA965e — external 24G SAS4 HBAs for direct connectivity to external SAS JBOD enclosures; up to 2 HBA465e supported in certain riser configurations.
BOSS-N1 internal boot — Boot Optimized Storage Subsystem with Hardware RAID 1 across 2 × M.2 NVMe SSDs; no PCIe slot consumed; independent of all data drive bays.
Software RAID (S160) — OS-level NVMe RAID option for configurations where a hardware RAID controller is not required.
Five Chassis Configurations — Up to 16 EDSFF E3.S Gen5 NVMe Drives
4 × 3.5-inch SAS/SATA front bays — up to 80 TB raw; supports 7.2K and 10K HDDs plus SATA SSD; ideal for high-capacity nearline and archival storage at lowest cost per TB.
8 × 2.5-inch NVMe SSD front bays — up to 122.88 TB all-NVMe; PCIe Gen5 direct-attach NVMe SSDs for performance-dense deployments.
10 × 2.5-inch SAS/SATA/NVMe front bays — up to 153.6 TB; universal backplane accepts SAS, SATA, or NVMe drives for maximum configuration flexibility.
14 × EDSFF E3.S Gen5 NVMe front bays — up to 89.6 TB; E3.S form factor delivers higher drive density than 2.5-inch U.2 in the same 1U space.
16 × EDSFF E3.S Gen5 NVMe front bays — up to 102.4 TB; maximum NVMe density with full PCIe Gen5 bandwidth per drive.
Rear bay options — 2 × 2.5-inch SAS/SATA (up to 30.72 TB) or 2 × E3.S Gen5 NVMe (up to 12.8 TB); rear module adds capacity without consuming front bays.
Hot-swap across all configurations — front and rear drive carriers are hot-swappable for live replacement without system downtime.
BOSS-N1 — Dedicated NVMe OS Boot, All Data Bays Free
BOSS-N1 module — Boot Optimized Storage Subsystem (NVMe generation); dedicated M.2 NVMe SSD carrier mounts in a rear internal slot separate from all front data bays.
Hardware RAID 1 mirrored pair — two M.2 NVMe SSDs in hardware-mirrored configuration provide OS boot redundancy; one drive failure does not interrupt operation.
No PCIe slot consumed — BOSS-N1 connects via a dedicated internal connector, leaving all expansion card slots free for NICs, HBAs, and GPUs.
Hot-serviceable design — BOSS-N1 module is accessible from the rear without powering down the server; M.2 SSD replacement does not require full chassis disassembly.
UEFI Secure Boot integration — BOSS-N1 supports UEFI Secure Boot; integrates with iDRAC9 Lifecycle Controller for OS deployment and recovery workflows.
Frees all front bays for data — because the OS lives on BOSS-N1, all front drive bays and rear drive modules are available exclusively for application data storage.
Up to 3 × 75W Single-Width GPUs — PCIe Gen5 Acceleration in 1U
Up to 3 × 75W single-width GPUs — Config8 (R2A+R4 E3 paddle card) riser supports three NVIDIA single-width GPUs across slots 1 and 2 in the 1U chassis.
Single-width (SW) form factor — all GPU slots accept single-width cards; full-height single-width (FH/SW) cards fit in full-height riser configurations (Config3).
PCIe Gen5 GPU interface — in Gen5-capable riser configurations the GPU connects via PCIe Gen5 x16 for maximum bandwidth between processor and accelerator.
75W TDP — no auxiliary power required — GPU draws all power directly from the PCIe slot; no riser-mounted auxiliary power connectors needed in a 1U chassis.
Typical GPU use cases — AI inference, visualization, video transcoding, GPU-accelerated HPC, and VDI graphics offload for user session density improvement.
GPU count depends on riser selection — available GPU slots vary by riser; Config5 (R3P) and Config7 (R2A 1CPU min) support 1 GPU; Config6 (R2S+R3S) supports 2; Config8 supports up to 3.
Up to 4 PCIe Slots — Eight Dual-Processor Riser Configurations
Up to 4 × PCIe slots — x8 or x16 lane width; supports both Gen4 and Gen5 expansion cards across all riser configurations depending on processor connectivity.
Eight riser configurations — Config1 (R2A+R3A), Config2 (R2P+R3P), Config3 (R1P+R4P), Config4 (R2Q), Config5 (R3P), Config6 (R2S+R3S), Config7 (R2A), Config8 (R2A+R4 E3) — each optimized for different workload profiles.
