IBM HX5 User Manual
Optimized for virtualization and database applications with maximum
memory and compute capacity in a blade
Product Guide
August 2010
Product Overview
CONTENTS
Product Overview 1
Selling Features 2
Key Features 6
Key Options 13
HX5 Images 15
HX5 Specifications 16
The Bottom Line 18
Server Comparison 20
For More Information 21
Legal Information 21
IBM BladeCenter HX5
Scalable 2-to-4-socket blade server optimized for
virtualization/consolidation, database, and ERP
Suggested uses: Front-end and mid-tier applications requiring high performance (CPU, memory
or I/O), enterprise-class availability, and extreme flexibility and power efficiency.
IBM® has been designing and implementing chipsets under the X-Architecture® name since
2001. eX5 technology represents the fifth generation of products based on the same design
principle IBM began with in 1997: to offer Intel® Xeon® processor-based systems that are
expandable, offer “big iron” reliability, availability, and serviceability (RAS) features, with
extremely competitive price/performance.
The eX5 technology is primarily designed around three major workloads: database servers,
server consolidation using virtualization services, and Enterprise Resource Planning (application
and database) servers.
If you’re using industry-standard servers to run business critical applications, the systems that
run these applications need the type of technology designed into IBM’s eX5 technology systems.
The eX5 chipset represents a $100M+ investment in designing a flagship offering that can
harness the power of 4-socket-and-up 64-bit x86 (x64) Xeon processors. The eX5 family
includes a scalable perf ormance blade server with the ability to scale from a single-wide blade
(30mm) to a double-wide blade (60mm). Maybe you’d like to start out with a 2-socket blade and
possibly add sockets later, if your needs change. Or perhaps you need more than a 2-socket
blade, but don’t want to get locked into a monolithic 4-socket blade—again, in case your
requirements change. With IBM, you can start at 2 sockets and grow to 4 if needed. And if you
require a variety of 2- and 4-socket servers in your data center, you only have to qualify one
server for all these workloads. This can save much time and effort and speed deployment. You
can also save money on software licensing by virtualizing a 4-processor server into many VMs,
rather than using multiple 2-processor servers. Huge amounts of memory also enable more or
larger VMs, and larger databases (especially databases stored entirely in memory).
Reducing an entire server into a little over .5U of rack space (i.e., up to 14 servers in 9U) does
not mean trading away features and capabilities for smaller size. The IBM BladeCenter® HX5
blade server offers features comparable to many 1U rack-optimized full-featured servers, and
then some: The HX5 supports up to two of the latest high-performance 8-core, 6-core, or 4-core
Intel Xeon 7500 and 6500 Series processors. The Xeon processors are designed with up to
24MB of shared cache and leading-edge memory performance (up to 1066MHz using MAX5,
depending on processor model) to help provide the computing power you require to match your
business needs and growth. In addition, the HX5 supports up to 128B of registered double data
rate III (DDR3) ECC (Error Checking and Correcting) memory in 16 DIMM slots, with multiple
levels of IBM Active Memory™ protection, f or the highest levels of performance and availability.
Active Memory protection includes IBM Memory ProteXion™, IBM Chipkill™ memory, memory
scrubbing, memory rank sparing, and memory mirroring.
Because business requirements change and a 2-socket server that meets those needs today
may not meet them in the future, the HX5 was designed to be upgradeable to meet the diverse
needs of multiple workloads. For compute-intensive workloads, it can be configured as a 2wide blade server with up to 4 processors / 32 cores, 32 DIMMs (256GB), 4 PCIe cards, 16 I/O
ports, and 4 solid-state drives (SSDs). Conversely, for memory-intensive workloads, the
HX5 can be configured as a 2-wide server consisting of 1 blade server and 1 MAX5 memory
expansion blades, with up to 40 DIMMs (320GB), or as a 4-wide server comprising 2 blade
servers and 2 MAX5 memory expansion blades, with up to 4 processors / 32 cores, 80 DIMMs
(640GB), 4 PCIe cards, 16 I/O ports, and 4 SSDs. IBM FlexNode partitioning allows a
physical 4-blade configuration to be remotely reconfigured by software into two logical 2-socket
servers, as needs change or for daily peak and off-peak workloads.
The HX5 supports VMware ESXi preloaded on a standard USB flash drive. It operates in a
diskless configuration, offers a smaller memory footprint, extremely high performance, and
stronger security, making getting a system up and running in a virtualized environment faster and
easier than ever before.
