The ExpressBox 3600-AB provides two independent PCIe five slot subsystems that can be managed separately by two different hosts. The EB3600-AB is ideal for Virtual Desktop VDI and other high density GPU applications that demand ‘mission critical’ features such as N+1 redundant power and cooling and high speed PCIe Gen 3 x16 connectivity.
Magma’s ExpressBox 3600-AB provides optimal server grade GPU (and other HPC peripheral) expansion to two independent hosts needing autonomous control of the PCIe devices in either side “A” or “B” of the chassis. Ten total expansion slots are supported by the EB3600-AB with five for either host. Each segment “A” or “B” may be individually powered up or down for individual servicing of either host’s resources. Each segment is independently managed through Magma’s Express I/O manager, which also provides system level alarm monitoring and control though a user GUI and/or SNMP.
The EB3600-AB is optimized for server class applications and supports a variety of server class GPUs. Power and cooling are designed for up to 300W per GPU. Cooling is designed to support passive cooled GPUs. All critical components are hot swappable and designed for RAS applications. The EB3600-AB supports full Gen 3 x16 (128 Gbps) communication between each host and all peripherals connected to the host.
Your privacy is assured. Information about subscribers to MAGMA email newsletters or any form of communication is treated as strictly confidential and will not be sold, shared, or passed on to any other entity. Should you unsubscribe, your details will be erased from our records.
|Interconnect Bandwidth||PCIe Gen 3 x16: 128 Gbits/sec to all peripherals and host link||Enclosure|
|Dimensions||16.8” Wide x 7” High (4U) x 22.125” Deep|
|Weight||50 lbs/22.7kg||Host Connection|
|Host side||Any PCIe based host-equipped computer (preferably Gen 3 x16)|
|Expansion side||All host links are Gen 3 x16 capable and will train down to Gen 2 or 1 as needed|
|Switchless host link adapter design for minimal host/expansion latency|
|1m or 2m cable lengths are supported for host interface||Power Supply|
|Additional Hot swappable 1200W@120V/ 1500W@240V AC power PSU can be ordered as spares.||Environmental|
|Ambient Temperature||0º to 50º C|
|Storage Temperature||-55º to 125º C|
|Relative Humidity||0% to 90% non-condensing||Regulatory Compliance|
|FCC||Class A Verified|
|CE||Certified||Supported Operating Systems|
|30 day||Money back guarantee|
|1 year (extendable)||Return to factory||Package Contents|
It is recommended that systems be configured with a minimum of 8 GB of system memory, with a suggestion of 16 or 32 GB is appropriate for most of today’s applications and operating systems.
GPUs require additional system memory to support memory copy operations, and the system memory should add to the base memory configuration, whatever BAR (Base Address Register) space memory that each GPU requires. An NVIDIA K40, for example, requires 16 GB of BAR1 space and if a base system is configured with 16 GB of system memory, the user should add 16 GB (32 GB total) of system memory PER K40 added through expansion. If the user is configuring 4 K40s, for instance, this would require 64GB additional system memory above and beyond what the base operation system and application requires. For 4 K40s and base system memory of 16GB, this would be 80Gb of total system memory to support the base application as well as the GPU memory requirement.
NOTE: Always check the minimum system requirement of the GPU. This information is available from the vendors / manufacturers website.
The host interface is standard half height/half length. Length: 7 inches (17.78 cm) Height: 2.25 inches (5.72 cm).
No. The host interface card uses low latency retimers (no switch) for cable interfacing, resulting in a lower latency than competitive switch based solutions.
Magma has designed a special Gen 3 cable with improved signaling characteristics over previous generation cables. This x8 or x16 cable interface allows true 0 bit error rate transmission for ultimate performance at Gen 3 signaling rates.
1 meter and 2 meter cable lengths are available as ordering options.
For the ExpressBox 3600-10 and 3600-AB, up to 10 peripherals are supported. For double wide (GPU) usage, up to 9 GPUs are supported. For the EB3600-P, up to 8 peripherals/GPUs are supported (two per partition).
Any PCI Express supported GPU may be used. The ExpressBox 3600 supported NVIDIA Tesla, GeForce and Quadro GPUs and full access to all GPUs is provided at the rear of the chassis. AMD FirePro and Intel Phi co-processors are supported.
The chassis provides N+ 1 redundant cooling fan. 4 fans are supplied, each with 175 cfm.
Although optimized for double wide GPU modules, any PCI Express board may be used within the system. The ExpressBox 3600 support Gen 3, Gen 2 and Gen 1 boards of any data width (x1/x4/x8/x16).
The enclosure is 16.8” (42.7 cm) wide x 22.125 (56.2cm) deep x 7” (17.8 cm) high (4U).
Up to 4 1200/1500W hot swappable AC input PSUs may be configured. For 120V operation, each PSU is specified for 1200W; For 240V operation, each PSU is specificed for up to 1500W. At 120V, this provides a total of 4800W, or 3600W N+1 redundant. For 240V operation, total power is 6000W or 4500W N+1 redundant.
Nine pairs of PCI AUX cables are supplied with the system, with each slot being supported by two 8-pin PCI AUX connectors. Total cable pairs: 18.
Looking for more information to download or print? Product drivers? Below you will find everything from data sheets to warranties. If you still cannot find what you are looking for, please contact Magma directly and we will assist you.
Please submit your contact information to access the following documentation. Submit here.
The documentation below is available without submitting contact information.
If you are having trouble with your product installation please go to our drivers and utilities knowledge base.
If you still can't find what you are looking for, please contact us through our customer support hub.