






(List below serves as quick links to each section)
4U rack server
Processor | Max: 112 CoresUp to 4x 2nd Gen Intel Xeon Scalable
Memory | Max: 15.36TB48 DDR4 RDIMM, LRDIMM 2933Mhz Max 6TB
Up to 12 NVDIMM, Max 384GB
Up to 24 Intel® Optane™ DC persistent memory PMem, max 12.28TB
(15.36TB max with PMem and LRDIMM)
Supports registered ECC DDR4 DIMMs only
Up to 8x 2.5” SAS SATA (HDD/SSD) max 61.44TB
Up to 24 x 2.5” SAS/SATA (HDD/SSD) max 184.32TB
Up to 32 SAS/SATA (HDD/SSD) with 4 NVME SSD max 245.76TB
None
Boot Options:BOSS
USB
IDSDM (Virtualized Platforms only)
Up to 12 slots, (6x16 slots or 2x16 + 10x8 slots)
Remote management:iDRAC9 with Lifecycle Controller
LOM / rNDC:4 x 1GbE, 4 x 10GbE, 2 x 10GbE+2 x 1GbE, or 2 x 25GbE
RAID controllers:S140 (SW RAID | SATA, NVMe)
PERC H330, H730, H730P, H740P, H750, H840
HBA330, HBA350, HBA355i, HBA355e, 12 Gbps SAS HBA
Up to 4 double-width GPUs or
8 single-width full-height FPGAs
(Click to View)
The PowerEdge R940xa drives GPU database acceleration to deliver real-time decisions for compute-intensive applications. By combining four CPUs with four GPUs, the R940xa delivers consistently high performance for demanding applications. The R940xa allows you to dynamically scale capacity and performance as your business-critical workloads evolve including:
The Dell EMC OpenManage portfolio helps simplify IT operations across your data center, delivering intelligent, automated management of routine tasks. Combined with unique agent-free management capabilities, the R940xa is simply managed, freeing up time for high profile projects.
The PowerEdge R940xa accelerates applications to deliver real-time decisions. The R940xa combines four CPUs with four GPUs in a powerful 1:1 ratio to drive database acceleration. With up to 6TB of memory and four-socket performance, the R940xa delivers consistent and fast response times. Scale on-premises capacity to offset rising cloud fees and security risks.
8 Bay Chassis - 8x2.5" SAS/SATA (HDD/SSD)
Supports the Front USB module and the ODD
32 Bay Chassis - 24x2.5" SAS/SATA (HDD/SSD) + 8x2.5" SAS/SATA (HDD/SSD)
32 Bay Chassis - 20x2.5" SAS/SATA (HDD/SSD) + 4x2.5" Universal SAS/SATA/NVMe (HDD/SSD) + 8x2.5" SAS/SATA (HDD/SSD)
This configuration includes the SAS Expander to support the NVMe front drives
In the 32 Bay configuration, there will be two backplanes installed. The upper and lower backplanes. The Upper Backplane consists of 24-slots. The 24 Bay chassis supports the SAS Expander which supports up to 4 SAS/SATA or NVMe Drives. The lower backplane only supports the "Lower" 8 Bay backplane. Please note that the 8 Bay Upper and Lower Backplanes have very different layouts for the drives.
8 Bay
32 Bay
32 Bay w/ Expander for NVMeThe R940xa has 6 Hotswap Fans and 4 different Air Shrouds. If you install NVDIMMs the battery for the NV will be placed on top of the shrouds.
The cooling fans are integrated into the system to dissipate the heat generated by the functioning of the system. These fans provide cooling for the processors, expansion cards, and memory modules
The cooling fan assembly houses the array of cooling fans. A failure in the server’s cooling system can result in the server overheating and may lead to damage

Each CPU has 12 memory DIMM slots for a total of 48 DIMMS. Those DIMMs are organized into six different channels so there are 2 DIMMs per channel. For best performance, all memory channels should be populated with the same number of DIMMs, either 6 or 12 DIMMs per CPU.
For more information on all the Rules and Restrictions with each memory type. Please read the manuals below.

RDIMM/LRDIMM
(Registered DIMM)
(Load Reduced DIMM)
Registered DIMM - Provides for higher capacity options and advanced RAS features. It is the most commonly used DIMM type, and offers the best mix of frequency, capacity, and rank structure choices
Load Reduced DIMM - Provides maximum capacity beyond that of an RDIMM but at a higher power consumption. Uses a buffer to reduce memory loading to a single load on all DDR signals, allowing for greater density.
RDIMMs and LRDIMMs must not be mixed.
DCPMM
(Intel Optane DC persistent memory)
Provides a large memory capacity at an affordable price. Any application can take advantage of DCPMM in Memory Mode with a compatible operating system. Unlock more performance and persistency when using an application that supports App Direct Mode. DCPMM is used along with RDIMMs or LRDIMMs, and a maximum number of 6 DCPMMs can be used per CPU. This persistent memory technology does not require a battery.
DCPMM comes in 3 different memory sizes, 128 GB, 256 GB, and 512 GB.
PMems' cannot be mixed with other PMems capacities or NVDIMMs
PMems of different capacities are not allowed.
1600W and 2400 W PSU Must be installed
PMem does not support GPU configurations.
NVDIMM-N
(Non-Volatile DIMM)
Non-Volatile DIMM - Provides a persistent memory solution with NAND and DRAM that maintains data in power loss, system crash, or normal shutdown. This solution requires a battery as a power source for an AC loss condition. It can be used along with RDIMMs.
NVDIMM-Ns or RDIMMs must not be mixed with LRDIMMs.
Maximum of 12 NVDIMM-Ns can be installed in a system.
The system has up to 12 PCIe slots. Depending on the riser you install you can have either x8 or x16 PCIe width slots. If you plan to install any kind of GPU you will need the x16 Risers installed. But, if you install any of the single-wide FPGA you can install the x8 PCIe riser.