Please see the Legal Information section for important notices and information.
1.
Optimized for virtualization and database applications with maximum
memory and compute capacity in a blade
IBM’s eX5 technology-based systems are the ideal solution for scale-up database-serving
applications on Microsoft® Windows® with Microsoft SQL Server® or IBM DB2®, as well as Linux®
with Oracle or DB2. Database hosting demands ultimate server reliability features, and once
installed, they tend to grow and grow, requiring ever greater levels of availability. eX5 servers
provide exactly that degree to availability.
Another strong application area for the eX5-based systems is enterprise server consolidation
activities workloads, including SAP and Oracle. eX5 systems can offer considerable savings over
UNIX deployments, using our certified solution stacks on either Windows or Linux.. Larger
servers need more processor, memory and I/O resources, which make maximum use of any
applicable virtual machine software licensing fees and deliver superior system utilization levels.
The name of the game in consolidation activities is to deploy the fewest new servers possible
and help IT staff manage more images with the same or fewer overall people.
An integrated dual-port Gigabit Ethernet1 controller is standard, providing high-speed data
transfers and offering TOE (TCP Offload Engine) support, load-balancing and failover
capabilities. Via optional expansion cards, each blade can also connect to additional Ethernet,
10Gb Ethernet, SAS, Fibre Channel, iSCSI, InfiniBand™, and other high-speed communication
switches housed in the chassis. This blade is designed with power management capability to
provide the maximum uptime possible for your systems. In extended thermal conditions or power
brownouts, rather than shut down completely, or fail, the HX5 automatically reduces the
processor frequency to maintain acceptable thermal and power levels.
All HX5 models offer impressive features at an equally impressive price, including up to two 1.8-
inch fixed solid-state drives (SSDs) with RAID-0/1 support, and one optional internal USB
flash drive (for embedded hypervisor]. Additional direct-attach storage is available via the
BladeCenter S chassis. Moreover, the HX5 is optimized for diskless operation, offering each
blade server access to essentially unlimited external storage capacity via Fibre Channel, SAS, or
iSCSI.
A single BladeCenter H chassis supports up to 14 hot-swappable 30mm-wide HX5 blades in
only 9U (BladeCenter H) of rack space, or up to 12 in the 12U BladeCenter HT high-speed
telecommunications chassis. In addition to the blade servers, these chassis also hold up to 10
(BladeCenter H) or 8 switches/bridges (BladeCenter HT) internally. The BladeCenter S,
designed for SMB and mid-market customers, takes integration and affordability to a new level,
combining up to 12 hot-swap SAS/SATA HDDs (with optional SAS card) and up to 6 blade
servers and 4 switches. Not only can this save significant data center space (and therefore the
cost of floor space and rack hardware) compared to 1U servers, it also consolidates
switches/bridges and cables for reduced complexity and lower cabling costs, and it allows clients
to manage everything in the solution as one. Using a BladeCenter H chassis, up to 56 HX5
servers (112 processors/896 cores) can be installed in one industry-standard 42U rack; but the
value of BladeCenter extends far beyond high density data center environments.
The various BladeCenter chassis are designed to monitor environmental conditions in the
chassis and each blade and send alerts to the administrator. Advanced standard features, such
as Active Memory, Predictive Failure Analysis, light path diagnostics, hot-swap redundant
SSDs and HDDs, power supplies and blower modules with Calibrated Vectored Cooling™;
IPMI 2.0 support, including highly secure remote power control; text-console redirect over
LAN, next-generation BIOS (UEFI), an Advanced Management Module (upgradeable with a
redundant AMM), IBM Systems Director management software including IBM Systems
Director Active Energy Manager™. Remote Deployment Manager and IBM ServerGuide™
help maintain system availability with increased uptime.
If you need scalable, highly manageable, high-performance virtualization in a space- or powerconstrained environment, the HX5 is the ideal system.
Selling Features
1
Actual data transfer speed will vary and is often less than the maximum possible. Gigabit Ethernet transfer speed requires support on both
system and server, and appropriate network infrastructure.
2
Based on Intel measurements.
Please see the Legal Information section for important notices and information.
Price/Performance
• There is an HX5 model to fit all budgets. The HX5 offers a choice of high-performance 4-, 6-, and 8-
core Xeon processors with dual integrated memory controllers, clock rates of 1.86GHz to 2.26GHz,
and 12MB, 18MB or 24MB of integrated Level 3 cache. Xeon 7500 series processors offer up to 3X
better performance2 than the previous-generation 7400 series processors and up to 10X better
performance than the single-core processors of a few years ago that you may still be using.
