Compaq DS10 User Manual

Compaq AlphaServer DS10 Systems
Technical Summary
1 System Overview Features and Benefits
2 Third-Generation Alpha Chip
Chip Operation Alpha 21264 Features
3 Architecture
System Block Diagram
4 System Board
Component and Connector L ocations
5 Memory
Memory Options Memory Configura tion Memory Performance Considerations
6 System I/O
Block Diagram of I/O Control PCI Slots I/O Ports
7 System Control
Control Panel
Storage
RAID (Redundant Array of Independent Disks)
8 Server Management
Operational Management Platform Management Error Reporting
Security Reliability and Availability Features
Processor Features Memory Features I/O Features System Features
9 Clustering
PCI to Memory Channel Interconnect Operating System Support
Performance
Sources of Performance Information Information for Compaq Partners
10 Service and Support
Hardware Warranty Software Warranty
11 Compaq AlphaServer DS10 System Diagrams
12 System Features at a Glance
13 Physical Characteristics
Compaq AlphaServer DS10 Systems
The Compaq AlphaServer DS10 system is an entry-level system offering 64-bit computing with the Alpha processor. This single processor system is ideal for ISP/Internet appli­cations, telecom applications, software development, and replicated site applications. The 3-U system box can be mounted in a cabinet or used on a desktop.
Compaq AlphaServer products use the 64-bit Alpha RISC architecture that supports multiple operating systems: Tru64 UNIX, OpenVMS, and Linux.
For more information on Compaq AlphaServer DS10 systems, see
http://www.compaq.com/alphaserver
System Overview
The Compaq AlphaServer DS10 systems are available with the 600 MHz Alpha chip, the 21264A (marketed as 600 MHz but it actually runs at 616 MHz). Memory begins at 256 Mbytes and can be increased to 2 Gbytes. Second-level cache is 2 Mbytes. The switch-based system interconnect exploits the full potential of the Alpha chip.
The system meas ures 17 x 19 x 5 inches (3U) a nd can be placed on a desktop or rackmounted in a choice of three M­series cabinets, along with additional disks. In the largest cabinet (79 inches) up to 12 systems can be installed, or if more storage is desi red, there can be up to 6 StorageWorks shelves and up to 8 systems. A rackmount slide kit is included with each system.
Six drive bays are available for storage devices. Each system includes a CD-ROM and floppy; a 5.25-inch r emovable media device can be added. With four 1-inch disks, t here can be 218 GB of storage in the syste m box.
There are four full-length PCI slots: three 64-bit slots and one 32-bit slot. Integr ated on the system board are two 10/100 Mbit fast Ethernet controllers, an IDE controller, two serial ports, one parallel port, remote management console, and keyboard and mouse ports.
Systems can be purchased with the Tru64 UNIX or OpenVMS operating systems insta lled. Or they can be purchased without any operating system, allowing customers to inst all Linux.
Features and Benefits
Performance
The Alpha chip, the world’s fastest microprocessor, is offered with a switch-base d interconnect that supports one 600 MHz processor and up to 2 Gbytes of memory. This switch-based system provides a memory bandwidth of up to 1.3 Gbytes/sec (peak) using a 128-bit memory bus running at 77 M Hz. The peak I/O bandwidth is 250 Mbyes/sec.
Multiple Operating Systems
No other server offers the flexibility of running so many operating systems: Tru64 UNIX, OpenVMS, and L inux. It’s ideal as a development tool as well as a Web server or file server, or for remote applica tions or E-comme rce applications.
Package and Price
All the features of AlphaServers are now available in a package that goes anyw here at a price attractive to everyone. Start with one and then add to your base as your needs demand. Use the DS10 as a dedicated system or set up a cluster.
DS10 Workstations
The 600 MHz system is also offered as a workstation and is called the AlphaStation DS10.
Third-Generation Alpha Chip
The third genera tion of the Alpha microprocessor, the Alpha 21264, is a superscalar, superpipeline d i mplementation of the Alpha architecture. The first offering of this chip, which was manufactured using the CMOS-6 process, was known as EV6 and now the EV67 (21264A) chip is available, which uses the CMOS-7 process. Over 15.2 million transistors are on one die.
