|
HyperPower
Blade Cluster Configuration |
|
Blades |
40
blade nodes based on GreenBlade
Systems configured with GPU
expansion blades |
|
Processors |
Intel®
Xeon® 5500/5600 Series
CPU and NVIDIA® Tesla™
M2050 GPU cards |
CPU/GPU
Cores |
Up
to 480 CPU / 35,840 GPU cores |
|
Memory
Capacity |
Up
to 3.84TB CPU ECC DDR3 memory
(using 8GB DIMMs)
Up to 240GB GPU GDDR5 memory |
Performance |
Total
of 46.83TF double-precision
(CPU and GPU total performance)
Up to 5.63TF CPU performance
(with X5670 CPUs)
Up to 82.4TF GPU single-precision
performance
Up to 41.2TF GPU double-precision
performance |
Storage
Capacity |
Up
to 80 internal 2.5" HDDs,
equal to 40TB of local storage |
|
Networking |
Standard:
Gigabit Ethernet
Optional: Integrated QDR InfiniBand |
Rack
Configuration |
Standard
42U/19" rack, 2U rack
space available for switches |
|
RAS
Features |
Improved
reliability, availability
and serviceability with:
Shared N+1 power and cooling
system design
Shared cooling (3+3) design
Hot-swappable blade nodes,
power supplies, and cooling
fans
Integrated platform management
module monitors and controls
individual blades, platform
power supplies and platform
fans |
|
Rack
Level Power Consumption |
16kW
to 32kW, depending on system
configuration |
|
Management |
Integrated
IPMI 2.0 remote management |
|
Certification |
Intel
Cluster Ready certified |
Options |
Additional
storage
Cluster management software
Linux or Windows operating
system
Cluster software compilers
development tools
Installation and onsite
maintenance services |
|
Services
and Support |
Consulting, pre-integration
and testing with pre- and
post-sales support with
2-year warranty on parts
and labor |