• Up to 128GB of registered DDR3 ECC memory per blade operates at 800MHz to 1066MHz
(depending on the system configuration), for high performance and wide memory bandwidth.
• By scaling to 2 blade servers, the HX5 can support up to 256GB of memory—enough for even the
most demanding virtualization, database, or ERP needs. If even that’s not enough memory, using
two blade servers and two MAX5 memory expansion blades, the HX5 can scale to 640GB.
2.
Optimized for virtualization and database applications with maximum
memory and compute capacity in a blade
• Embedded virtualization (optional on all models) offers extremely high performance, enhanced
security, and a zero-gigabyte HDD footprint. (In other words, no mechanical HDD to fail.)
• Solid state drives (SDDs) use only 2W of energy per drive, vs. 9-10W for 2.5-inch HDDs. This is as
much as 80% less energy than a HDD would use (with a corresponding reduction in heat output).
• The extremely high degree of integration in the various BladeCenter chassis reduces the need for
server components, replacing numerous fans, KVM and Ethernet cables, power supplies, external
switches and other components with fewer shared hot-swap/redundant components in the
BladeCenter chassis itself. This integration also can greatly reduce the amount of power
consumed and heat produced, relative to an equivalent number of 1U servers. This can
significantly reduce a data center power bill. The reduced data center footprint can also save on
infrastructure cost.
• The midplanes used in all chassis provide high-speed blade-to-blade, blade-to-switch-module
and module-to-module communications internally as well as externally. The midplanes used in
the BladeCenter H and BladeCenter HT chassis provide four 10Gb data channels to each blade,
and supports high-speed switch modules, including 4X InfiniBand and 10Gb Ethernet.
• The various BladeCenter chassis use ultrahigh efficiency power supplies. Most industry-standard
servers use power supplies that are between 70-75% efficient at converting power from AC wall
current to the DC power used inside servers. BladeCenter power modules are up to 92% efficient.
This helps save even more money, as more of the power input you are paying for is used for
processing, rather than released into the data center as waste heat that requires even more energy
to cool.
• BladeCenter design also reduces the number of parts required to run the system. Sharing fans,
systems management, and optical media means fewer parts to buy and maintain, and fewer items
that can fail and bring the overall solution down.
Flexibility
The HX5 has the ability to grow with your application requirements, thanks to:
• Up to two multi-core Xeon processors (up to 16 cores) per HX5 blade server; up to 2 blades (4
processors, 32 cores) per server.
• The ability to grow from a single-wide 2-socket server blade to a double-wide 4-socket server
configuration, optimized for compute-intensive workloads with up to 4 processors/32 cores,
256GB of memory, 4 PCIe cards, 16 I/O ports, and 4 SSDs), or for memory-intensive workloads
with up to 2 server blades and two MAX5 memory expansion blades (4 processors/32 cores, 640GB
of memory, 4 PCIe cards, 16 I/O ports, and 4 SSDs).
• A choice of processor speeds (1.86 or 2.0GHz), and shared L3 cache sizes (12MB, 18MB, or
24MB).
• Up to 128GB of system memory in 16 DIMM slots. Using multiple blade servers and MAX5 memory
expansion blades, the HX5 can support up to 256GB (2 server blades), 320GB (1 HX5 and 1
MAX5), or 640GB (2 HX5 servers and 2 MAX5 blades).
• Up to two internal hot-swap 1.8-inch solid state drives, and access to terabytes of external storage
via the BladeCenter S chassis or IBM System Storage™ SAN and NAS storage devices. SSDs
consume only 10-20% of the energy required by 2.5-inch HDDs.
• Two Gigabit Ethernet ports standard, plus optional 2-port or 4-port expansion cards or a
BladeCenter PCI Express I/O Expansion Unit 3. One HX5 model includes a 10Gb expansion card
(optional in the other models).
In addition, the various BladeCenter chassis offer a high degree of flexibility:
• They support configurations that are 30mm or 60mm wide, with a variety of I/O options, depending
on need.
• When installed in a BladeCenter H high-speed switch bay, the optional Multi-Switch Interconnect
Module doubles the number of Gigabit Ethernet and Fibre Channel connections to every blade in
the chassis (up to 8 or 12 ports, depending on the blade server).