PCIe Gen5 primary slots — Riser 2P, 3P, 1P, 4P, 2Q, 2S, 3S all provide PCIe Gen5 x16 links to Processor 1 or Processor 2 for CPU-direct high-bandwidth connectivity.
Full Height (FH) slot options — Config3 (R1P+R4P) provides two full-height half-length PCIe Gen5 slots for full-height GPU or FPGA accelerators.
Low Profile (LP) slots — most configurations use low-profile half-length slots for 1U chassis compatibility; LP NICs, HBAs, and SAS controllers all supported.
External SAS / NIC adapters — HBA465e and H965e external adapters supported in up to 3 slots depending on riser; InfiniBand (Mellanox NDR200) also supported in PCIe slots.
OCP 3.0 slot independent — dedicated integrated OCP 3.0 slot is separate from riser PCIe slots; accepts OCP 3.0 cards up to 25 GbE × 4 without consuming a riser slot.
OCP 3.0 Networking — Up to 25 GbE per Slot Plus PCIe NIC Options
Dedicated OCP 3.0 slot — one integrated OCP 3.0 card slot occupies a dedicated position on the system board independent of riser PCIe slots; supports both SCFF and full-size OCP cards.
OCP 3.0 speeds — 1 GbE × 4, 10 GbE × 2, 10 GbE × 4, 25 GbE × 2, or 25 GbE × 4; card selection depends on workload bandwidth requirements.
Optional 2 × 1 GbE LOM card — integrated LAN-on-Motherboard (optional) adds 1 GbE × 2; can be installed alongside an OCP card for dual-NIC management and data network separation.
PCIe NIC expansion — NICs up to 100 GbE (Mellanox, Broadcom) or 400 GbE (Mellanox) can be installed in PCIe expansion slots for high-bandwidth HPC fabrics.
InfiniBand support — Mellanox NDR200 InfiniBand HCA supported in PCIe slots for HPC fabrics at 200 Gbps per port.
Dedicated iDRAC9 1 GbE port — separate rear-panel Ethernet port dedicated to out-of-band management; never shares bandwidth with production network traffic.
Multi-rail topology — install LOM + OCP + PCIe NICs simultaneously for separate management, storage, and production network paths in a single 1U chassis.
800W to 1800W — Platinum & Titanium Hot-Swap Redundant PSUs
800W Platinum Mixed Mode — AC (100–240V) or 240V DC; entry-level redundant PSU for standard air-cooled configurations; C13 power cord.
1100W Titanium Mixed Mode — AC or 240V HVDC; highest efficiency tier at 96%+ (Titanium 80 PLUS); recommended for dense workloads in AC environments.
1100W LVDC — -48V to -60V DC input for telecom and HVDC data center power environments; C13 power cord.
1400W Platinum Mixed Mode — AC or HVDC for high-TDP dual-processor configurations; derated to 1050W at 100–120V low-line AC input.
1800W Titanium — 200–240V AC or 240V HVDC only; highest available wattage; C15 power cord; required for full dual-socket DLC configurations at maximum TDP.
N+1 hot-swap redundancy — dual hot-swap PSUs provide N+1 redundancy; one PSU failure does not interrupt server operation or require shutdown.
Tool-free PSU replacement — PSU slides out and replaces from the rear without powering down the server; no tools required for field PSU swaps.
SmartCooling — Standard Air Cooling + Optional Direct Liquid Cooling
Up to 4 sets of hot-plug fan modules — each fan module is dual-fan; system supports STD (Standard) or HPR Gold (High Performance Gold) fan types based on CPU TDP and drive configuration.
HPR Gold fans for high-TDP configurations — HPR Gold fans provide higher airflow for configurations with high-TDP processors or dense NVMe storage; required for certain CPU/drive combinations per the thermal restriction matrix.
Air shroud required for all air-cooled configs — directs airflow from front intake fans across CPUs, memory, and storage before exhausting out the rear; caution: never operate without the air shroud installed.
Optional Direct Liquid Cooling (DLC) — liquid cooling heat sink modules replace air heat sinks for 300W+ TDP CPU configurations; cold plates connect to facility coolant manifold via rear-panel coolant tubes.
DLC is a rack solution — requires rack manifolds and a Cooling Distribution Unit (CDU) installed in the rack infrastructure; DLC is not a standalone server option.
SmartCooling algorithm — iDRAC9 dynamically adjusts fan speed based on inlet temperature, CPU utilization, drive load, and ambient temperature sensors for optimal acoustic and thermal balance.
