Alpha 21364 to Ease Memory Bottleneck: 10/26/98

ثبت نشده
چکیده

With processor speeds rapidly approaching the gigahertz mark, the shortcomings of today’s memory architectures are becoming all too apparent. Waiting a few hundred nanoseconds to retrieve data from main memory is tolerable for a 100-MHz CPU, but this delay turns into hundreds of cycles for a 1-GHz processor. Compaq’s Alpha processors are likely to be the first to reach that speed, and at this month’s Microprocessor Forum, the company previewed its solution to this problem, embodied in the 21364. The 21364, code-named EV7, will use multiple Direct RDRAM channels to pump information from main memory right into the processor, greatly reducing latency. The 21364 certainly won’t be the first processor to include a DRAM controller; recent examples range from Cyrix’s MediaGX to Sun’s forthcoming UltraSparc-3. But the Alpha design presents a unique and elegant combination of a large on-chip L2 cache, direct access to DRAM, and a high-speed interprocessor connection. We believe this combination will become a common design for high-performance multiprocessor systems in the next decade. The 21364, which will wrap this new system interface around a 0.18-micron 1-GHz version of the current 21264 CPU core, is still early in its design phase. At the Forum, Alpha architect Pete Bannon said he expects the chip to tape out in late 1999, with shipments in late 2000 or early 2001. Thus, its main competition will come from future high-end processors such as UltraSparc-4 and Intel’s Merced. The Alpha chip’s projected workstation scores of 60 SPECint95 (base) and 100 SPECfp95 (base) set a high bar for competitors to leap, but the 21364 will truly soar in multiprocessor servers, where its lead could be greater.

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

The Alpha 21364 network architecture

Advances in semiconductor technology have let microprocessors integrate more than a 100 million transistors on a single chip. The Alpha 21364 microprocessor uses 152 million transistors to integrate an Alpha 21264 processor core, a 1.75-Mbyte second-level cache, cache coherence hardware, two memory controllers, and a multiprocessor router on a single die, as Figure 1a shows. In the 0.18-micron ...

متن کامل

Testability Features of the Alpha 21364 Microprocessor

The custom testability strategy of the Alpha 21364, Hewlett-Packard’s most recent Alpha microprocessor, builds upon its Alpha 21264 embedded core. Several additional DFT features integrate to meet the testing challenges of the new generation.

متن کامل

A Power Model for Routers: Modeling Alpha 21364 and InfiniBand Routers

As interconnection networks proliferate to many new applications, a low-latency high-throughput fabric is no longer sufficient. Applications are becoming powerconstrained. In this paper, we propose an architecturallevel power model for interconnection network routers that will allow researchers and designers to easily factor in power when exploring architectural trade-offs. We applied our model...

متن کامل

A High Performance Parallel IP Lookup Technique Using Distributed Memory Organization and ISCB-Tree Data Structure

The IP Lookup Process is a key bottleneck in routing due to the increase in routing table size, increasing traıc and migration to IPv6 addresses. The IP address lookup involves computation of the Longest Prefix Matching (LPM), which existing solutions such as BSD Radix Tries, scale poorly when traıc in the router increases or when employed for IPv6 address lookups. In this paper, we describe a ...

متن کامل

A High Performance Parallel IP Lookup Technique Using Distributed Memory Organization and ISCB-Tree Data Structure

The IP Lookup Process is a key bottleneck in routing due to the increase in routing table size, increasing traıc and migration to IPv6 addresses. The IP address lookup involves computation of the Longest Prefix Matching (LPM), which existing solutions such as BSD Radix Tries, scale poorly when traıc in the router increases or when employed for IPv6 address lookups. In this paper, we describe a ...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

عنوان ژورنال:

دوره   شماره 

صفحات  -

تاریخ انتشار 1998