• Xeon processor-based HX5 blades can be used in the same chassis as Intel processor-based
HC10, HS12, HS20, HS21, HS21 XM, HS22, HS22V, and HS40 blades; AMD Opteron processor-
based LS20, LS21, LS22, LS41 and LS42 blades; IBM PowerPC® processor-based JS20, JS21,
and JS22 blades; POWER6® processor-based JS12, JS23, and JS43 Express blades; POWER7®
processor-based PS700, PS701, and PS702 blades; Cell Broadband Engine™ processor-based
QS21; and IBM PowerXCell™ processor-based QS22 blades. Depending on the blade servers
used, the various BladeCenter chassis support Microsoft Windows, Linux, Novell Netware, IBM
AIX® and Sun Solaris 10 operating systems in the same chassis.
Please see the Legal Information section for important notices and information.
3.
Optimized for virtualization and database applications with maximum
memory and compute capacity in a blade
• Most HS/LS/JS/QS blade servers ever released by IBM are supported in every BladeCenter chassis
ever released, going back to 20023. Every switch module released by IBM is equally compatible.
(Ask HP and Dell how far back their compatibility goes.)
• A blade server has access to as many as 10 communication switches/bridges in a BladeCenter
H or 8 in a BladeCenter HT chassis. (Up to 4 switches in a BladeCenter E, S, or T chassis.) And the
switches can be Ethernet, iSCSI, SAS, InfiniBand, Fibre Channel, or anything else designed and
ServerProven® for BladeCenter use. Switches, bridges and interface cards are currently available
from such vendors as Brocade, Cisco, Intel, McData, Nortel, QLogic, Cisco Topspin and others, in
addition to IBM.
Manageability
• The HX5 blade server includes an Integrated Management Module (IMM) to monitor server
availability, perform Predictive Failure Analysis, etc., and trigger IBM Systems Director alerts. The
IMM performs the functions of both the Baseboard Management Controller (BMC) of earlier
systems, and the Remote Supervisor Adapter II, as well as remote presence/cKVM.
• Each BladeCenter chassis includes an Advanced Management Module to provide additional
systems management capabilities, including Web-based out-of-band control; virtual floppy and CDROM support; latest OS failure screen capture; LDAP and SSL support; and remote redirection of
video, text, keyboard and mouse.
• Integrated industry-standard Unified Extensible Firmware Interface (UEFI) next-generation BIOS.
New capabilities include:
Human readable event logs – no more beep codes
Complete setup solution by allowing adapter configuration function to be moved into UEFI
Complete out-of-band coverage by the Advance Settings Utility to simplify remote setup
• Integrated industry-standard IPMI 2.0 support works with the IMM to alert IBM Systems Director to
anomalous environmental factors, such as voltage and thermal conditions. It also supports highly
secure remote power control.
• Integrated Trusted Platform Module (TPM) provides a highly secure start-up process from power-
on through hand-off to the operating system boot loader. ACPI support is provided to allow ACPIenabled operating systems to access the security features of this module. (TCG V1.2-compliant.)
• IBM Systems Director Active Energy Manager, an IBM-exclusive, is designed to take
advantage of new system power management features, by monitoring actual power usage and
providing power consumption capping features. More accurate power usage data helps with data
center construction planning and the sizing of power and cooling needs, as well as allowing you to
use available power more efficiently.
• The HX5 supports remote presence/concurrent KVM (cKVM) and concurrent media (cMedia)
access by multiple administrators at once, via the IMM.
• IBM Systems Director is included for proactive systems management and works with both the
blade’s internal IMM and the chassis’ management module. It comes with a portfolio of tools,
including IBM Systems Director Active Energy Manager, Service and Support Manager, and
others. In addition, IBM Systems Director offers extended systems management tools for additional
server management and increased availability. When a problem is encountered, IBM Systems
Director can issue administrator alerts via e-mail, pager, and other methods.
3
Some older chassis may require power module and management module upgrades.
Please see the Legal Information section for important notices and information.
Availability and Serviceability
• BladeCenter chassis are designed for operation with greatly reduced potential for single points
of failure. Most aspects of operation, from blade servers to communication modules, to
management modules, to power and blower modules, are hot-swappable. The midplane
connections are redundant and the other features can be made so, when used in pairs.
