PACT Unveils The eXtreme Processor Platform

October 13, 2000

NEWS BRIEFS

San Jose, CALIF. — PACT Corporation, a fabless semiconductor and intellectual property company based in Munich, unveiled the eXtreme Processor Platform (XPP) for advanced, processing-intensive high-bandwidth applications. Based on an innovative architecture, the eXtreme Processor enables designers to easily use massively parallel processing to reach unprecedented performance and bandwidth levels.

The first eXtreme Processor, the XPU128, has demonstrated performance that is up to 50 times greater than conventional sequential processors and 20 times higher than digital signal processors (DSPs), which are basically sequential processors fine-tuned for signal processing. The first XPP device achieves sustainable peak performance in excess of 50 Giga Operations per second (GigaOps), making it the world’s most powerful 32-bit processor. “The XPP can be an enabler for a number of new digital signal processing (DSP) applications that were not practical until now. For example, it will likely be a serious market contender for the next generation of base station smart antennas, an application that many feel will explode with the emergence of third-generation cellular,” said Will Strauss, president of Forward Concepts, a market research firm in Tempe, Arizona. “But PACT’s XPP is not limited to the DSP market, since it is a new class of processing engine with unrivaled computing performance that is ideally suited for many other computing-intensive applications, as well.”

The scalable architecture of the XPP enables broad applicability to a wide range of applications. For applications where cost or power consumption is paramount, PACT provides intellectual property (IP) cores that are easily integrated into standard cell and system-on-a-chip devices. XPP IP cores are ideal for the next generation of multifunctional high bandwidth devices, including Internet appliances, PDAs, and mobile telephones for which extreme processing power enables higher levels of functionality and enhanced user interfaces.

For applications that require maximum processing power, PACT supplies eXtreme Processing Unit (XPU) devices. XPP components are targeted at midrange applications such as 3G cellular base stations and voice over IP (VoIP), where they help deliver higher call quality and the ability to handle a much larger number of calls than previously thought to be possible. In addition, cascaded arrays of XPU devices supply the processing performance and bandwidth demanded for extreme applications such as the next generation of supercomputers.

Conventional processors, such as RISC or CISC, use a sequential model to process data. These processors store operands in registers and fetch instructions from memory; then move the operands into the processing unit where the processor performs calculations; and then send results into another register. This model is efficient for sequential applications where different operations are applied to each data word.

However, the sequential model is very inefficient for high-speed broadband applications such as mobile communications, which apply the same algorithm to each data word in a multi-channel data stream. The processor overhead – the process of moving each word into a central processing unit, fetching an instruction, and moving the word to a result register – creates a huge bottleneck that severely limits system performance and bandwidth.

To break the bottleneck, XPP’s reconfigurable parallel data flow processor sends high-speed data streams through an array of processing elements at full speed. The processor performs calculations during each clock cycle. The results then flow to the next processing element. This process continues until all calculations specified in the algorithm have been completed and all of the processed data has flowed out of the XPP.

In addition to raw broadband processing power, the XPP platform offers design methodologies that enable designers to easily and efficiently use high bandwidth parallel processing. A major breakthrough implemented in the XPP platform is the Wave Reconfiguration technology that transparently configures processing elements as data flows through the array.

PACT’s Wave Reconfiguration model is ideal for emerging multi-service broadband networks that use packetized data to support voice, audio and video. For example, if the first packet flowing through the network contains voice data, the XPP is configured for the voice-processing algorithm. The voice packet streams through the array and is processed at full speed. As the next packet enters the array, a token attached to the front of the packet identifies it as a video packet. This token rides just in front of the wave of video data and triggers reconfiguration of each processing element one clock cycle before the data arrives. The result is a wave of reconfiguration that programs the processing elements with the video algorithm at precisely the right time as data flows through the array.

Wave Reconfiguration allows programmers to easily make tradeoffs between performance and density and to minimize power consumption. For example, conditional branch execution can be implemented by programming each branch into the array for maximum speed or using Wave Reconfiguration to configure processing elements as branch evaluations are completed.