Thermal restriction matrix — not all CPU SKUs are compatible with all drive configurations; Dell’s thermal restriction matrix defines valid CPU/drive/fan combinations to maintain system stability.
Front, Rear & Internal I/O — USB 3.0, VGA, iDRAC Direct
Front panel — 1 × iDRAC Direct (Micro-AB USB 2.0) for local BIOS/iDRAC management without network access; 1 × USB 2.0; 1 × VGA for direct KVM console connection.
Rear panel — 1 × USB 3.0 (9-pin, SuperSpeed); 1 × USB 2.0 (4-pin); 1 × VGA; 1 × dedicated iDRAC9 Ethernet; OCP NIC port(s); optional LOM NIC port(s).
Optional serial COM port — 9-pin DTE, 16550-compliant serial adapter installs in an expansion bracket slot; required for legacy serial console management.
Internal USB — 1 × internal USB 3.0 port for internal USB flash media or hypervisor boot drives; accessible from inside the chassis.
iDRAC Direct LED — iDRAC Direct port includes a status LED: solid blue = connected, blinking blue = active data transfer.
Dual-VGA layout — one VGA port on the front panel and one on the rear; front VGA is useful for initial BIOS setup, rear VGA for permanent rack KVM switch cabling.
Integrated Matrox G200 graphics — 16 MB frame buffer; supports resolutions up to 1920 × 1200 at 60 Hz for management console use.
Cyber Resilient Architecture — Silicon Root of Trust & AMD Infinity Guard
Silicon Root of Trust — cryptographic hardware anchor in the iDRAC9 silicon verifies firmware integrity at every boot; prevents execution of unsigned or tampered firmware images.
AMD Secure Encrypted Virtualization (SEV) — hardware-level per-VM memory encryption; even a hypervisor compromise cannot access guest VM memory in plaintext.
AMD Secure Memory Encryption (SME) — full-system memory encryption at the memory controller; all DRAM encrypted transparently without application modification.
Secured Component Verification — hardware integrity check validates that all server components match the factory configuration; detects substituted or counterfeit parts at startup.
Cryptographically signed firmware — every firmware image is signed by Dell; iDRAC9 validates signatures before installation and before execution at each boot.
System Lockdown — locks all firmware and configuration changes when enabled; requires iDRAC9 Enterprise or Datacenter license; prevents unauthorized modification during live operation.
TPM 2.0 / Secure Boot / Secure Erase — TPM 2.0 (FIPS and CC-TCG certified) stores cryptographic keys and enables measured boot; Secure Erase cryptographically sanitizes drives to NIST 800-88 standard before decommission.
iDRAC9 — Out-of-Band Lifecycle Management & Redfish API
iDRAC9 embedded in every server — dedicated management processor independent of the main CPU and OS; remains accessible even when the server is powered off or the OS is unresponsive.
RESTful Redfish API — industry-standard API enables infrastructure-as-code management; integrates with Ansible, Terraform, and third-party orchestration platforms at scale.
iDRAC Direct — local management via Micro-AB USB port on the front panel; connect a laptop directly without needing network access or preconfigured BIOS credentials.
Quick Sync 2 wireless module — optional Bluetooth/Wi-Fi module enables OpenManage Mobile (OMM) to configure and monitor server health from a mobile device.
OpenManage Enterprise (OME) — Dell’s multi-server management console provides one-to-many lifecycle management including firmware updates, power capping, and configuration compliance reporting.
Lifecycle Controller — manages firmware updates, OS deployment, hardware configuration, and system logs without an OS installed; works in conjunction with iDRAC9 for bare-metal provisioning.
OpenManage integrations — plugins for VMware vCenter, Microsoft SCOM, ServiceNow, Red Hat Ansible, Terraform, and Windows Admin Center for ecosystem-native server management workflows.
Windows Server, RHEL, Ubuntu, SLES & VMware ESXi
Microsoft Windows Server (with Hyper-V) — certified for Windows Server 2019 and 2022; Hyper-V role supported for in-OS VM consolidation; tested and validated by Dell Technologies.
Red Hat Enterprise Linux (RHEL) — RHEL 8.x and 9.x certified; includes Dell-specific kernel drivers and hardware enablement packages for EPYC 9004 platform features.
Canonical Ubuntu Server LTS — Ubuntu 20.04 LTS and 22.04 LTS certified; ideal for Kubernetes, OpenStack, and DevOps CI/CD pipeline deployments.
SUSE Linux Enterprise Server (SLES) — SLES 15.x certified; common in SAP HANA and mission-critical enterprise Linux deployments requiring long-term support.