In our discussion here, t he Alpha 21264 designation applies to the EV6 and the EV67 chips, unl ess we need to distinguish betw een the two. Designed f or performance, the Alpha 21264 achie ves this goal by carefully st udi ed and simulated ar chitectural and circuit analysis. The 21264 memory system also enable s t he high per­formance levels. On-chip and off-chip caches provide for very low latency data access, which allows for very high bandwidth data access. The 21264A 2-Mbyte off-chip cache runs at 205 MHz.
Internal to each chip is a 64-Kbyte instruction cache (I-cache) and a 64-Kbyte data cache (D-cache).
I-cache. 64 Kbytes, two-way set -associative, vir tually
addressed cache with 64-byte blocks
D-cache. 64 Kbytes, two-way set -associative, virtually
indexed, physically tagged, writeback cache with 64-byte blocks
Chip Operation
Several key design choices were made i n the chip architect ure to maximize performance: Four instructions are fetched each cycle, and then how those instructions are handled boosts the speed of execution. Register renaming assigns a unique storage location with each write reference to a register, avoiding register dependencies that can be a potential bottleneck to processor performance.
Another design fea t ure, out-of-or der execution, permits instructions to execute in an order different from the order that the instructions are fetched. In effect, instructions execute as soon as possible. This allows for faster e xe cution since critical path computations ar e started and completed as soon as possible.
In addition, the Alpha 21264 employs speculative execution to maximize performance. It speculatively fetches and executes instructions even though it may not know immediately whether the instruction will be on the final execution path. This is particularly use ful, for instance , when the 21264 predicts branch directions and speculatively executes down the predicted path. The sophisticated bra nch prediction in the 21264 coupled with the specul ative and dynamic e xecution extracts the most instruction parallelism from applications.
For more inform ation about the chip, see
http://www.compaq.com/alphaserver/download/ev6chip.pdf
Alpha 21264 Features
Out-of-order instruction execution
Large (64 Kbyte) on-chip data and instruction caches
Improved branch prediction through intuitive execution
Register renaming
Increased bandwidth for high-speed access to second-level
cache and system memory
Motion video instructions
Square root and divide instructions
All instructions are 32 bits long and have a regular
instruction format
Floating-point unit, supports DIGITAL and IEEE floating-
point data types
80 integer registers, 64 bits wide
72 floating-point re gisters, 64 bits wide
Architecture
The traditional bus interconnect has been replaced by a switch­based interconnect system. With a bus design, the processors, memory, and I/ O modules share the bus. As the number of bus users increases, the transactions interfere with one another, increasing latency and decreasing aggregate bandwidt h. However, with a switch-based system t here is no degradation in performance as the number of CPUs, memory, and I/O users increase. Although t he users increase, t he speed is maintained.
With a switch-based, or point-to-point interconnect, the performance remains constant, even though the number of transactions multiplies. The switched system inte rconnect uses a set of complex chips that route the traffic over multiple paths.
The chipset consists of one C-chip, one P-chip, a nd t wo D­chips.
C-chip. Provides the command interface from the CPU.
D-chips. Provide the data path for the CPU, main
memory, and I/O.
P-chip. Provides the interface to the PCI bus.
System Block Diagram
This chipset, similar to those used in the AlphaServer DS20 and ES40 systems, supports up to one CPU and up to 2 Gbytes memory. Interleaving occurs when at least two memory arrays are used.
The PAD bus, the interface between the P-chip and the D­chips, is 32 data bits with 4 check bits. The 128-bit memory bus supports two memory arrays, yielding a 1.3 Gbyte/sec system bandwidth. Tr ansactions are ECC prot ected. Upon the receipt of data, the receiver checks for data integrity and corrects any errors.
C-chip
CPU
B-cache
Command, Address, and Control lines for each Memory Array
Control lines for D-chips
CAP Bus
CPU Data Bus
P-chip
64 bit PCI
PAD Bus
2 D-chips
Memory Data Bus
1 or 2
Memory
Arrays
PKW
1400B-99
Loading...
+ 11 hidden pages