Bluetooth low energy v5.4 Baseband Controller, Protocol Software Stack and Profiles IP
AMD shows 64-bit processor architecture
AMD shows 64-bit processor architecture
By Will Wade, EE Times
August 10, 2000 (7:27 p.m. EST)
URL: http://www.eetimes.com/story/OEG20000810S0035
SAN MATEO, Calif. Advanced Micro Devices Inc. took the wraps off its 64-bit microprocessor architecture on Thursday (Aug. 10) and said it will have chips available by late next year. While trailing archrival Intel Corp. in the race to deliver a 64-bit chip, AMD hopes to gain a marketing advantage with its architecture, a simple extension of existing 32-bit designs, by making it easy for designers to implement. "There is a fairly big demand for 64-bit processors, and our customers have been telling us that they want them to be done in a straightforward manner," said Bob Mitton, division marketing manager for enterprise products at AMD (Sunnyvale, Calif.). "They don't want us to reinvent the entire architecture." Mitton's comment is a clear snipe at Intel, which has developed a new 64-bit architecture, termed IA-64, from scratch. While Intel's project has taken more time and money than AMD's X86-64 development effort, Intel has frequently s tated that a brand-new design will offer higher performance than a modification of existing 32-bit instruction sets. While analysts give some weight to this argument, AMD counters by stating that its own architecture is backwards-compatible with all existing systems and software, and will be much easier for engineers to design into products. "AMD is taking a significantly different step from Intel," said Dean McCarron, principal analyst for Mercury Research (Scottsdale, Ariz.). "Unlike Intel, which will have both IA-32 and IA-64 architectures and will have two different product lines going into the future, AMD will only have one product line on its road map." Not only has AMD's development efforts been shorter and less expensive than Intel's, Mitton said he expects to see designers embrace his company's X86-64 products and slip them into existing system designs without significant modifications, and to run existing 32-bit applications with no changes at all. While moving into the 64-bit realm is a m ajor accomplishment, it did not require major changes to AMD's 32-bit architecture. Mitton explained that the new design features some new instruction prefixes to modify the instructions' performance at the 64-bit level, and it doubled the number of registers to total 16. "This is a pretty simple, but elegant, approach," said McCarron. "It is much easier than what Intel has done." While Intel's IA-64 chips will also be able to run 32-bit applications, the chips will need to modify 32-bit instructions to do so. Nathan Brookwood, principal analyst for Insight64 (Saratoga, Calif.), said this likely will mean that Intel's 64-bit processors will process 32-bit applications less efficiently than a core designed specifically for that task. But what Intel may lack in 32-bit performance, it will certainly make up for with a broad 64-bit line. The first IA-64 chip, originally code-named Merced, will be available to OEMs under the Itanium brand name beginning this quarter, and systems based on that device will likely appear early next year. Beyond the Itanium, Intel's road map calls for successive 64-bit chips called McKinley, Northfield, Deerfield, and a fifth chip that has not been publicly named. With AMD only now releasing the architectural details of its X86-64 design, Mitton said there will be samples of its first implementation, the Sledgehammer, sometime next year, and volume production is expected in 2002. Intel and AMD have been competing head-to-head in the desktop PC market, but they are starting to tread upon different paths with their 64-bit architectures. Intel will continue to push its 32-bit microprocessors in the consumer space, while promoting its 64-bit chips for servers, workstations and high-end desktop systems. AMD will initially target high-end PCs, servers and workstations with its 64-bit devices, but Brookwood and McCarron are agreed that the company will begin to offer the technology to mid-level systems as well. If so, AMD will be offering 64-bit performance against Intel's 32-bit chips in the mid-range desktop market. While this may allow AMD to claim a marketing advantage, Brookwood said that few if any standard applications will actually be able to use 64-bit power because the programs have been written for 32-bit chips. Brookwood said the Sledgehammer may be viewed as the successor to the Athlon, AMD's current flagship processor. "Sledgehammer will be AMD's next 32-bit processor, and it will also do 64-bit applications," he said. Beyond running applications faster, there are other features that will help 64-bit chips' popularity. The current crop of 32-bit processors can only address up to 4 Gbytes of system memory, and that's a limiting factor for today's high-end servers and systems that power major databases. But that number will double with each additional bit of processing power, so a 64-bit MPU is mathematically capable of supporting 18 exebytes of system memory (that's 1018 bytes of DRAM), according to Brookwood. While nobody will need that much memory anytime soon, PC OEMs do need to deliver more system memory. "That's one of the real reasons for moving to a 64-bit processor," Brookwood said.
Related News
- AMD licenses Mips' 64-bit processor technology for future products
- Toshiba Announces Availability of Reference Board for the TX4939XBG-400 Embedded PCI-Based Microprocessor, Its Newest 64-Bit RISC Processor
- China's BLX unveils 64-bit processor
- CPU Tech announces World's first configurable family of true 64-bit processor cores for Deeply Coupled Computing
- MIPS Technologies and Agilent Technologies Announce Agreement on 32- and 64-Bit MIPS-Based Embedded Processor Cores
Breaking News
- HPC customer engages Sondrel for high end chip design
- Ubitium Debuts First Universal RISC-V Processor to Enable AI at No Additional Cost, as It Raises $3.7M
- TSMC drives A16, 3D process technology
- Frontgrade Gaisler Unveils GR716B, a New Standard in Space-Grade Microcontrollers
- Blueshift Memory launches BlueFive processor, accelerating computation by up to 50 times and saving up to 65% energy
Most Popular
- Cadence Unveils Arm-Based System Chiplet
- Eliyan Ports Industry's Highest Performing PHY to Samsung Foundry SF4X Process Node, Achieving up to 40 Gbps Bandwidth at Unprecedented Power Levels with UCIe-Compliant Chiplet Interconnect Technology
- TSMC drives A16, 3D process technology
- CXL Fabless Startup Panmnesia Secures Over $60M in Series A Funding, Aiming to Lead the CXL Switch Silicon Chip and CXL IP
- Blueshift Memory launches BlueFive processor, accelerating computation by up to 50 times and saving up to 65% energy
E-mail This Article | Printer-Friendly Page |