FeiTeng
Chinese supercomputing CPUs by Phytium and NUDT
FeiTeng is a family of central processing units designed in China for high-performance computing. Developed by Tianjin Phytium Technology and the National University of Defense Technology, these chips power service nodes in major supercomputers like Tianhe-1A and Tianhe-2.
- Used in Tianhe-1A and Tianhe-2 supercomputers
- FT-1000 has 8 cores and SPARCv9 compatibility
- FT-1500 uses 40 nm process and 16 cores
- FT-1500A switches to ARMv8 architecture
- Developed by NUDT and Tianjin Phytium
What it is
FeiTeng refers to a series of microprocessors built for supercomputing tasks in China. Tianjin Phytium Technology and the National University of Defense Technology developed these chips. They serve as central processing units or systems on a chip, handling complex calculations for large-scale scientific models. The family includes early SPARC-compatible designs and later ARM-based processors. These components are critical for operating service nodes within high-performance computing clusters.
How it came about
The project began under the direction of Professor Xing Zuocheng at the National University of Defense Technology. Early versions aimed for binary compatibility with Intel Itanium 2. The second generation, FT64, combined a CPU with a stream processor for acceleration in YinHe supercomputers. A 2012 report noted that the design leveraged work from the OpenSPARC project. In 2020, the team announced the S2500 processor, continuing the roadmap for future high-performance silicon.
Inside the chip
The FeiTeng-1000 uses the SPARCv9 instruction set with eight cores capable of running 64 threads. It features HyperTransport links and PCIe 2.0 connectivity. The later Galaxy FT-1500 shifts to an OpenSPARC base with 16 cores running at 1.8 GHz. This chip supports 256-bit vector operations and fused multiply-add instructions. It includes substantial cache memory, with 16 KB L1 and 512 KB L2 per core, plus a shared 4 MB L3 cache. The FT-1500A variant adopts the ARMv8 architecture with 16 cores and integrated Ethernet controllers.
How it was made
Manufacturing processes have evolved across the family generations. The FeiTeng-1000 was built using 65 nm technology and contains approximately 350 million gates. The subsequent Galaxy FT-1500 utilized a more advanced 40 nm process. These fabrication details highlight the progression in density and efficiency. The chips integrate multiple memory controllers and high-speed interconnects directly onto the die, reducing the need for external components in server configurations.
Where it went
These processors are primarily deployed in national supercomputing facilities. The Tianhe-1A system employs 2,048 FeiTeng-1000 chips to manage service nodes alongside Intel and Nvidia hardware. The Tianhe-2 supercomputer uses 4,096 Galaxy FT-1500 processors. These installations demonstrate the role of FeiTeng in managing data flow and control tasks within massive parallel computing environments, rather than performing all raw number crunching alone.
Versions and family
The family includes the initial Itanium-compatible designs, the FT64 accelerator, and the SPARC-based FeiTeng-1000. The Galaxy FT-1500 introduced higher core counts and OpenSPARC heritage. The FT-1500A marked a shift to ARM64 architecture. The S2500 represents a later iteration announced in 2020. Each variant addresses different performance needs, from early acceleration tasks to modern multi-core server operations with enhanced vector processing capabilities.
Why it matters
FeiTeng processors have enabled China to build some of the world's fastest supercomputers. Their integration into the Tianhe series demonstrates domestic capability in high-performance computing hardware. By utilizing open architectures like SPARC and ARM, the designers created viable alternatives to proprietary western processors. This contributes to technological independence in critical infrastructure sectors such as scientific research and national defence computing.
Questions people ask about FeiTeng
Who designed the FeiTeng processors?
The chips were developed by Tianjin Phytium Technology. The design team was directed by Professor Xing Zuocheng from the National University of Defense Technology in China.
What instruction sets do FeiTeng chips use?
Early models like the FeiTeng-1000 used the SPARCv9 architecture. Later variants, such as the FT-1500A, adopted the ARMv8 instruction set. Some designs also leveraged OpenSPARC resources.
Which supercomputers use FeiTeng processors?
The Tianhe-1A supercomputer uses 2,048 FeiTeng-1000 processors. The Tianhe-2 system employs 4,096 Galaxy FT-1500 chips. They typically handle service node operations within these large clusters.
Good reading and sources
The facts on this page come from the Wikipedia article "FeiTeng", written by its contributors and used under the CC BY-SA 4.0 licence, and from its Wikidata record (CC0). EveryChip rewrote the text in its own words, with AI assistance, and compares every page with the article to make sure no passage is copied. It may be reused under the same CC BY-SA 4.0 licence with a link back.
It may contain mistakes, and figures change over time. Nothing here is investment or buying advice. If something is wrong, tell us. Last refreshed 2026-11-14.