• Fast PC3-10600 DDR3 ECC memory offers multiple layers of Active Memory protection, including
Chipkill error correction, Memory ProteXion (redundant bit steering), memory scrubbing, memory
rank sparing, and memory mirroring for high availability
• HX5 blade servers support the use of Chipkill-enabled ECC (error checking and correcting)
memory. Chipkill memory can be up to 16X better than standard ECC memory at correcting some
types of memory errors. This can help reduce downtime caused by memory errors.
• Solid-state drives offer up to triple the availability (MTBF rates) of conventional SAS HDDs. This
can lessen the need for redundant drives.
• IPMI 2.0 supports highly secure remote system power on/off using data encryption. This allows an
administrator to restart a server without having to visit it in person, saving travel time and getting the
server back up and running quickly and securely.
4.
Optimized for virtualization and database applications with maximum
memory and compute capacity in a blade
• Environmentally tuned blower modules in the chassis adjust to compensate for changing thermal
characteristics. At the lower speeds they draw less power and suffer less wear. Equally important in
a crowded data center, temperature-controlled blowers produce less ambient noise in the data
center than if they were constantly running at full speed.
• Text and graphics console redirection support allows the administrator to remotely view HX5 text
and graphic messages over serial or LAN connections.
• A standard three-year (parts and labor) limited onsite warranty4 affords you peace of mind and
greater potential investment protection.
Right, Open, Easy, Green
You need to make IT decisions that will drive business success. You face management challenges and
technological complexity such as space constraints, power and cooling limitations, heterogeneous
environments and I/O connectivity issues. IBM brings together the widest choice of compatible chassis,
blade servers, storage and networking offerings and solution providers in the industry to help you build
an open and flexible IT environment. And regardless of the size of your business, you want to be up and
running 24/7. With built-in redundancy, innovative power and cooling and the latest I/O and
management tools, IBM BladeCenter is easy to own—so you can focus on your business demands and
stay ahead of the competition.
The RIGHT choice, tailored to fit your diverse needs:
• It’s flexible and modular. As needs evolve, a one-size-fits-all solution doesn’t work.
– Meet your needs with BladeCenter: everything from a high-performance data center to a small
office with limited IT skills—IBM has you covered
– Get flexibility with 5 compatible chassis and 5 blade types supporting multiple I/O fabrics, all
managed from a common point
• It’s robust and reliable, providing redundancy throughout and the information you need to keep your
business up and running.
– Provide redundancy for no single point of failure with IBM BladeCenter
– Preserve application uptime with IBM Predictive Failure Analysis and light path diagnostics
– Make decisions based on accurate data for quick problem diagnosis with First Failure Data
Capture
OPEN and innovative, for a flexible business foundation:
• It’s comprehensive, providing broad, fast, and reliable networking and storage I/O with BladeCenter
Open Fabric.
– Match your data center needs and the appropriate interconnect using a common management
point, and 5 I/O fabrics to choose from
– Extract the most from your third-party management solutions by utilizing the BladeCenter Open
Fabric Manager
• It’s collaborative, enabling you to harness the power of the industry to deliver innovation that matters.
– Get flexibility from a myriad of solutions created by Blade.org members and industry leaders
that have downloaded our open specification
EASY to deploy, integrate and manage:
• It enables efficient integrated management, which allows you to minimize costs with the tools you
need for effective management.
– Automate OS installation and BIOS updates remotely with IBM Systems Director tools
– Administer your blades at the chassis or rack level with the Advanced Management Module
– Plug into your enterprise management software
• It enable deployment simplicity without tradeoffs by speeding the deployment of new hardware in
minutes rather than days, using BladeCenter Open Fabric Manager
– Get significantly faster deployment of servers and I/O than from rack solutions
– Reduce costly downtime with integrated failover capability
– Manage from a single point of control via the Advanced Management Module
– Use with virtually all IBM switches, blades and chassis
GREEN today for a better tomorrow:
• It offers control via powerful tools that help you optimize your data center infrastructure so you can be
responsive.
4
For terms and conditions or copies of the IBM Statement of Limited Warranty, call 800-772-2227 in the U.S. In Canada call 800-426-2255.
Telephone support may be subject to additional charges. For warranties including onsite labor, a technician is sent after IBM attempts to
resolve the problem remotely. International warranty service is available in any country in which this product is sold.
Please see the Legal Information section for important notices and information.
5.