VMware ESXi — ESXi 7.0 and 8.0 certified; validated for vSAN, NSX-T, and vSphere Distributed Resource Scheduler (DRS) for automated VM workload balancing.
OS deployment via Lifecycle Controller — iDRAC9 Lifecycle Controller enables one-click OS installation without an external boot server or USB media; supports PXE boot for network-based OS provisioning.
Full OS support list — visit www.dell.com/ossupport for the complete and current list of certified operating systems and hypervisors for the PowerEdge R6625.
Universal Rail Kit — 1U 19-Inch Rack Ready with CMA Option
1U rack form factor — 42.8 mm height (1.685 in); 482 mm width (18.97 in); standard 19-inch EIA rack compatible with all major rack vendors and data center standards.
Depth: 822.89 mm (32.4 in) with bezel — verify rack depth before installation; most modern 1000 mm deep racks accommodate the R6625 without clearance issues.
Rail sizing compatibility matrix — Dell publishes an Enterprise Systems Rail Sizing and Rack Compatibility Matrix with specific rail types, adjustability ranges, and rack mounting flange details for all PowerEdge models.
Cable Management Arm (CMA) option — optional CMA organizes rear cables and allows the server to be extended from the rack on the rails without disconnecting cables during servicing.
Tool-less rail design — standard PowerEdge rails support tool-free installation in square-hole racks; adapter kits available for round-hole and threaded-hole racks.
Maximum weight: 20.4 kg (44.97 lbs) fully loaded — plan rack load capacity accordingly; 15.8 kg without drives and PSU for initial rack mounting before component population.
Quick Resource Locator (QRL) — QR code on the system information tag links to setup videos, service manual, and configuration-specific support resources for fast field servicing.
R6625 vs. R6525 — What’s New in 16th Generation
4th Gen AMD EPYC 9004 Genoa (SP5) vs 2nd/3rd Gen EPYC (SP3) — xGMI3 interconnect doubles from 16 GT/s to 32 GT/s; PCIe Gen5 replaces Gen4/Gen3 for higher per-lane I/O bandwidth.
DDR5 at 4800 MT/s vs DDR4 at 3200 MT/s — 24 DDR5 DIMM slots deliver up to 6 TB total capacity; the R6525 used 32 DDR4 slots but at lower per-slot bandwidth.
EDSFF E3.S Gen5 NVMe support added — the R6625 adds up to 16 × E3.S Gen5 front bays and 2 E3.S rear bays; the R6525 had no EDSFF configuration option.
PERC 12 (H965i) added — the H965i brings 24G SAS4 RAID and NVMe RAID capability; the R6525 used PERC 11 (H745/H840) only.
Optional Direct Liquid Cooling — R6625 supports DLC for 300W+ TDP processor configurations; the R6525 was air-cooling only.
Secured Component Verification added — hardware integrity check for component validation at boot; not available on R6525.
OCP 3.0 full-size slot plus optional LOM — R6625 supports full-size OCP 3.0 up to 25 GbE × 4 plus an optional 2 × 1 GbE LOM card; the R6525 had no LOM riser support.
| Feature | R6625 (Gen 16) | R6525 (Gen 15) |
|---|---|---|
| Processor | AMD EPYC 9004 Genoa (SP5) | AMD EPYC 2nd/3rd Gen (SP3) |
| CPU Interconnect | xGMI 32 GT/s | xGMI 16 GT/s |
| Memory | 24 DDR5 slots, up to 6 TB, 4800 MT/s | 32 DDR4 slots (RDIMM/LRDIMM) |
| EDSFF Storage | Up to 16 × E3.S NVMe Gen5 | Not supported |
| PCIe Generation | Gen5 (primary) + Gen4 | Gen4 / Gen3 |
| Top RAID Controller | PERC H965i (PERC 12, 24G SAS4) | H840 / H745 (PERC 11) |
| Direct Liquid Cooling | Optional DLC | Air cooling only |
| GPU Options | Up to 3 × 75W SW | Up to 3 × 75W SW/LP |
| Security | Silicon Root of Trust + SCV | Silicon Root of Trust |
| Management | iDRAC9 | iDRAC9 |
| Form Factor | 1U rack | 1U rack |
ProSupport & ProDeploy — End-to-End Service Coverage
ProSupport for Enterprise — 24×7 access to Dell Technologies senior engineers with direct escalation to engineering; remote diagnostics, firmware analysis, and field dispatch coordination.
