Virtualization & Private Cloud — Pack more VMs per rack unit with up to 2x 144 E-cores (288 total) and 8TB DDR5 memory, delivering industry-leading VM density in a compact 1U chassis.

High-Performance Computing (HPC) — Run simulation, modeling, and research workloads across 288 E-cores or 172 P-cores with full PCIe Gen5 interconnect bandwidth between processor domains.

Scale-Out Database — Drive OLTP, analytics, and in-memory databases with dual-socket DDR5 bandwidth and up to 20x EDSFF E3.S Gen5 NVMe for extreme I/O throughput in a 1U node.

AI Inference & GPU-Accelerated Workloads — Deploy up to 3x NVIDIA L4 24GB GPUs for AI inference, video transcoding, and edge AI with full PCIe Gen5 x16 bandwidth per accelerator.

Cloud-Scale Microservices — Handle containerized applications and Kubernetes orchestration with dual-socket core density and flexible front/rear I/O configurations for diverse network fabrics.

Big Data & Analytics — Power Hadoop, Spark, and streaming analytics pipelines with up to 20x E3.S Gen5 NVMe drives and dual-socket memory bandwidth exceeding 800 GB/s aggregate.

Software-Defined Storage — Deploy ultra-dense 1U all-flash NVMe storage nodes with up to 20x EDSFF E3.S Gen5 drives in distributed or hyperconverged architectures.

32 DIMM Slots — 288-pin DDR5 RDIMM sockets across 16 memory channels (8 per processor) — doubling the available bandwidth and capacity vs single-socket platforms.

E-Core Configuration — Supports up to 2TB max using 32GB or 64GB RDIMMs at 6400 MT/s (1DPC) or 5200 MT/s (2DPC). 32GB minimum capacity with 1 DIMM per channel configuration.

P-Core Configuration — Supports up to 8TB max using 16GB through 256GB RDIMMs including 96GB, 128GB, and 256GB 8R DIMMs. Enables memory mirroring and Fault Resilient Mode (FRM) for RAS-critical deployments.

RDIMM Capacities — 16GB (1R), 32GB (2R), 64GB (2R), 96GB (2R), 128GB (2R), 256GB (8R) — all ECC DDR5 at 1.1V. 16GB, 96GB, 128GB, and 256GB supported with P-Core only.

DDR5 Exclusively — Supports registered ECC DDR5 DIMMs only; no DDR4 compatibility. All DIMM slots must be populated with identical DIMMs (one Dell PN) — mixing configurations are not allowed.

Memory Mirroring & FRM — Memory mirroring (full redundancy) supported on P-Core only. Fault Resilient Mode available on P-Core with 8 or 16 DIMMs per processor for mission-critical uptime requirements.

H965i Front DC-MHS — High-performance PERC 12 internal RAID controller using the DC-MHS connector; optimized for NVMe E3.S configurations without occupying a PCIe riser slot.

H365i Front DC-MHS — Standard PERC 12 internal front RAID controller via DC-MHS; suitable for SAS/SATA drive configurations with RAID 0, 1, 5, 6, 10, 50, 60 support.

H975i Front DC-MHS — Premium PERC 12 internal controller with higher cache and throughput; ESXi 9.0 or later required for full support.

External Controllers — PERC H965e (RAID) and HBA 465e (pass-through) for connecting external 12Gb SAS JBOD enclosures and MD-series storage arrays.

No Tri-Mode — PowerEdge 17th Gen does not support mixing SAS, SATA, and NVMe behind the same controller; select the controller matched to your drive protocol.

External Storage — Supports external USB tape, NAS/IDM appliance software, and 12Gb MD-series JBOD connections for storage expansion beyond direct-attach capacity limits.

Up to 3x NVIDIA L4 24GB 72W GPU — Single-width Low Profile GPUs connected via PCIe Gen5 x16 slots, enabling AI inference, video transcoding, edge AI, and GPU-accelerated database operations in a 1U chassis.

72W Single-Width Form Factor — NVIDIA L4 operates within the 75W PCIe slot power budget without requiring auxiliary power cables — simplifying rack cabling and PSU capacity planning.

