Details of new Godson 3B1500 microprocessor will be presented at upcoming IEEE International solid-state Circuits Conference (IISCC) by Loongson Technology. Godson 3B1500 is a MIPS64-based processor, that offers 35% higher floating-point performance than its predecessor, Godson 3B, while staying within the same thermal envelope. The CPU is manufactured on 32nm technology, which is a big leap from 3B's 65nm technological process. Development of new microprocessor was partially funded by Chinese Academy of Sciences.
Godson 3B was detailed at the last year's IISCC conference. This MIPS64-based processor is built on 65nm technology, and it has 8 CPU cores, implemented as two quad-core nodes. Each node has its own shared L2 cache, DDR2/DDR3 memory controller and HyperTransport controller. The total size of L2 cache in the package is 4 MB. The processor is clocked at 1GHz, and has maximum performance 128GFLOPS when using double-precision accuracy. Godson 3B has 40 Watt Thermal Design Power.
According to IISCC's advanced program information, Godson 3B1500 will be clocked 35% higher, or at 1.35 GHz, while having the same 40 Watt TDP. Processor's performance for double-precision calculations was increased to 172.8GFLOPS. Based on other sources, the 3B1500 will have smaller die size, 182.5mm2 as opposed to 299.8mm2 on the 3B. The size of L2 cache on the chip hasn't changed, but the processor will have 8 MB level 3 cache. A picture of this processor can be found in this PDF file, located on Loongson company website.