Loongson recently unveiled two new quad core processors Loongson-3A2000 & Loongson-3B2000 targeting respectively consumer electronics and server, and on the company’s new GS464E architecture using MIPS64 cores.
Both processors are 4-way superscalar processors built on a 9-stage, super-pipelined architecture with in-order execution units, two floating-point units, a memory management unit, and a crossbar interconnect. According to Imagination Technology, Chinese media reported that 3A2000 CPU was designed at high-performance consumer electronics products such as desktop computers and laptops, 64-bit embedded and DSP applications, and network routers, while 3B2000 will be used in 8 to 16 core server systems, meaning there would be 2 to 4 quad core processors in these servers.
I could not find anything about the new processors on Loongson product page, but a white paper describes GS464E architecture in details. Unfortunately, the document is entirely in Chinese, except the abstract and some charts and tables:
Loongson GS464E is the most recent processor architecture introduced by Loongson Technology. In this paper, we describe the key aspects of the GS464E architecture. Compared to the previous GS464 architecture, GS464E focuses on improving the performance of memory access and branch prediction, using enlarged queues, caches, and TLBs in the processor. Support for the MIPS DSP instruction set and virtualization is also provided. Specifically, the memory subsystems have a 3-level cache hierarchy each with an LRU replacement policy, and also support multi-processor cache coherence. With the aforementioned optimization features, GS464E has become an innovative, high-performance processor architecture.
And some results are really impressive with over 2000% improvement in memory transfer over the previous generation Loongson-3A (GS464).
However, I could not find any power consumption data such as TDP or performance per watt, so I’m not sure the new processor excels there, especially since the processors are apparently manufactured with a 40nm process technology. However, two new chips to be released in 2016 – 3A3000 and 3B3000 – will be built using 28nm process node and clocked at 2GHz.
Some other interesting features of the processor / architecture include:
- MIPS64 Release 3 instructions
- LoongBT, a binary translation technology that enables developers to run x86 and ARM code
- LoongVM instructions for custom virtual machines
- LoongSIMD instructions for 128- and 256-bit vector arithmetic operations
The new processors should soon be found in devices sold in China.
Jean-Luc started CNX Software in 2010 as a part-time endeavor, before quitting his job as a software engineering manager, and starting to write daily news, and reviews full time later in 2011.