Full PCIe Gen5 Bandwidth — Each GPU slot operates at full x16 Gen5 bandwidth (32 GT/s) — no lane sharing required — enabling GPU-direct storage access and NVLink-style high-bandwidth workloads.

Workload Fit — Real-time AI inference at the edge, video analytics pipelines, GPU-accelerated SQL, and multi-model AI serving with NVIDIA Tensor Core acceleration on 24GB GDDR6.

Riser Configuration Dependent — GPU availability depends on the riser configuration selected (R2b, R4b for full-height). Review riser compatibility matrix when configuring GPU + storage combinations.

Up to 2x OCP NIC 3.0 Slots — Front I/O configurations support 2x OCP slots (positions 31 and 32); rear I/O configurations provide 1x OCP slot at the rear. All OCP 3.0 cards connect on PCIe Gen5 x16.

Port Speed Options — 1GbE x4, 10GbE x2/x4, 25GbE x2/x4, 100GbE x2 — select based on fabric topology and application bandwidth requirements.

Shared iDRAC NIC Port — The primary OCP NIC card shares a port with iDRAC for out-of-band management, reducing required physical ports in certain deployment configurations.

Dedicated iDRAC Ethernet — Dedicated 1Gb iDRAC management port (DC-SCM) available at front (FIO) or rear (RIO) for independent management-plane traffic that doesn’t share data-plane bandwidth.

OCP 3.0 vs 2.0 — OCP 3.0 uses PCIe Gen5 (vs Gen3 on OCP 2.0) and supports up to x16 lanes in a small form factor, enabling next-generation 100GbE and future 200GbE throughput without consuming a full PCIe slot.

Up to 4x Dual Fan Modules — Hot-swappable dual fan sets with N+1 redundancy; continuous operation with one failed fan module — no downtime required for fan replacement.

Standard (STD) Fans — Default option for lower-TDP processor configurations (e.g. 6507P at 150W). Quietest idle acoustic profile for office-adjacent or edge deployments.

High Performance Silver (HPR SLVR) Fans — Required for processor TDPs above approximately 205W. Supports up to 330W E-Core and 350W P-Core configurations with extended heatsink.

1U Extended & L-Type Heatsinks — 1U EXT heatsink supports processor TDPs up to 270W; L-Type heatsink required for TDPs ≥270W (e.g. 6780E at 330W, high-TDP P-Core SKUs).

Optional Direct Liquid Cooling (DLC) — The 6780E 144-core 330W E-Core processor requires DLC in configurations with 16x or 20x E3.S drives. DLC module connects at the rear of the chassis with coolant supply and return lines.

ASHRAE A2 Standard — 10–35°C operating range for conventional data center environments with ISO Class 8 air filtration. Extended ambient support varies by processor TDP and configuration.

iDRAC Closed-Loop Control — iDRAC dynamically adjusts fan speeds using inlet temperature, CPU/DIMM/drive sensors, and system inventory — minimizing acoustic output and energy consumption across load levels.

Silicon Root of Trust — Hardware-anchored cryptographic boot chain verifies firmware integrity at the silicon level before any software loads — protecting against supply chain attacks and firmware tampering from factory to site.

TPM 2.0 (FIPS, CC-TCG Certified) — Trusted Platform Module 2.0 provides hardware-based key storage, attestation, and cryptographic operations for drive encryption and platform integrity verification.

Secure Boot — UEFI Secure Boot validates bootloader and OS kernel signatures, blocking unauthorized operating systems and rootkits from loading during the boot process.

Data at Rest Encryption — Self-Encrypting Drive (SED) support with local or external key management for NIST-compliant data protection across SAS, SATA, NVMe, and E3.S drive types.

Secured Component Verification — Factory-to-site hardware integrity check ensures no components were replaced or tampered with during shipping — critical for supply chain zero-trust requirements.

System Lockdown (iDRAC10) — Requires iDRAC10 Enterprise or Datacenter; locks all firmware update and configuration change pathways, preventing unauthorized modifications even with physical chassis access.