Optimized for virtualization and database applications with maximum
memory and compute capacity in a blade
– Understand your power requirements with IBM Power Configurator
– Monitor, control and virtualize your power with IBM Systems Director Active Energy Manager
– Reduce data center hot spots with the IBM Rear Door Heat eXchanger
– Optimize and future-proof your data center with IBM Data Center Energy Efficiency services
• Our eco-friendly servers and services can help you be environmentally responsible.
• Become more energy efficient with IBM expertise
Key Features
Multicore Intel Xeon Processors
The HX5 ships with 1 or 2 high-performance Intel Xeon 7500 or 6500 Series processors per
blade. By connecting 2 blades together, the HX5 servers also go a step further by allowing you
to increase the number of sockets from 2 to as many as 4. The choice of processors includes:
• 130W 8-core Xeon processor model X7560 at 2.26GHz, with 64-bit extensions, low power draw per
core (16.25W), 6.4 GTps (gigatransfers per second) QPI speed, 978MHz memory access, dual
integrated memory controllers, 24MB of shared L3 cache, and Intel Turbo Boost and Hyper-
Threading technology; supported in BladeCenter H, HT, and S chassis
• 130W 8-core Xeon processor model X6550* at 2.0GHz, with 64-bit extensions, low power draw per
core (16.25W), 6.4 GTps QPI speed, 978MHz memory access, dual integrated memory controllers,
18MB of shared L3 cache, and Intel Turbo Boost and Hyper-Threading technology; supported in
BladeCenter H, HT, and S chassis
• 95W 8-core Xeon processor model L7555 at 1.86GHz, with 64-bit extensions, extremely low power
draw per core (11.9W), 5.86 GTps QPI speed, 978MHz memory access, dual integrated memory
controllers, 24MB of shared L3 cache, and Intel Turbo Boost and Hyper-Threading technology;
supported in BladeCenter H, HT, and S chassis
• 105W 6-core Xeon processor model E7540 at 2.0GHz, with 64-bit extensions, low power draw per
core (17.5W), 6.4 GTps QPI speed, 978MHz memory access, dual integrated memory controllers,
18MB of shared L3 cache, and Intel Turbo Boost and Hyper-Threading technology; supported in
BladeCenter H, HT, and S chassis
• 105W 6-core Xeon processor model E7530* at 1.86GHz, with 64-bit extensions, low power draw per
core (17.5W), 5.86 GTps QPI speed,978MHz memory access, dual integrated memory controllers,
12MB of shared L3 cache, and Intel Turbo Boost and Hyper-Threading technology; supported in
BladeCenter H, HT, and S chassis
• 95W 4-core Xeon processor model E7520* at 1.86GHz, with 64-bit extensions, reduced power draw
per core (23.75W), 4.8 GTps QPI speed, 800MHz memory access, dual integrated memory
controllers, 18MB of shared L3 cache, and Intel Hyper-Threading technology; supported in
BladeCenter H, HT, and S chassis
• 105W 6-core Xeon processor model E6540* at 2.0GHz, with 64-bit extensions, low power draw per
core (17.7W), 5.86 GTps QPI speed, 1066MHz memory access, dual integrated memory controllers,
18MB of shared L3 cache, and Intel Turbo Boost and Hyper-Threading technology; supported in
BladeCenter H, HT, and S chassis
• 105W 4-core Xeon processor model E6510* at 1.73GHz, with 64-bit extensions, reduced power
draw per core (26.25W), 4.8 GTps QPI speed, 800MHz memory access, dual integrated memory
controllers, 12MB of shared L3 cache (available via CTO) , and Intel Hyper-Threading technology;
supported in BladeCenter H, HT, and S chassis
* This processor is limited to single-chassis configurations.
Note: Because of the integrated memory controllers the former front-side bus (FSB) no longer
exists.
With the Xeon 7500 series processors, Intel has diverged from its traditional Symmetric
Multiprocessing (SMP) architecture to a Non-Uniform Memory Access (NUMA) architecture. The
Xeon 7500 processors are connected through three serial coherency links called QuickPath
Interconnect (QPI). QPI is capable of up to 6.4GTps (gigatransfers per second), depending on
the processor model.
The 4-core Xeon processors contain 4 complete processor cores and 8 threads. Each
processor contains one L3 cache, shared by all the cores. The cores appear to software as
separate physical processors. Four-core processors can offer more than double the performance
of a same-speed 2-core Xeon processor (depending on workload). Similarly, 8-core Xeon
processors contain 8 processor cores and 16 threads. They can offer more than double the
performance of four-core Xeon processors (again, depending on workload). Likewise, the six-
core processors contain six processor cores.