For the control portion of applications, Wave Reconfiguration enables the XPP to instantly configure resources to efficiently process sequential operations eliminating the need for a separate sequential processor. For power sensitive applications, Wave Reconfiguration turns off clocks for unconfigured elements and processing elements dynamically power themselves down when data streams are not active.

The XPP contains many additional features that free the designer from chip-level details and make it simple to use. Automatic DataFlow Synchronization ensures that all data inputs are valid before nodes are calculated eliminating the need for lengthy timing verification and optimization. A configurable internal event network enables control of program flow using flags like conventional processors.

The XPP uses a flow-graph style design methodology familiar to microprocessor and DSP design engineers. PACT’s Native Mapping Language (NML) allows users to easily express arithmetic constructs targeted to XPP processing elements. Tools that enable even higher levels of abstraction are in development including C compilers and a flow-graph style language called LELA, developed for PACT by Professor Niklaus Wirth, the inventor of PASCAL.

PACT’s first device, the XPU128, contains an array of 128 parallel processing elements implemented in advanced reprogrammable technology. Each element performs high-speed, 32-bit signed and unsigned operations that execute in a single clock cycle.

The XPU128’s I/O are organized into eight modules, each containing dual 32-bit high-bandwidth channels. The device contains 32 independent 1K-byte embedded SRAMs for high-speed access to source data, destination data, and look-up table contents. These internal SRAMs can be configured as FIFO buffers to support burst-oriented applications. Multiple interfaces to external RAMs provide high distributed memory bandwidth.

A point-to-point internal terabit dataflow network connects XPU resources without the bottlenecks of a bus-oriented architecture. This network makes one-to-many and many-to-one connections between neighboring or distant XPU resources anywhere on the device. A separate second configurable event network transports control flow information. And a third network provides transparent and background reconfiguration capabilities for not only the processing elements, but also for the array’s interconnect. The XPU128 offers high levels of processing power and bandwidth:

— 51.2 GigaOps (32-bit)

— 12.8 GigaMACs (32-bit)

— 3.2 Gigabyte/s external memory bandwidth

— 12.8 Gigabyte/s internal memory bandwidth

— 6.4 Gigabyte/s sustained I/O bandwidth

— 0.4 Terabyte/s internal data bandwidth

The XPU’s modular and truly scalable architecture enables its performance to scale much more efficiently than conventional processors. PACT’s product roadmaps show devices with over 400 GigaOps per second in 2002 and over the PetaOps (1,000,000 GigaOps) level within the next decade.

“Vendors of classical processors rely primarily on higher clock frequencies to achieve incremental performance gains,” said Martin Vorbach, PACT’s co-founder, chief technology officer and the inventor of the XPP. “XPP achieves a second dimension of performance gains by scaling not only with clock frequency, but also with the number of processing elements. Utilizing advanced process technology, we can soon put over 1000 processing elements on a single chip.”

XPP intellectual property cores and PACT’s development system with NML are available for license today. The LELA programming language will be available in Q1 2001. XPU128 devices will be available in production quantities in Q1 2001.

PACT is a fabless semiconductor and intellectual property vendor developing extreme performance processor platforms. The company’s patented breakthrough technology results in a new class of processor, which delivers extreme bandwidth and processing performance for a wide range of applications including portable communication devices, next generation base stations and supercomputers. The company was founded by Marcel Kreutler and Martin Vorbach in Germany in 1996. PACT’s eXtreme Processor Platform, XPP is covered by over 20 international patents in both processor architecture and reconfigurability. PACT currently has offices in Munich, San Jose, New York, Atlanta and Tokyo. The company can be reached at http://www.pactcorp.com .

============================================================

Subscribe to HPCwire's Weekly Update!

Be the most informed person in the room! Stay ahead of the tech trends with industry updates delivered to you every week!