Chassis Intrusion Detection & Secure Erase — Intrusion switch logs chassis access events; Secure Erase cryptographically sanitizes all storage media prior to decommission for data compliance.

Multi-Factor Authentication (MFA) — Enforces MFA across iDRAC and management interfaces, reducing credential-based attack surface in Zero Trust IT environments.

Canonical Ubuntu Server LTS — Long-term support Ubuntu for stable enterprise Linux deployments with full iDRAC10 driver support and Lifecycle Controller integration.

Red Hat Enterprise Linux (RHEL) — Industry-standard enterprise Linux with Dell-certified driver stacks and one-click OS deployment via Lifecycle Controller.

SUSE Linux Enterprise Server (SLES) — High-reliability Linux for SAP workloads and mission-critical enterprise applications requiring long-term kernel stability.

VMware ESXi — Leading type-1 hypervisor for virtualized infrastructure; Dell-certified with OpenManage Integration for vCenter for lifecycle management directly from the vSphere console. Note: H975i PERC requires ESXi 9.0 or later.

Microsoft Windows Server & Windows Server Datacenter — Full Microsoft certification including Hyper-V virtualization, Windows Admin Center integration, and NVIDIA L4 GPU support for virtual desktop and AI inference workloads.

OEM-Ready — Available in OEM and OEMR configurations; hardware can be customized from bezel to BIOS to packaging for white-label or branded customer deployments.

Processor Core Explosion — Intel Xeon 6 E-Core (up to 144 cores per socket, 288 total) vs 4th/5th Gen Xeon Scalable (up to 32 cores per socket, 64 total) — up to 4.5× more cores in the same 1U dual-socket form factor.

Memory Speed & Capacity — DDR5 at 6400 MT/s (1DPC) vs 5200 MT/s on R660; up to 8TB max (P-Core) vs ~4TB on R660 — more capacity and approximately 23% faster bandwidth for data-intensive applications.

EDSFF E3.S NVMe Added — R670 supports up to 20x EDSFF E3.S Gen5 NVMe drives vs R660 with no E3.S support — enabling next-generation NVMe storage density in the same chassis.

DLC Support Added — R670 adds optional Direct Liquid Cooling for the highest-TDP 330W E-Core processor SKUs vs R660 which is air-only — enabling denser rack deployments without thermal throttling.

PCIe Generation Upgrade — R670 delivers PCIe Gen5 across all expansion slots vs a mix of Gen4/Gen5 on R660 — doubling per-lane bandwidth for accelerators, NVMe, and 100GbE NICs.

iDRAC10 vs iDRAC9 — R670 ships with iDRAC10 offering enhanced automation, expanded Redfish API coverage, and next-generation security features including strengthened System Lockdown.

Higher Maximum PSU Wattage — R670 supports up to 1800W PSUs vs 1500W max on R660 — necessary for dual high-TDP processor configurations with GPU complement and DLC module.

Titanium PSU Efficiency — Up to 1800W Titanium PSUs achieve industry-leading efficiency at load; combined with Platinum options across the wattage range to match infrastructure power delivery.

1% Power Monitoring Accuracy — Dell’s 1% monitoring accuracy (5× the 5% industry standard) enables right-sizing of power infrastructure and UPS capacity — reducing stranded capacity and energy waste.

N+1 Fan Redundancy & Closed-Loop Cooling — iDRAC dynamically sets minimum required fan speeds using real-time sensor data — reducing idle power consumption vs fixed-speed cooling alternatives.

Intel Xeon 6 E-Core Efficiency — Efficient-core architecture delivers superior performance-per-watt vs prior Xeon generations; more workloads per kilowatt-hour means fewer servers and a smaller data center footprint.

Optional DLC Reduces Cooling Overhead — Direct Liquid Cooling moves processor heat directly to facility cooling loops at higher efficiency than air-based cooling for the highest-TDP configurations.

Dell EIPT — Enterprise Infrastructure Planning Tool calculates precise power consumption across server, storage, and networking at specific workload levels — enabling accurate TCO modeling before purchasing decisions.

Recycled Materials & Responsible Sourcing — Dell incorporates recycled content in products and packaging and offers end-of-life take-back to support circular economy goals.