Each processor includes two integrated memory controllers, to reduce memory bottlenecks and
improve performance.
Intelligent Power Capability powers individual processor elements on and off as needed, to
reduce power draw.
Please see the Legal Information section for important notices and information.
6.
Optimized for virtualization and database applications with maximum
Mill
Mill
Mill
Mill
QPI (x4)
Mill
Mill
Mill
Mill
memory and compute capacity in a blade
Intel’s Virtualization Technology (VT) integrates hardware-level virtualization hooks that allow
operating system vendors to better utilize the hardware for virtualization workloads.
DDR3 Registered Memory with Active Memory Protection
The HX5 uses registered double data rate III (DDR3) VLP (very-low-profile) DIMMs and provides
Active Memory features, including advanced Chipkill memory protection, for up to 16X better
error correction than standard ECC memory. In addition to offering triple the memory
bandwidth of DDR2 or fully-buffered memory, DDR3 memory also uses less energy. DDR2
memory already offered up to 37% lower energy use than fully buffered memory. Now, a
generation later, DDR3 memory is even more efficient, using 10-15% less energy than DDR2
memory.
The HX5 supports up to 128GB of memory in 16 DIMM slots. Redesign in the architecture of the
Xeon 7500 series processors bring radical changes in the way memory works in these servers.
For example, the 7500 series processors integrate 2 memory controllers inside each
processor, resulting in four memory controllers in a two-socket system. Each memory controller
has four memory channels, one per pair of DIMMs. Depending on the type of memory,
population of memory, and processor model, the memory may be clocked at 978MHz or
800MHz.
Redesign in the architecture of the x7500 series processors bring radical changes in the way
memory works in these servers. For example, the Xeon 7500 series processors integrate two
memory controllers inside each processor, resulting in four memory controllers in a twosocket system. Each processor has four memory channels.
Note: If only one processor is installed, only eight DIMM slots (up to 64GB) can be used. There
are two ways to expand memory beyond 8 DIMMs. Adding a second processor not only doubles
the amount of memory available for use, but also doubles the number of memory controllers,
thus doubling the system memory bandwidth. If you add a second processor, but no additional
memory for the second processor, the second processor has to access the memory from the first
processor “remotely,” resulting in longer latencies and lower performance. The latency to access
remote memory is almost 75% higher than local memory access. So, the goal should be to
always populate both processors with memory.
Alternatively, you can add a MAX5 memory expansion blade, containing 24 more DIMM slots
with up to 192GB of memory. This offers up to 32 DIMMs and 256GB of memory with only a
single processor. Adding the second processor allows the use of 8 more DIMM slots inside the
HX5 and a grand total of 40 DIMMs and 320GB of memory in a double-wide (60mm) blade
server. Expanding further, 2 HX5 server blades and 2 MAX5 memory blades can create one 4-
socket, 640GB powerhouse in a 4-wide blade server.
Brook 8
Brook 7
SMI 7
Brook 6
Brook 5
SMI 5
SMI 6
SMI 8
Brook 4
Brook 3
SMI 4
Brook 2
Brook 1
SMI 3
SMI 1
SMI 2
Memory
Controller-2
24MB L3 Cache
Co
re Core Core Core
QPI
Memory
Controller-4
Memory
Controller-3
24MB L3 Cache
Co
re Core Core Core
QPI
Co
re Core Core Core
Notes: DIMMs must be installed in matching pairs. Also, each CPU requires at least 2 DIMMs. It
is important to ensure that all memory channels in each processor are populated. The relative
memory bandwidth decreases as the number of channels populated decreases. This is because
the bandwidth of all the memory channels is utilized to support the capability of the processor.
So, as the channels are decreased, the burden to support the requisite bandwidth is increased
on the remaining channels, causing them to become a bottleneck.
For peak performance:
• Always populate processors with equal amounts of memory to enable a balanced NUMA system
• Always populate both memory channels on each processor with equal memory capacity
• Ensure an even number of ranks are populated per channel
Power guidelines:
• Fewer larger DIMMs (for example 8 x 4GB DIMMs vs. 16 x 2GB DIMMs will generally have lower
power requirements
• x8 DIMMs (x8 data width of rank) will generally draw less power than equivalently sized x4 DIMMs
Please see the Legal Information section for important notices and information.
Memory
Controller-1
Co
re Core Core Core
7.
Loading...