HPE Apollo pc40 GPU Server
The HPE Apollo pc40 Server is a 1U dual socket server featuring up to four NVIDIA Tesla GPUs and based on the Intel Xeon Processor Scalable Family. The HPE Apollo pc40 Server is an optimized, industry-standard server for deep learning and high performance computing (HPC) workloads that uses the PCIe form factor for greater choice in GPU selection.
Enjoy quicker and easier deployments with systems configured, integrated, and tested by Hewlett Packard Enterprise. Using industry-standard IPMI 2.0 system management and the HPE Performance Software Suite, the server is easy to use and manage. The HPE Apollo pc40 Server delivers cost-effective deep learning and HPC performance with high compute density (up to four GPUs per 1U rack space), helping to turn data into insights and insights into actions.
Flexible Entry-level Server Configurations Tailored to HPC and Deep Learning
The HPE Apollo pc40 Server features up to four NVIDIA Tesla PCIe graphic processing units (GPUs) to accelerate mixed-application HPC computing as well as deep learning workloads. The PCIe form factor can allow for a greater selection of accelerators and GPUs than alternate form factors.
Optimize the server for required workloads by choosing from the available CPUs in the Intel Xeon Processor Scalable Family, up to twelve 2666MHz DDR4 DIMMs, optional network adapters, and up to two SFF hard drives or solid state drives.
Fully integrated and factory tested clusters are delivered with a complete cluster management software stack to simplify deployment and management, and can be custom configured for perfect right-sizing.
Multiple network topologies are supported in cluster builds: Fat Tree, Hypercube, and Enhanced Hypercube for increased flexibility.
The server is readily integrated into a standard cluster environment.
Right-sized Compute for Infrastructure Efficiency
The HPE Apollo pc40 Server is air-cooled for easy integration into standard server environments.
The system is designed for energy efficiency with dual 2000W Titanium Power Supplies to drive these powerful GPUs.
Select the right-sized network topology for the target cost model. For example, Fat Tree can be a strong, general purpose fabric which is effective for small systems, but can be expensive for large systems. Choose based on needs.
Integrated IPMI 2.0 enables standards-based systems management.
Expertise and Resources for Operational Efficiency
The HPE Apollo pc40 Server alongside HPE Pointnext services offerings, deliver expertise in HPC, data management, visualization, systems management, and high performance data analysis (HPDA).
Expect responsive global support.
Enables custom solutions, as one size does not fit all requirements.
Committed account and technical resource teams.
Know your Server - CTO Configuration Support
Need help with the configuration? Contact us today!
HPE Apollo pc40 Base Knowledge.
This 1U GPU-intense rack server has 2 Sockets and 12 DIMMs total along with 4 GPU slots and 2 hot-swap SSD drives. In the rear, it has 2 PSUs and the option to install 2 Low Profile PCIe cards. Next to the two low-profile PCI slots is the 4th DW GPU slot. IF this 4th GPU is not installed in the Rear it frees up 2 Full-Height PCIe slots. The 3 other GPUs have internal slots right behind the front bezel and 7 fans for optimal airflow.
12 DIMM Slots Available
Type DDR4 Load Reduced (LRDIMM), or Registered (RDIMM)
DIMM Slots Available 12 DIMM slots available
Maximum (LRDIMM) 1.5 TB (12x 128 GB)
Maximum Internal Storage
| Capacity |
Quantity |
| HPE 1TB SATA 7.2K SFF RW 512n DS HDD |
2x 1 TB |
| HPE 2TB SATA 7.2K SFF RW 512e DS HDD |
2x 2 TB |
| HPE 240GB SATA MU SFF RW DS SSD |
2x 240 GB |
| HPE 480GB SATA MU SFF RW DS SSD |
2x 480 GB |
| HPE 960GB SATA MU SFF RW DS SSD |
2x 960 GB |
| HPE 1.92TB SATA MU SFF RW DS SSD |
2x 1.92 TB |
| HPE 3.84TB SATA RI SFF ST DS SSD |
2x 3.8 TB |
HPE Apollo pc40 PCI/GPU Risers
There are no risers with this system. But it does come with 2 PCIe Low-Profile slots. If the 4th GPU is not installed in the server, that slot provides 2 Full-Height PCIe Slots.
GPU Modes.
Consider selecting 16GB memory DIMMs for Apollo pc40 with 4 NVIDIA GPUs. NVidia design guidelines state a requirement for at least 64GB main memory per CPU to ensure optimal application performance. The GPU slots are as follows, 3 Internal slots right behind the 7 Fans and Bezel. While the 4th GPU sits in one double-wide GPU or if the GPU is not installed it becomes two full-height PCIe Gen3 x16 slots.
Networking Abilities
The system comes with 2x 10Gb NIC RJ-45 ports. You do have the option to install other network cards in the 2 low-profile slots in the rear. Choose between 1 or 2 ports and 10Gb speeds of 100Gb with InfiniBand Adapters.
HPE Apollo pc40Cooling
In front of the 3 Internal GPUs sits 7 1U fans providing all the air flow you'll need with 4 GPUs installed.
HPE Apollopc40 PSU
The Apollo uses Two 2000W Titanium level power supplies to support the 4 GPUs with 12 DIMMs.
HPE Apollo pc40 Server Management
UEFI - Configure and boot your servers securely with industry-standard Unified Extensible
Firmware Interface (UEFI).
HPE Insight Cluster Management Utility- HPE Insight Cluster Management Utility (CMU) is a Hewlett Packard Enterprise- licensed and Hewlett Packard Enterprise-supported suite of tools that are used for lifecycle management of hyperscale clusters of Linux systems. CMU includes software for the centralized provisioning, management, and monitoring of nodes. CMU makes the administration of clusters user-friendly, efficient, and effective.
Intelligent Provisioning – Hassle-free server and OS provisioning for 1 or few servers with Intelligent Provisioning.