MLPerf Inference 4.0 Results Showcase GenAI; Nvidia Still Dominates

March 28, 2024

There were no startling surprises in the latest MLPerf Inference benchmark (4.0) results released yesterday. Two new workloads — Llama 2 and Stable Diffusion XL — were added to the benchmark suite as MLPerf continues Read more…

Q&A with Nvidia’s Chief of DGX Systems on the DGX-GB200 Rack-scale System

March 27, 2024

Pictures of Nvidia's new flagship mega-server, the DGX GB200, on the GTC show floor got favorable reactions on social media for the sheer amount of computing power it brings to artificial intelligence.  Nvidia's DGX Read more…

Call for Participation in Workshop on Potential NSF CISE Quantum Initiative

March 26, 2024

Editor’s Note: Next month there will be a workshop to discuss what a quantum initiative led by NSF’s Computer, Information Science and Engineering (CISE) directorate could entail. The details are posted below in a Ca Read more…

Waseda U. Researchers Reports New Quantum Algorithm for Speeding Optimization

March 25, 2024

Optimization problems cover a wide range of applications and are often cited as good candidates for quantum computing. However, the execution time for constrained combinatorial optimization applications on quantum device Read more…

NVLink: Faster Interconnects and Switches to Help Relieve Data Bottlenecks

March 25, 2024

Nvidia’s new Blackwell architecture may have stolen the show this week at the GPU Technology Conference in San Jose, California. But an emerging bottleneck at the network layer threatens to make bigger and brawnier pro Read more…

Who is David Blackwell?

March 22, 2024

During GTC24, co-founder and president of NVIDIA Jensen Huang unveiled the Blackwell GPU. This GPU itself is heavily optimized for AI work, boasting 192GB of HBM3E memory as well as the the ability to train 1 trillion pa Read more…

MLPerf Inference 4.0 Results Showcase GenAI; Nvidia Still Dominates

March 28, 2024

There were no startling surprises in the latest MLPerf Inference benchmark (4.0) results released yesterday. Two new workloads — Llama 2 and Stable Diffusion Read more…

Q&A with Nvidia’s Chief of DGX Systems on the DGX-GB200 Rack-scale System

March 27, 2024

Pictures of Nvidia's new flagship mega-server, the DGX GB200, on the GTC show floor got favorable reactions on social media for the sheer amount of computing po Read more…

NVLink: Faster Interconnects and Switches to Help Relieve Data Bottlenecks

March 25, 2024

Nvidia’s new Blackwell architecture may have stolen the show this week at the GPU Technology Conference in San Jose, California. But an emerging bottleneck at Read more…

Who is David Blackwell?

March 22, 2024

During GTC24, co-founder and president of NVIDIA Jensen Huang unveiled the Blackwell GPU. This GPU itself is heavily optimized for AI work, boasting 192GB of HB Read more…

Nvidia Looks to Accelerate GenAI Adoption with NIM

March 19, 2024

Today at the GPU Technology Conference, Nvidia launched a new offering aimed at helping customers quickly deploy their generative AI applications in a secure, s Read more…

The Generative AI Future Is Now, Nvidia’s Huang Says

March 19, 2024

We are in the early days of a transformative shift in how business gets done thanks to the advent of generative AI, according to Nvidia CEO and cofounder Jensen Read more…

Nvidia’s New Blackwell GPU Can Train AI Models with Trillions of Parameters

March 18, 2024

Nvidia's latest and fastest GPU, codenamed Blackwell, is here and will underpin the company's AI plans this year. The chip offers performance improvements from Read more…

Nvidia Showcases Quantum Cloud, Expanding Quantum Portfolio at GTC24

March 18, 2024

Nvidia’s barrage of quantum news at GTC24 this week includes new products, signature collaborations, and a new Nvidia Quantum Cloud for quantum developers. Wh Read more…

Alibaba Shuts Down its Quantum Computing Effort

November 30, 2023

In case you missed it, China’s e-commerce giant Alibaba has shut down its quantum computing research effort. It’s not entirely clear what drove the change. Read more…

Nvidia H100: Are 550,000 GPUs Enough for This Year?

