[ad_1]
Although Godson claims its The latest 3A5000 and 3C5000 processors Is based on its Proprietary architecture developed in-house And no longer relying on the MIPS architecture, the code the company uses to enable the new CPU in Linux is actually the same as the code it uses for MIPS-based chips. In addition, even on paper, CPU developers cannot show software developers the advantages of their architecture.
Historically, Loongson’s CPU relied on the company’s various types of LoongISA architecture, which is a custom subset of the MIPS64 architecture. This strategy allows the company to maintain compatibility with programs designed for MIPS64, including software for high-performance computing applications, while introducing its own extensions to improve the performance of contemporary applications.
But in order to make a major leap in performance, Loongson needs a completely new architecture, which is why it developed LoongArch. The architecture continues to be compatible with MIPS, but it has approximately 2,000 proprietary instructions, binary conversion extension instructions (LBT), vector processing extension instructions (LSX), advanced vector processing extension instructions (LASX), and virtualization extension instructions (LVZ).
However, instead of writing new code to enable LoongArch-based CPUs in Linux, the company continued to use old code written for MIPS64-powered processors, which caused some setbacks in the community, the report said Fronix.
“You keep saying’not MIPS’, but all I see is a blind copy of the MIPS code,” a software developer wrote To Loongson. “This is still the same outdated, broken MIPS code, but with a different name.”
Since Loongson’s 3A5000 and 3C5000 CPUs based on LoongArch can execute code designed for the MIPS64 platform, and there may not be too many differences between the company’s LoongArch and MIPS64 platforms, there is nothing wrong with enabling them in Linux using verified codes . However, the use of old code means that any new features of the platform are hidden.
“What is that? Another MIPS legacy? If it obviously applies to every root interrupt, why must it be every interrupt?” Another developer wanted to know. “You obviously still have some static partitioning of the interrupt space, which is unacceptable for the new architecture.”
As of now, it remains to be seen when Loongson will enable the new features of its LoongArch architecture in Linux.
[ad_2]
