Java Solaris Communities About Sun How to Buy United States Worldwide

Throughput Computing

FAQ




  1. What is Throughput Computing?
  2. What is the benefit of Throughput Computing to customers?
  3. Why is Throughput Computing necessary?
  4. How does Throughput Computing work?
  5. What is chip multithreading (CMT)? How does it differ from chip multiprocessing (CMP) and simultaneous multithreading (SMT)?
  6. Why is Throughput Computing a reality now?
  7. What is Sun's competitive advantage to deliver Throughput Computing?
  8. What applications will work best for CMT-based designs?
  9. What is Sun's strategy to manufacture CMT-based processors?
  10. Will customers and ISVs have to change their software applications in order to take advantage of CMT technology?
  11. When are CMT based systems expected to roll out?
  12. What is Niagara?
  13. Is Niagara still SPARC?
  14. Do I have to change my application to run on Niagara?
  15. When will Niagara-based systems become available?

1. What is Throughput Computing?

Throughput Computing is the underlying strategy of Sun's new family of UltraSPARC processors designed to significantly increase real-world application performance while helping to cut the cost and complexity of network computing. These processors maximize throughput - the aggregate amount of work done - for network computing workloads by incorporating the technology of chip multithreading (CMT). CMT integrates the power of symmetric multiprocessing (SMP) on to a single chip, allowing a single processor to execute several software threads simultaneously.

Increasingly, this approach is being adopted by other processor vendors, as evidenced by the announcement of first generation dual-core processors from AMD and Intel, while Sun's UltraSPARC road map delivers second-generation CMT in early 2006, delivering both multiple cores and multiple threads per core.

Back to top


2. What is the benefit of Throughput Computing to customers?

Throughput Computing is designed to offer customers significantly improved processing efficiency by allowing multi-threaded applications to run much faster and enabling a single processor to run multiple applications using Solaris containers. Being able to purchase systems that are so much more efficient can help enable customers to experience a variety of savings, including lower cost of acquisition as well as dramatically reduced space, power and maintenance requirements. In addition, having fewer to manage, simpler, single-processor systems can also help improve reliability and availability. Because CMT-based servers will be so much more efficient, our customers will be able to focus their IT budgets on building competitive advantage through the deployment of new business services, and the enhancement of existing services.

Back to top


3. Why is Throughput Computing necessary?

The demand for network services continues to rise, in some cases exponentially, while IT budgets and resources remain limited, coupled with contraints in data center power and HVAC. Companies need to do more - in some cases increasing throughput by orders of magnitude instead of by percentages--while still keeping within budget. Sun's Throughput Computing strategy is designed to deliver orders of magnitude more throughput while reducing the cost and complexity of network computing.

Back to top


4. How does Throughput Computing work?

Moore's law has traditionally seen processor performance to double every two years as a result of the doubling of the number of the transistors that can be crammed onto a slice of silicon. Unfortunately, memory speeds have only been doubling every six years, creating an ever-increasing gap. As a result, today's processors are stalled as much as 75 percent of the time while they wait for memory to fetch data. With Throughput Computing, when a thread must wait for the memory, the affected core simply starts processing another thread. CMT processors transcend the clock-speed focus of traditional chip design to deliver systems with significant price/performance improvements and higher reliability while lowering the cost of computing by allowing fewer systems to handle the same workload.

Back to top


5. What is chip multithreading (CMT)? How does it differ from chip multiprocessing (CMP) and simultaneous multithreading (SMT)?

Today's traditional single-core processors can only process one thread at a time, spending a majority of time waiting for data from memory. In sharp contrast, chip multithreading (CMT) refers to a processor's ability to process multiple software threads. A CMT processor could implement this multithreaded capability using a variety of methods, such as (i) having multiple cores on a single chip (CMP), (ii) executing multiple threads on a single core (SMT), or (iii) combination of both CMP and SMT.

Back to top


6.Why is Throughput Computing a reality now?

Moores Law states that the number of transistors that can be delivered with a single processor doubles every 2 years. This increased transistor budget enables Sun's processor design team to create processors that more accurately reflect the way commercial applications are written.

Most commercial applications are multi-threaded, meaning each user or process transacted by the application is represented by a software thread. Enabling multiple threads to be executed simultaneously is the goal of CMT and is made possible as the number of transistors available increases, allowing multiple cores with multiple threads to be created in a single processor.