August 17, 2023

The GPU Squeeze continues to place a premium on Nvidia H100 GPUs. In a recent Financial Times article, Nvidia reports that it expects to ship 550,000 of its lat Read more…

Shutterstock 1285747942

AMD’s Horsepower-packed MI300X GPU Beats Nvidia’s Upcoming H200

December 7, 2023

AMD and Nvidia are locked in an AI performance battle – much like the gaming GPU performance clash the companies have waged for decades. AMD has claimed it Read more…

DoD Takes a Long View of Quantum Computing

December 19, 2023

Given the large sums tied to expensive weapon systems – think $100-million-plus per F-35 fighter – it’s easy to forget the U.S. Department of Defense is a Read more…

Synopsys Eats Ansys: Does HPC Get Indigestion?

February 8, 2024

Recently, it was announced that Synopsys is buying HPC tool developer Ansys. Started in Pittsburgh, Pa., in 1970 as Swanson Analysis Systems, Inc. (SASI) by John Swanson (and eventually renamed), Ansys serves the CAE (Computer Aided Engineering)/multiphysics engineering simulation market. Read more…

Choosing the Right GPU for LLM Inference and Training

December 11, 2023

Accelerating the training and inference processes of deep learning models is crucial for unleashing their true potential and NVIDIA GPUs have emerged as a game- Read more…

Intel’s Server and PC Chip Development Will Blur After 2025

January 15, 2024

Intel's dealing with much more than chip rivals breathing down its neck; it is simultaneously integrating a bevy of new technologies such as chiplets, artificia Read more…

Baidu Exits Quantum, Closely Following Alibaba’s Earlier Move

January 5, 2024

Reuters reported this week that Baidu, China’s giant e-commerce and services provider, is exiting the quantum computing development arena. Reuters reported � Read more…

Leading Solution Providers

Contributors

Comparing NVIDIA A100 and NVIDIA L40S: Which GPU is Ideal for AI and Graphics-Intensive Workloads?

October 30, 2023

With long lead times for the NVIDIA H100 and A100 GPUs, many organizations are looking at the new NVIDIA L40S GPU, which it’s a new GPU optimized for AI and g Read more…

Shutterstock 1179408610

Google Addresses the Mysteries of Its Hypercomputer 

December 28, 2023

When Google launched its Hypercomputer earlier this month (December 2023), the first reaction was, "Say what?" It turns out that the Hypercomputer is Google's t Read more…

AMD MI3000A

How AMD May Get Across the CUDA Moat

October 5, 2023

When discussing GenAI, the term "GPU" almost always enters the conversation and the topic often moves toward performance and access. Interestingly, the word "GPU" is assumed to mean "Nvidia" products. (As an aside, the popular Nvidia hardware used in GenAI are not technically... Read more…

Shutterstock 1606064203

Meta’s Zuckerberg Puts Its AI Future in the Hands of 600,000 GPUs

January 25, 2024

In under two minutes, Meta's CEO, Mark Zuckerberg, laid out the company's AI plans, which included a plan to build an artificial intelligence system with the eq Read more…

Google Introduces ‘Hypercomputer’ to Its AI Infrastructure

December 11, 2023

Google ran out of monikers to describe its new AI system released on December 7. Supercomputer perhaps wasn't an apt description, so it settled on Hypercomputer Read more…

China Is All In on a RISC-V Future

January 8, 2024

The state of RISC-V in China was discussed in a recent report released by the Jamestown Foundation, a Washington, D.C.-based think tank. The report, entitled "E Read more…

Intel Won’t Have a Xeon Max Chip with New Emerald Rapids CPU

December 14, 2023

As expected, Intel officially announced its 5th generation Xeon server chips codenamed Emerald Rapids at an event in New York City, where the focus was really o Read more…

IBM Quantum Summit: Two New QPUs, Upgraded Qiskit, 10-year Roadmap and More

December 4, 2023

IBM kicks off its annual Quantum Summit today and will announce a broad range of advances including its much-anticipated 1121-qubit Condor QPU, a smaller 133-qu Read more…

  • arrow
  • Click Here for More Headlines
  • arrow
HPCwire