X-bit labs

Intel Conroe – the Axe for Intel Pentium 4?

Category: CPU

by Anton Shilov

[ 02/22/2004 | 12:43 PM ]

In about two or three years from now Intel is going to launch a chip featuring next-generation Pentium M-like architecture for desktop computers, The Inquirer web-site reports. If this happens, the NetBurst desktop processors may become much less popular, or be totally canned.

Apparently, a processor code-named Merom still exists in Intel’s roadmap, unlike previously suggested. Moreover, there is a flavour of the Merom chip for desktops – a processor internally called Conroe due to come in late 2006.

Merom itself reportedly is not a yet another Banias-like architecture, but a completely revamped core also intended for mobile computers with relatively low power consumption, but still with exceptionally high performance per clock, about 20% - 30% higher than that of predecessors, according to The Inquirer. Intel’s microprocessor code-named Conroe is expected to remove certain power constraints and probably widen thermal envelope of the Merom. Additional performance tweaks are also possible to bring extra speed, but the conception of a chip will still remain – a low-power highly efficient central processing unit.

It is not clear whether Conroe microprocessor – a chip with all the capabilities one would expect from a desktop CPU – is a next big Intel’s desktop processor, or just a possible option to complement the Santa Clara, California-based chipmaker’s mainstream line powered by another micro-architecture.

Earlier there were rumours about a chip architecture code-named Nehalem supposed to succeed the whole NetBurst family of products. It was said that the Nehalem would greatly expand the idea of the Hyper-Threading, would feature the hardware security LaGrande technology, the IA32e extensions, as well as other architectural innovations.

The Nehalem was projected to emerge in late 2004, but considering Prescott’s delays and postponements of Tejas, it will hardly emerge until late 2005 or late 2006. Furthermore, the Pentium 4 E “Prescott” showed the world that the NetBurst architecture still has some headroom for development and the industry is looking forward to see more advanced Hyper-Threading, Vanderpool, new instructions and other innovations in future incarnations of the Pentium 4, such as Intel Tejas. In case Intel is capable of making the NetBurst innovations reality, this will give the architecture an opportunity to successfully live on for another couple of years.

Historically Intel has had two general lines for its desktop processors: Pentium and Celeron. However, those chips shared the same architecture most of the time and were more or less equal in terms of capabilities.

These days the situation on the CPU market is changing because usage models of personal computers are transforming. Eventually this may lead CPU companies to offer numerous products lineups for different types of PCs at the same time.

Nowadays there is a group of users who demand a lot of computing power at any cost and with any specifications – those buy Intel Pentium 4 Extreme Edition or AMD Athlon 64 FX processors that may cost from $700 to $1000 and require larger and louder cooling system, powerful PSUs, huge PC cases and so on. There are also a group of customers, who wish to have a small and stylish personal computer that may not be as fast as the most expensive boxes, but that would be trendy and miniature. Because of thermal and power consumption issues modern 90nm Pentium 4 can hardly satisfy all the possible requirements of the second group at a reasonable price.

In case the trend towards increasingly power-hungry desktop chips continues, future incarnations of high-speed processors will not be able to address certain markets at all, while mobile chips probably have different targets to achieve rather than to serve as desktop CPUs. This factor may catalyze processor companies to start offering different families of their desktop products tailored for a definite usage model, but not a line of mainstream chips sharing one micro-architecture intended to satisfy all possible requirements. Every company, however, would prefer to have one micro-architecture across all segments because of the nature of mass market and cost-efficiency.

Intel did not comment on the story.