Performance Efficiency AI Accelerator for Mobile and Edge Devices
ARM in partner discussions over next-gen architecture
ARM in partner discussions over next-gen architecture
By Peter Clarke, Semiconductor Business News
September 11, 2002 (3:20 p.m. EST)
URL: http://www.eetimes.com/story/OEG20020910S0035
LONDON -- ARM Holdings plc, the developer and licensor of the ARM family of RISC processor cores, is in discussions with partners over what features and technologies should be included in the next generation of its microprocessor architecture, and Intel and Texas Instruments are thought to be two of the partners with whom ARM is consulting. This is despite the fact that the current architecture, version six, is only just being fashioned into "ARM11" processor core variants and any next generation architecture is not expected to turn up in silicon for three or four years. Nonetheless, as silicon manufacturing capability is accelerating towards being able to offer billions of transistors on a chip, so architecture and implementation engineers need to determine what features will become desirable and worth adding to an architecture at different levels of integration complexity. Increased instruction code density, which helps drive down power co nsumption and cost, remains a number one priority for ARM so that, together with increased support for multiprocessing and network-style massively parallel processing are likely to be key features of the next generation architecture, according to Simon Segars, executive vice president of engineering at ARM. ARM, already the possessor of a low power CPU, made a significant contribution to increasing code density and further reducing power consumption when it introduced the "Thumb" 16-bit instruction format for its processors in the mid-1990s. It remains to be seen whether ARM can pull off another power saving coup in this decade. Speaking at a press conference held here today (September 10, 2002) Segars, jokingly said that observers might jump to the conclusion that the next generation of the ARM processor architecture would be called version seven but that it was a little too early to say. Whether or not ARM chooses to stick with the obvious and utilitarian naming convention, Segars told SBN I> that consultations with partners on the next generation architecture had started. "We like to get partners involved at an early stage. We have identified a number of key features we want to look at," Segars said. He said a presentation around these key features had been given to the whole ARM licensee and developer community at the recent ARM developers forum, held over two days in Cambridge, England, in August. Some partners just want to license cores after the architectural and implementation engineering has been done, Segars said, while others want to get involved with discussions about what is important to them and what they want to build in silicon. "Implementation is very important," said Segars pointing out that information about the resources available at any given process technology node was important to effectively trading off the power, frequency and area implications of new architectural features. When asked if Intel's input was being sought Segars said: "Intel has a great t rack record of designing and building microprocessors." Intel, is an architectural licensee of the fifth version of the ARM architecture (ARMv5) and is committed to ARM in the guise of its XScale processor family. Although not an architectural licensee Texas Instruments has been one of the leading deliverers to the market of silicon which includes ARM cores. Both these companies have aggressive silicon manufacturing roadmaps. Segars declined to confirm that ARM is in discussions with Intel and Texas Instruments, but said: "It's not just about those two. It's about other people. The operating system people are very important and we need to consult with them. There's no point in us putting features into the architecture that could be implemented in silicon if the operating systems don't want to make use of them." Segars said that he expected the next generation of the ARM architecture to take at least a couple of years to solidify and then it would probably take another year or more for particular designs that implemented the architecture to be completed. At the Microprocessor Forum coming up in San Jose, California, October 2002, Segars is due to disclose details of the first implementations built round the ARM11 microprocessor core which is the first manifestation of the sixth version of the ARM architecture (ARMv6).
Related News
- Bringing Arm TCS23 to Life on Next-gen Flagship Smartphones with MediaTek and vivo
- NVIDIA Collaborates With SoftBank Corp. to Power SoftBank's Next-Gen Data Centers Using Grace Hopper Superchip for Generative AI and 5G/6G
- Arm announces new "Automotive Enhanced" processor designed for safe next-gen driver experiences
- Qualcomm, CEVA, ARM Race to Next-Gen Modem
- ARM Forecast to Accelerate Lead in the Automotive Sector Over the next Five Years
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 |