Intel faces a familiar challenge: do more with less. With compute capacity growing exponentially and chip size expectation shrinking, the new 5500 series delivers on both fronts. This white paper delivers test results that show increased performance and speed along with greater efficiency.
IT@Intel Brief Faster Chip Design with Intel® Intel Information TechnologyComputer Manufacturing Xeon® Processor 5500 Series64-Bit Computing Silicon chip design engineers at Intel face ongoing March 2009 Profile: Intel® Xeon® Processor challenges: integrating more features into ever- 5500 Seriesshrinking silicon chips and bringing products to . Up to 3.77x performance improvement market faster. Increased design complexity and for a single-threaded jobbusiness requirements lead to exponential compute capacity growth and increased cost, as shown in . Up to 4.98x performance improvement Figure 1. To support this growing demand and for a multi-threaded jobbetter manage costs, Intel IT conducts ongoing . Up to 11.39x throughput improvement performance assessments of compute servers for distributed job when all cores are usedbased on Intel® quad-core processors. To test the new Intel® Xeon® processor 5500 series, we recently ran a broad range of industry-leading end-to-end electronic design automation (EDA) applications with more than 50 Intel silicon design workloads.For a single-threaded job, the Intel Xeon processor 5500 series-based server was up to 3.77x faster than a server based on a 64-bit Intel® Xeon® processor (3.6 GHz) with a single-core. For a multi-threaded job, the performance gain was 4.98x. For a distributed job when all cores are used, the quad-core processor ran 4x as many EDA application jobs as the single-core processor, with a throughput improvement of up to 11.39x. This means we can replace up to 11 single-core processor-based servers with one quad-core processor-based server for some EDA applications and consequently lower our operational costs. And with significantly higher throughput, we can accelerate design cycles to achieve faster time to market.CostDesign Factors+ . Shorter product development time. Nano process technology Resulting . Reduced headcount and cost computingdemand . Increased pre-silicon veri?cation. High degree of design automationComputing . Global design team and collaborationdemand
Design complexity growth
- - Time +Figure 1. Design team challenges. Design factors, such as those listed at left, push the computing demand higher, as shown by the dotted line. Design and IT Environment front- and back-end EDA applications in the following areas:ChallengesTo keep up with ever-expanding computing needs . Simulationat Intel, we need to constantly increase the number . Synthesisof servers and workstations-however, just adding . Physical verification: design rule check (DRC) and more systems is not the solution. It is imperative layout versus schematic (LVS)that new systems deliver features specific to EDA computing. Our design methodologies use several . Silicon prototypingEDA applications that support both 64- and 32-bit . Optical proximity correction (OPC)processing capability. Each distinct feature of the computing platform-such as microarchitecture; Test Methodologyfrequency and cache size; number of cores; and memory type, capacity, and access rate-offers We ran tests on dual-socket servers based on Intel® unique benefits to each EDA application and to Xeon® processor X5570. This processor is based on design workload data. a new platform featuring the Intel® 5520 chipset; Figure 2 illustrates some of the enhancements that New systems must also make use of existing data boost EDA application performance.center investments by delivering adaptable utilization workloads, increasing end user productivity, and We used more than 50 EDA workloads that are operating within the available footprint and cooling vital to the design and manufacturing of Intel® space. processors, chipsets, and communication silicon products. These workloads require more than 250 To support chip design, Intel IT conducts ongoing hours of runtime on a 64-bit Intel Xeon processor-performance assessments to quantify and analyze based server (single-core). We used both 64- and the potential performance benefits of introducing 32-bit single and multi-threaded applications during new Intel quad-core processors into our EDA the end-to-end EDA performance analyses.computing environment. We run industry-leading
Intel® Xeon® Intel Xeon Intel® Xeon® Intel Xeon Intel® Xeon® Intel XeonProcessor Processor Processor Processor Processor 5570 Processor 5570(Quad- or Dual-Core) (Quad- or Dual-Core) (Quad-Core) (Quad-Core)6.4 GB/s 21-25 GB/s 25.6 GB/s per Intel® QuickPathInterconnect(Q... [download for more]
