NASHUA, N.H., 12 August 2005. The latest innovation from NextCom LLC is Flextreme, a mobile computing architecture incorporating AMD Quad Opteron processing and PCI Express.
Flextreme Opteron is a portable server and mobile workstation, boosting performance with dual 2.2-Gigahertz AMD Opteron processors, up to 16 Gigabytes DDRAM, up to 200-Gigabyte removable IDE or 146-Gigabyte Ultra320 SCSI storage, dual Gigabit Ethernet, optional Ultra160 SCSI, and dual head support, all in a briefcase-size package weighing under 22 pounds.
As the pace of the world speeds up, computer professionals and users have less and less time at their disposal, and all too often must perform several tasks all at once. The complexity of tasks that computers are called on to execute has also increased to the point where it often takes more computing power than is in even the fastest single-core processors to meet the needs of a fast-paced company's IT department.
For non-IT professionals also, the problems that computers are now called on to solve have become complicated enough so that increasingly, you want the computing power of a rackmount or tower-style server at your fingertips when you are away from your office. But both rackmount and tower computers are too cumbersome and heavy to be carried anywhere. Even using a fast connection to a server that has the proper computing horsepower, you can access what you need only with a smaller, far less powerful machine, and, in some circumstances, even fast connections may become unstable.
The answer is dual-core architecture, the solution that IT managers, computer professionals and serious power users of all stripes are looking for. In this new architecture, one processor contains two full execution cores, which means that two CPUs share the same space that in single-core architecture is occupied by one CPU.
In addition, dual-core architecture is optimal for multi-threaded software. The technique of multi-threading divides up programs so that independent tasks can run concurrently, yet separately. Each task is assigned to one of the two CPUs. Single-core computer architecture, in which a computer has only one CPU, can achieve this only by time-slicing, which means the CPU stops processing one thread in order to process another thread, switching back and forth between the two.
NextCom's Flextreme brand represents a major step forward in computing power. Besides the single-core processor model that is already available, Flextreme now comes with a dual-core or quad (two dual-cores ganged together) Opteron processor from Advanced Micro Devices (AMD). This means that both its CPUs can process two separate threads literally at the same instant, or both can switch back and forth between two or more threads, doubling the multi-threading capability of a single-core processor.
Coupled with the dual-core architecture is PCI Express X16, a bus standard that is faster and more efficient than its predecessor, PCI. In addition, Flextreme's briefcase-sized form factor gives it complete mobility so you can carry this processing power with you wherever you go.
Dual-core vs. single-core
The dual-core AMD Opteron processor inside Flextreme is based on AMD64 technology, which enables simultaneous 32- and 64-bit computing with no loss of performance.
In fact, the dual-core AMD Opteron uses less power and generates less heat than recent single-core processors, making it ideal for use in a mobile computer. Whereas a single-core processor has only one CPU, the dual-core architecture puts two CPUs side-by-side on one die, so that they share the same space that one CPU takes up in single-core processing. The two CPUs are linked independently of each other to the bus that connects them to the main memory on the motherboard, and each CPU has its own cache.
All this allows the user to multi-task on one machine at a significantly faster rate than is possible on a single-core machine. In theory, a dual-core processor would be twice as powerful as a single-core processor, but in everyday use, dual-core generally gives you about one-and-a-half times the speed of single-core. Speed is further improved by AMD's use of Direct Connect Architecture, which connects the processors straight to the memory controller and I/O.
How AMD's Direct-Connect Architecture works
The two crucial components of the Direct-Connect Architecture used by AMD Opteron technology are the System Request Queue and the "crossbar," so named after the design of old phone switching devices. The System Request Queue acts as an interface between the CPU cores and the crossbar. It prioritizes the requests of both CPU cores for contact with the crossbar switch, so that access to the system bus is managed more effectively, resulting in a smooth use of system resources.
The crossbar connects the CPUs to both the HyperTransport technology links and the Integrated DDR Memory Controller, which handles I/O to the outside world.
Special features of the Dual-Core Opteron architecture
The HyperTransport technology links and the Integrated DDR Memory Controller improve system performance in the following ways:
§ HyperTransport technology links. HyperTransport technology is a high-speed data transmission channel that minimizes latency and increases the speed of communication between integrated circuits up to 48 times that of previous technologies. It reduces how many buses are needed in a system, which in turn reduces I/O bottlenecks and allows the modern, faster microprocessors to make more efficient use of system memory in high-end multiprocessor systems. It also provides high peak bandwidth to each processor. The impact on peripheral drivers is minimal.
§ Integrated DDR Memory Controller. An integrated memory controller runs at the same core frequency as the processor. This has the effect of increasing the bandwidth that is directly available to the processor, and cutting down on latency. In an AMD Opteron, each processor has its own memory controller, so the memory bandwidth is scaled.
The increased ability to multi-task means that the user can run 32- and 64-bit applications as wished, with no loss in performance. Up to 3 coherent Hyper-Transport Technology links make for up to 24GB/s peak bandwidth per processor. The 256 TB of address space results in optimized performance. This is especially useful when it becomes necessary to hold large datasets, or several datasets at once, in memory. Lower heat output is a result of the 90nm Silicon on Insulator process technology used in the processor.
PCI Express X16
The peripheral component interconnect express (PCIe) routes data traffic through a switch mechanism instead of through the old, slower shared bus used by the previous generation of PCI; this speeds up the data flow between components and the main memory. The more "lanes" of communication are available, the faster the switch can "talk to" the peripheral devices connected to it; the X16 version of PCIe enables 16 bytes of data to flow in both directions at the same time. The combination of PCIe X16 and either one or two dual-core processors makes Flextreme the fastest, most powerful small computer on the market today.
Small size translates to portability
The dual-core Opteron processor will be found most often in rackmounted or tower systems, which are large and heavy enough to be confined to a computer room or a desk. Measuring 11.1 x 16.6 x 5.6 inches and weighing only 17 to 23 pounds, Flextreme eliminates the need to rely on a connection to a server to access files and perform complicated computing tasks. Its ruggedized, small form factor makes it completely portable, so that whether you stay at your workstation or go far afield, its 256 TB of address space and the power of dual-core Opteron processing will be ready to accompany you whenever and wherever you need it.
Flextreme's three Gigabit Ethernet ports give it plenty of connectivity. The two removable RAID 0/1 SATA hard disk drives run at twice the read transaction rate single disks, and data is written to both of them at the same time. For more information, see www.nextcomputing.com.