| Number of processors | NVMe | GPU | Riser size | Slot size | Slot quantity | Available slots | Height | Length |
|---|---|---|---|---|---|---|---|---|
| 4 | NA | GPU Ready/ Double wide accelerators capable | X16 PCIe Riser 1 | x16 | 2 | 2, 4 | FH | FL |
| x8 | 1 | 5 | FH | HL | ||||
| X16 PCIe Riser 2 | x16 | 2 | 9, 11 | FH | FL | |||
| x8 | 1 | 12 | FH | HL | ||||
| NA | Non-GPU/ Single wide FPGA | X8 PCIe Riser 1 | x8 | 5 | 1, 2, 3, 4, 5 | FH | HL | |
| X8 PCIe Riser 2 | x8 | 5 | 8, 9, 10, 11, 12 | FH | HL | |||
| Front NVMe | GPU Ready/ Double wide accelerators capable | X16 PCIe Riser 1 | x16 | 2 | 2, 4 | FH | FL | |
| X16 PCIe Riser 2 | x16 | 2 | 9, 11 | FH | FL | |||
| Front NVMe | Non-GPU/ Single wide FPGA | X8 PCIe Riser 1 | x8 | 4 | 1, 2, 3, 4 | FH | HL | |
| X8 PCIe Riser 2 | x8 | 4 | 8, 9, 10, 11 | FH | HL | |||
| 2 | NA | GPU Ready/ Double wide accelerators capable | X16 PCIe Riser 1 | x16 | 1 | 4 | FH | FL |
| X16 PCIe Riser 2 | x16 | 1 | 11 | FH | FL | |||
| NA | Non-GPU/ Single wide FPGA | X8 PCIe Riser 1 | x8 | 2 | 3, 4 | FH | HL | |
| X8 PCIe Riser 2 | x8 | 2 | 10, 11 | FH | HL |
With only 12 PCIe slots available it's nice to know you have a dedicated slot for the rNDC which expands the networking ability of the server. You can upgrade go from 1GbE to 10GbE or both at the same time with a mixed flexlom.
The right riser will be placed above this rNDC. If no riser is installed you must install the blanks appropriately.

The Dell PowerEdge R940xa Rack server supports PERC 9, 10, and 11 storage controllers. It also has three different Boot options. Boss, IDSDM, and an Internal USB drive. The IDSDM is used more for loading Hypervisors for the virtualization Workloads.
The BOSS adapter holds two M.2 SATA drives and loads the OS for the entire system. Another option that does the same thing is the internal USB Stick.
An optional USB memory key installed inside your system can be used as a boot device, security key, or mass storage device. To boot from the USB memory key, configure the USB memory key with a boot image and then specify the USB memory key in the boot sequence in System Setup
The IDSDM holds two Micro SD in Raid 1 with a vFlash 16GB drive in the back of the module. This IDSDM is not going to load the main OS but is perfect for loading the hypervisor in the virtualization workloads.
| Performance Level | Controller & Description |
|---|---|
| Entry | S140 (SATA, NVMe) SW RAID SATA |
| Value | HBA330 (Internal), 12Gbps SAS HBA (External) Memory: None x8 12 Gb SAS x8 PCIe 3.0, H330, H350, 12Gbps SAS HBA (External), HBA355e (External) Memory: None x8 12 Gb SAS x8 PCIe 3.0, HBA350i |
| Value Performance | H730P Memory: 2GB, NV 72-bit, 866MHz x8 12 Gb SAS x8 PCIe 3.0 , H750 |
| Premium Performance | H740P, 8x8 ROC Memory: 4/8GB, NV 72-bit |
Please view the PCI section above this. The PCIe section will explain the slot available with each riser configuration.
There are a few rules for GPU and FPGA with this system. To read all the rules please read the manuals below.

The Dell PowerEdge R940xa holds 4 Hot-Swap Power Supply Units (PSU). When identical PSUs are installed, power supply redundancy (2+2 – with redundancy or 4+0 – without redundancy) is configured in system BIOS. In redundant mode, power is supplied to the system equally from both PSUs when Hot Spare is disabled. When Hot Spare is enabled, one of the PSUs is put into the sleep mode when system utilization is low in order to maximize efficiency.
| AC (Platinum) | AC (Titanium) | DC | Mix Mode/HVDC (China and Japan only) |
|---|---|---|---|
| 1100 W | 1600 W | 2000 W | 2400 W |
| 750 W | 1600 W | 2600 W | |
| 1100 W | Mix Mode: 750 W (for China only) | ||
| 1100 W |
Height: 174.3 mm (6.87”)
Width: 441.16 mm (17.37”)
Depth: 835.9mm (32.9”)
Weight: 56.0kg (123.5lbs.)
*Dimensions do not include bezel
(List below serves as quick links to each section)