Because it's just the whole theater about Intel's upcoming 9th. Generation fits: Almost exactly 10 years ago, the manufacturer introduced Nehalem, after two years earlier, the first core processors had painstakingly healed the bloody nose that had been won with Netburst and the Pentium 4 before. Our articles read as this:
Two years ago, Intel achieved a stroke of genius with the introduction of the Conroe architecture (via the Core 2 Duo and Quad processors). The processor prime sacked the netburst debacle and Pentium 4 combined heat and power plant without any honors. During this time, Intel even announced a very ambitious plan to resume the rapid development of its mid-1990s architectures, which has gradually subsided.
The first phase of the plan was to launch a "refresh" of the architecture 12 months after its introduction in order to benefit from the technical advances in the structure. This was done with the Penryn (45 nm). Then, 24 months later, a brand new architecture with the code name Nehalem will be introduced. Intel has almost kept its promises, as the Core i7 processor will soon be released based on the Nehalem architecture.
Although the Conroe architecture offered high performance and reasonable power consumption, it was not perfect. It has to be said that the conditions for their vote were not ideal: when Intel became aware of the impasse into which pentium 4 maneuvered, it had to come up with something new in a hurry. For a company the size of Intel, this is far from easy. The engineering team from Haifa (Israel), which until then was responsible for the mobile architectures, was now promoted to the development team of the entire new product range of the giant from Santa Clara (USA).
What a heavy responsibility for this team, from which the future of the entire company depended! Given the tight schedule and pressure on them, the result achieved by Intel engineers is all the more remarkable. But this also explains why they had to compromise. In particular, the Conroe architecture still revealed its "mobile" roots in places, although it has been fundamentally revised compared to that of the Pentium M…
The architecture wasn't really modular: it was supposed to cover Intel's entire product range, from notebooks to servers, but in practice it was still virtually the same chip. The only point where there was still game was the second-level cache memory. The architecture was as clear as the Dual Core. The evolution to a quad-core version was based on exactly the trick Intel used for its first dual-core processors: two in one case.
The existence of the FSB was also a barrier to matching configurations with a large number of processors, as it was a bottleneck at the memory access level. And in the end, there was a small detail that could not be overlooked: one of the innovations of the Conroe architecture (the Macro-Op fusion, which made it possible to combine two x86 statements into a single op) did not work in 64-bit mode, standard mode of all servers. These compromises were understandable two years ago, but today Intel can no longer justify them, especially in the face of its rival AMD, for whom the server market is the last bastion…
Kommentieren