ProSupport Plus — adds SupportAssist automated case creation using iDRAC9 telemetry; predictive hardware failure detection reduces unplanned downtime before components fail.
ProDeploy Enterprise Suite — factory integration, rack-and-stack, and on-site deployment; server arrives pre-configured and deployed within the contracted window.
Dell Consulting Services — architecture design, migration planning, and workload optimization by Dell Technologies consultants across 170+ global service locations.
170+ country coverage — 60K+ certified engineers and partners; SLA options from next business day to 4-hour on-site parts and labor response.
Data Migration Services — assisted data migration from legacy infrastructure; supports physical-to-virtual (P2V) and storage migrations with project management.
Extended Life support — post-end-of-service-life support contracts maintain coverage on refurbished infrastructure aged beyond standard warranty windows.
Titanium PSUs & EPYC Genoa — Performance-per-Watt at Scale
AMD EPYC 9004 5nm efficiency — Genoa’s 5nm TSMC process delivers improved performance-per-watt over prior EPYC generations; more compute output per watt consumed in the 1U chassis.
Titanium 80 PLUS PSUs — the 1100W and 1800W Titanium PSUs achieve 96%+ AC-to-DC conversion efficiency; lower waste heat and reduced cooling load versus Platinum or Gold tier PSUs.
Optional Direct Liquid Cooling — DLC removes heat directly from the CPU at source; reduces air cooling dependency and can lower overall data center cooling energy consumption per rack.
iDRAC9 Power Manager — OpenManage Power Manager plugin enables real-time power capping, rack-level power budgeting, and per-server power consumption analytics.
ENERGY STAR and 80 PLUS certification — Dell PowerEdge products are designed to meet ENERGY STAR, 80 PLUS, Climate Savers, and EPEAT environmental certification standards.
High-density EDSFF reduces drive count per TB — E3.S drives deliver more capacity per 1U than legacy 2.5-inch drives; fewer drives equals less power consumption per TB of usable storage.
Sustainability commitment — Dell Technologies provides end-of-life electronics recycling, responsibly sourced materials, and circular economy programs for decommissioned server infrastructure.
Frequently Asked Questions — Dell PowerEdge R6625
The R6625 supports two AMD EPYC 4th Generation 9004 Series processors in a dual-socket (SP5) configuration. Each socket can host a CPU with up to 128 Zen4c cores, delivering up to 256 total cores per chassis. High-core-count Genoa SKUs make the R6625 ideal for dense virtualization, HPC, and NFV workloads requiring maximum thread count in a 1U form factor.
The R6625 supports five front chassis configurations: up to 4 × 3.5-inch SAS/SATA (80 TB), 8 × 2.5-inch NVMe (122.88 TB), 10 × 2.5-inch SAS/SATA/NVMe (153.6 TB), 14 × EDSFF E3.S Gen5 NVMe (89.6 TB), or 16 × EDSFF E3.S Gen5 NVMe (102.4 TB) front drives. Optional rear bays add up to 2 × 2.5-inch SAS/SATA or 2 × E3.S NVMe drives for additional capacity without consuming front bays.
Yes. The R6625 supports optional Direct Liquid Cooling (DLC) for high-TDP processor configurations exceeding standard air cooling limits. DLC uses liquid cooling heat sink modules that replace air heat sinks on the CPUs, with coolant tubes routed through the rear panel. DLC requires rack manifolds and a Cooling Distribution Unit (CDU) installed in the rack infrastructure and is not a standalone server option.
Yes. Express Computer Systems stocks professionally reconditioned refurbished Dell PowerEdge R6625 servers tested, cleaned, and configured to your specifications — ready to deploy at significant cost savings versus new. Shop refurbished Dell R6625 servers at ECS.
The R6625 (Gen 16) succeeds the R6525 (Gen 15) with AMD EPYC 4th Gen Genoa (SP5) replacing 2nd/3rd Gen EPYC (SP3). Key upgrades include DDR5 memory up to 6 TB at 4800 MT/s (vs DDR4), EDSFF E3.S Gen5 NVMe support with up to 16 drives (the R6525 had none), PCIe Gen5, PERC 12 (H965i) RAID controller, optional Direct Liquid Cooling, Secured Component Verification for hardware integrity, and a doubled xGMI interconnect (32 GT/s vs 16 GT/s).
Ready to Deploy the Dell PowerEdge R6625?
Express Computer Systems specializes in professionally refurbished Dell PowerEdge servers — tested, reconditioned, and configured to your exact specifications at a fraction of new pricing.
Start building your custom server today