This is complimented by the multi-threaded capability of the Solaris 10 Operating System, and the maturity of Sun's software development tools.

Only through delivery of this whole ecosystem, can throughput computing be made a reality.

Back to top


7.What is Sun's competitive advantage to deliver Throughput Computing?

Sun is uniquely positioned to deliver the most sophisticated implementations of Throughput Computing:

  • Sun's multithreaded SMP based systems are already providing optimal solutions for the thread-rich network computing environments today, utilizing the industry-leading threading capabilities in the Solaris OS. Because our CMT processors are based on our UltraSPARC product line and Solaris OS, this enables our customers to seamlessly transition to CMT-processor based systems without disrupting their SW model.
  • CMT processors also need lots of memory to support such computational capacity. 64-bit addressability like the UltraSPARC processor offers a significant benefit over the addressing limitations of a 32 bit processor. Only 64-bit architectures like that of the UltraSPARC processor can fully realize the potential of Throughput Computing.
  • For such a heavily threaded environment, one needs an operating system capable of handling and scheduling multiple threads. Bar none, Solaris has the best threading model in the commercial operating system market today.
  • A processor strategy focused on designing for CMT from the ground-up. Retrofitting CMT technology into an existing design focused on single-thread performance will yield a suboptimal solution.

Back to top


8. What applications will work best for CMT-based designs?

CMT based processors are designed for the thread-rich network computing environment. These processors maximize application throughput by processing multiple threads simultaneously on a single chip. Vertically-scaled, multi-threading applications benefit from having access to more threads, while horizontally-scaled, single -threading (or limited multi-threading) applications benefit from being able to run multiple instances in Solaris Containers on a single machine. As a result, almost all commercial applications such as database, web-based services, transaction processing, application serving, data-mining, ERP and CRM etc. potentially benefit from CMT.

Back to top


9. What is Sun's strategy to manufacture CMT-based processors?

As it has over the past 17 years, Sun continues to partner very closely with Texas Instruments (TI) to manufacture its industry-leading UltraSPARC processors. While Sun focuses on its core competency of microprocessor design, it relies on TI for its world-class semiconductor manufacturing. This fabless model also helps Sun to avoid the huge expense of building and maintaining a leading-edge fab every few years. The Sun/TI partnership is a non-competitive, mutually-beneficial relationship that truly works!

Back to top


10. Will customers and ISVs have to change their software applications in order to take advantage of CMT technology?

No. Backward binary-compatibility with the SPARC V9 ISA will be maintained on all UltraSPARC processors on the road map to ensure that the software written for a previous generation of SPARC processor will run seamlessly on the future generation processors.

Back to top


11. When are CMT based systems expected to roll out?

UltraSPARC IV processors, capable of handling two threads at a time, are available now to augment our current UltraSPARC III and UltraSPARC IIIi product lines. In CY2006, Sun is expected to introduce systems based on "Niagara," Sun's new, second-generation CMT processor, capable of handling up to 32 threads at a time.

Back to top


12. What is Niagara?

Niagara is the first of our second generation of CMT processors, and the first to use what we term "Radical CMT." With 8 cores, each running 4 threads for a total of 32 threads at a time, Niagara is a sharp departure from traditional processor design and a radical step in a new direction, delivering several times the throughput of current processor designs with lower power consumption and less heat dissipation.

Back to top


13. Is Niagara still SPARC?

Yes! Niagara is based on and entirely compatible with the SPARC v9 specification, and will run Solaris 10. This means that our customers' investments in SPARC and Solaris are protected by this ground breaking new technology. This is one of the biggest advantages Niagara has in the market � it will run the applications that are already available for the SPARC/Solaris platform unchanged!

Back to top


14. Do I have to change my application to run on Niagara?

No - Niagara is a fully-compliant SPARC v9 processor and, when it is released, will be branded with Sun's UltraSPARC name. Like all UltraSPARC processors, Niagara offers full binary compatibility.

If you have been thinking about making your application more highly-threaded, however, the multi-threading capabilities of Niagara, the enhanced threading tools offered in Sun Studio 10, and the Dynamic Tracing capabilities of Solaris 10 offer a perfect reason to re-examing the threading capabilities of your code.

Back to top


15. When will Niagara-based systems become available?

We are planning to ship Niagara-based systems in early 2006. Watch sun.com for news and announcements about this new processor and the systems it will enable!

Back to top

 

Give your business a boost