- Intel merges physical processor nuclei in a single design of super virtual nucleus
- Merged nuclei perform instructions in parallel before reorganizing to improve performance
- The approach targets a single effective effectiveness more to wire without widening the size of the nucleus
Intel has filed a patent for what he calls the super hearts defined by software, a technology that merges two or more physical processor cores into a single virtual “super nucleus”.
For the operating system, the merged nuclei appear as a single unit, but the instructions are divided and executed in parallel before being reorganized, aimed at improving single wire performance without the high costs of the construction of larger processors.
This approach resembles older concepts “reverse hyper-threading” from the Pentium 4 era, which suggests that Intel revisits past experiences with modern refinements.
Balance of efficiency and scale
The idea behind this approach is to improve the performance of a single thread by avoiding higher energy requirements associated with faster clock speeds or wider nuclei.
Intel design rather distributes workloads on several cores via shared memory and synchronization modules.
If the mechanism works, the company provides performance gains by Watt, allowing processors to switch between normal and super nuclei modes.
The observers compared Intel’s idea to the old multithreading in AMD cluster, although the methods differ.
AMD divides the nuclei into modules, while Intel’s proposal merges whole nuclei under software control.
Some also link the patent to the Royall Core canceled project of Intel, which would have driven up instructions raised by clock but became unsanished to manufacture.
By reviving such strategies, Intel seems to seek alternatives to brute force design extensions.
However, the lack of measured data makes it impossible to know if it could compete with the fastest CPU conceptions on the market.
The patent describes a small synchronization module inside each nucleus, supported by a reserved memory region called the address space of the Ver.
These manage registry transfers, control and data flow to guarantee the integrity of the instruction.
Software side, compilers or binary instrumentation divide the code into manageable blocks while inserting flow control controls.
Operating systems must then decide when a workload benefits from the Super Core mode, a requirement that could complicate planning and compatibility.
Without a large support of hardware and software, design may become unused feature.
Intel’s documentation does not consider clear performance gains, which only suggests that two closer nuclei could discuss the capacity of a wider nucleus under certain conditions.
Technology could interest researchers to explore specialized workloads, including scenarios in which a mining processor could seek improved efficiency in unique thread tasks.
However, for general IT, the lack of proven benchmarks leaves the promise uncertain, and if this really creates the best processor for demanding workloads remains an open question.
Via Toms equipment
